NBN should abolish speed tiers, says economist

114

blog Those of you with a more technical bent (hell, we hope that’s all of Delimiter’s readers at this point) will be aware that the speed tiers set by NBN Co — 12Mbps, 25Mbps, 50Mbps, 100Mbps and eventually 1Gbps, plus all the upload tiers — are actually artificial. There’s no real reason why the NBN needs to offer speeds at these discrete tiers, apart from commercial reasons: Giving retail ISPs options to offer customers, giving end users familiar options comparable to existing ADSL/HFC broadband services and so on.

However, according to one economist, the NBN might actually see higher uptake if just one flat speed (presumably 100Mbps, initially) was provided. The ABC reports (we recommend you click here for the full article):

NBN Co is offering five different speeds at different prices, which telecommunications consultant and former Telstra chief economist John de Ridder says is artificial because, once the network is built, it does not cost more to deliver 100 mbps over the NBN than it does the slower speeds. “They could offer the same speed, unlimited speed basically, to everybody but they choose to ration it out, charging more for higher speeds,” he said.

The difficulty with Ridder’s argument is that it’s actually not based on evidence. He claims in the ABC’s article that most people won’t pay more to take up the NBN’s higher speed tiers. In fact, the evidence so far shows the opposite: Most will. In October last year NBN Co revealed that 44 percent of NBN customers who have so far signed up to use the network had opted for 100Mbps options, and another survey published in June last year showed that 85 percent of Australian consumers wanted to be able to connect to the Internet at speeds of 50Mbps or higher.

However, that’s not to say that de Ridder’s idea doesn’t have merit. It does: A lot of merit. I’ve written previously that the current tranche of NBN plans is based directly on thinking imported from the previous ADSL broadband framework; when actually there’s a clear opportunity to do things differently; better; with the new technology.

I don’t know whether de Ridder’s idea of abolishing NBN speed tiers and charging only for actual usage of the network is economic; and I certainly don’t think Australians are ready for such a radical change to their broadband habits. But I do think it’s the kind of idea which retail ISPs should be modelling over and experimenting with; in much the same way as ISPs like Exetel have started to experiment with NBN plans featuring unlimited downloads.

114 COMMENTS

  1. Speed tiers are there for a reason – increase competition, because not every RSP will be wanting to target every customer base.

    It’s also there to make sure different technologies don’t get confused with getting faster speed or not.

    • > Speed tiers are there for a reason – increase competition, because not every RSP will be wanting to target every customer base.

      I find it doubtful that speed tiers increase competition. Essentially the cost for RSPs is driven by the amount of data that it is downloaded at, not the speed, and that bulk of the cost comes from people who download more, particularly during peak time. Consider these numbers for someone on a 1Gbps plan, how much time they are using their connection at full speed:
      – 50GB quota = 0.22 minutes a day
      – 100GB quota = .44 minutes a day
      – 1TB quota = 4.44 minutes a day
      – 5TB quota = 22.22 minutes a day

      As you can see someone on more than double the current average usage in Australia would be barely touching the network.

      The key constraint for network performance is the purchase of CVC. If an RSP chooses to support 100Mbps, then they need to purchase at least 200Mbps, as explained in Simon Hackett’s NBN Points of Interconnect and the future of competition blog post.

      The second point that needs to be realised is that the wholesale pricing is an artificial construct. AVC is not based on the cost to install and CVC is priced because NBNCo are expecting that data usage will grow and they want to benefit from that. If NBNCo mandated that every RSP had to purchase 2Gbps of CVC then the CVC price could drop dramatically.

      > It’s also there to make sure different technologies don’t get confused with getting faster speed or not.

      I’m not sure the point you are making here. Are you suggesting that confusion could exist between fibre, fixed wireless and satellite options provided by NBNCo? I think it will make plans simpler – just choose your quota.

      The ironic fact is that by introducing speed tiers, FTTN, HFC, 4G and even approaching half of ADSL2+ connections will be faster than half of NBNCo fibre connections according to the corporate plan

      • “Essentially the cost for RSPs is driven by the amount of data that it is downloaded at, not the speed”

        Of course, the entire tech industry has been completely blindsided!

        Seriously dude, you were wrong a year or two back when you first thought of this and are still wrong today – how about you actually do some research!

        • > Seriously dude, you were wrong a year or two back when you first thought of this and are still wrong today – how about you actually do some research!

          Actually it has been over 4 years. We’ve been through this argument on whirlpool. The data is there and you failed to refute it.

          How about you acknowledge that after 4 years, I might actually have a point.

          • You are fundamentally wrong on this point:
            “Essentially the cost for RSPs is driven by the amount of data that it is downloaded at, not the speed”

            I am a network tech (among many other hats) at a business that is effectively a small ISP.

            We don’t pay for data. We pay a flat rate for our internet. Just like every other ISP in the universe. The amount we pay directly dictates the speed we get.

            We ration our available speed; by putting limits on the amount of data our users use but in reality the amount they download is completely irrelevant. It is entirely the speed at which they download at any particular [ie peak] time.

            The model the NBNCo has chosen; is the Wholesale model that every isp uses. But they have extended it in both directions.
            Right now:
            IE, an ISP buys bandwidth from (lets say) America, and then they buy the one link they have access to to a customer (the copper). Because the copper is so limited, no one sells it in speed tiers [any more].

            In the future:
            an ISP buys bandwidth from America, and now they buy the bandwidth to the customer that they can afford.

            If reality was how you painted it; then my work would have 10gigabit fibre connection to the internet (it is a direct point to point fibre to their datacentre as it is); because it cost our wholesaler the exact same amount as the 100 megabits we buy today.

          • > We don’t pay for data. We pay a flat rate for our internet. Just like every other ISP in the universe. The amount we pay directly dictates the speed we get.

            But we aren’t discussing the wholesale model, we are discussing the retail model. Your comments are perfectly valid when discussing CVC.

            As you acknowledged data quotas are the primary way of managing load and will continue to be under the NBN. What is not required is speed tiers because the reality is once you have a few customers on 1Gbps connections, there isn’t any network advantage in having low quota customers on slower speeds. Speed shaping will kick in before they can significantly affect the network.

            > We ration our available speed; by putting limits on the amount of data our users use but in reality the amount they download is completely irrelevant. It is entirely the speed at which they download at any particular [ie peak] time.

            What every RSP relies on is that not everyone is downloading all the time (even during peak time). As I’ve previously explained someone on a 1Gbps connection with only 50GB of quota will use the network for less than 14 seconds per day at full speed – barely a blip on the network load.

            What do you sell to your customers? Under your misguided argument, 100Mbps would only support an average of 10 ADSL2+ customers. I would guess you have significantly more customers or you are offering a very premium service. As you have indicated it is the speed they download during peak time, but it is extremely rare for all customers to be downloading at the same time. People’s different routines cause this to happen naturally and quotas enforce it.

            If we take what you are saying to its logical conclusion, then an RSP would need to purchase the same quantity of CVC as the AVC speed.

            > The model the NBNCo has chosen; is the Wholesale model that every isp uses. But they have extended it in both directions.
            Right now:

            Yes it is the wholesale model, but as you correctly point out NBNCo have chosen to extended it to the retail end as well (just like Telstra did prior to competition in ADSL).

            > If reality was how you painted it; then my work would have 10gigabit fibre connection to the internet (it is a direct point to point fibre to their datacentre as it is); because it cost our wholesaler the exact same amount as the 100 megabits we buy today.

            You have misrepresented my argument. The 100Mbps connection you have today is comparable to NBNCo’s CVC charges. Conceptually the internet is a distribution network. Just like with water and power if everyone turns on all their taps / appliances then the distribution network suffers (lower water pressure / brown outs). At each level in the network, operators seek to provide sufficient capacity to meet their client’s demands at an affordable price.

            As other posters have pointed out, the largest component of the cost is the installation of the fibre. The terminators are a small fraction of the cost and the advantage of fibre is that if you “upgrade” the terminators the speed can increase dramatically. Your data centre could upgrade the fibre from 100Mbps to 10Gbps for a fraction of cost of speed multiplier, but they would also need to increase their backhaul to support those higher speeds, under the assumption that during peak periods you will be constantly using the connection at near to capacity as possible without causing congestion (budget RSPs have networks with congestion). Contrast this with NBNCo who are paid by RSPs to transfer the data off their network via CVC charges. Higher speeds means NBNCo will receive more for CVC, but since it’s costs won’t increase dramatically it can simply cut CVC charges dramatically. Unlike your provider NBNCo doesn’t have to pay for additional backhaul from upstream providers.

            An RSP aggregates many connections to even out the load resulting in greater utilisation of backhaul. Contrast this with a customer, where most of the time the connection is idle. For example at 1Gbps a 5TB quota provides less than 23 minutes of full speed access a day. What this does open up is for off-peak quotas to grow substantially. People who are prepared to download during off-peak times could start to see currently inconceivable quota levels.

          • But we aren’t discussing the wholesale model, we are discussing the retail model. Your comments are perfectly valid when discussing CVC.

            NBNCo are a wholesaler. What retailer chose to do with NBNCo’s products are not NBNCo’s problem.

            As you acknowledged data quotas are the primary way of managing load and will continue to be under the NBN. What is not required is speed tiers because the reality is once you have a few customers on 1Gbps connections, there isn’t any network advantage in having low quota customers on slower speeds. Speed shaping will kick in before they can significantly affect the network.

            I’m sorry but do you even understand how peak loading works. The majority of people use the internet at the same time of day, this means that the chance of a collision is very high. How much CVC a RSP buys is determined by their peak loading. The quota is a method of cross sub to simplify the costs to consumers. In other words quota is not directly proportional to the required CVC. If it was your argument might have merit, but it isn’t.

            What every RSP relies on is that not everyone is downloading all the time (even during peak time).

            Correct, but…

            As I’ve previously explained someone on a 1Gbps connection with only 50GB of quota will use the network for less than 14 seconds per day at full speed – barely a blip on the network load.

            … this extension of the argument assumes uniform predictable loading, i.e. user A always uses the network at 5% of her capacity between 6 and 7 p.m. in the evening. In reality users do not uniformly load the network.

            What do you sell to your customers? Under your misguided argument, 100Mbps would only support an average of 10 ADSL2+ customers.

            In no way does his argument imply this.

            Yes it is the wholesale model, but as you correctly point out NBNCo have chosen to extended it to the retail end as well (just like Telstra did prior to competition in ADSL).

            Once again, NBN are a wholesaler. They should be extending the wholesale model. The only reason Telstra didn’t with ADSL2+ is because they saturated demand. There was no demand for slower speed tiers. The same will eventually happen with NBNCo, but not from day one.

            You have misrepresented my argument.

            Somewhat hypocritical.

          • > I’m sorry but do you even understand how peak loading works. The majority of people use the internet at the same time of day, this means that the chance of a collision is very high. How much CVC a RSP buys is determined by their peak loading. The quota is a method of cross sub to simplify the costs to consumers. In other words quota is not directly proportional to the required CVC. If it was your argument might have merit, but it isn’t.

            You keep claiming this, yet you provide absolutely zero evidence. I’ve shown you that someone with a 5TB quota has less than 23 minutes online at full speed per day. The worst peak period is in the evening – lets say 7pm-11pm. Now that is 16 times longer than a person with a 1TB quota can be online.

            Sure CVC demand is not directly driven by quota, but as the customer base grows larger, the other factors become small.

            >> What every RSP relies on is that not everyone is downloading all the time (even during peak time).
            > Correct, but…
            >> As I’ve previously explained someone on a 1Gbps connection with only 50GB of quota will use the network for less than 14 seconds per day at full speed – barely a blip on the network load.
            > … this extension of the argument assumes uniform predictable loading, i.e. user A always uses the network at 5% of her capacity between 6 and 7 p.m. in the evening. In reality users do not uniformly load the network.

            It just so happens if you have a large enough customer base it does become somewhat uniform. Family A might have dinner between 6-7pm, while another family has dinner between 7-8pm. On some nights people will be home, etc. It works now. It will work with the NBN.

            >> What do you sell to your customers? Under your misguided argument, 100Mbps would only support an average of 10 ADSL2+ customers.
            > In no way does his argument imply this.

            Actually it does, he makes a false assumption about end users fully loading their network during peak. This is simply not the case, except for very high quota users and not the case at all for fast speeds. This is not the case. If you move from 5TB to a more realistic quota for today of 500GB and that becomes 2.5 minutes. Not very much time at all.

            >> Yes it is the wholesale model, but as you correctly point out NBNCo have chosen to extended it to the retail end as well (just like Telstra did prior to competition in ADSL).
            > Once again, NBN are a wholesaler. They should be extending the wholesale model.

            Except that NBNCo is selling end-points on a network which have very different characteristics to backhaul.

            > The only reason Telstra didn’t with ADSL2+ is because they saturated demand. There was no demand for slower speed tiers. The same will eventually happen with NBNCo, but not from day one.

            WRONG. Telstra moved because competition had already marched too far ahead and they were loosing customers. NBNCo don’t have to worry about that. What you may not realise is that the NBNCo numbers are based entirely on delivering 7% ROI. Providing faster speeds won’t dramatically change NBNCo expenses. Higher revenue just means having to cut charges. There is zero real incentive for NBNCo to deliver higher speeds, infact because higher speeds are slightly more demanding there is actually a perverse incentive not to deliver higher speeds.

          • You keep claiming this, yet you provide absolutely zero evidence. I’ve shown you that someone with a 5TB quota has less than 23 minutes online at full speed per day. The worst peak period is in the evening – lets say 7pm-11pm. Now that is 16 times longer than a person with a 1TB quota can be online.

            It’s basic knowledge for anyone who’s done any network engineering. Someone with 5TB of quota does not “only use the network for 23 minutes a day”, because they will be bursting. They don’t use their entire connection at once. Or even in discrete chunks. Most users do not actually use anywhere near their allocated quota.

            In fact in Australia they only use about 15% on average. Therefore, your continued premise that “problem users” will be shaped before they can significantly contribute to network congestion is false.

            People who buy more quota might not actually use more quota.

            In other words, when a new user is added to the network, their quota will have little relation on the decision by the ISP to provision more backhaul. Only the peak loading determines that.

            It just so happens if you have a large enough customer base it does become somewhat uniform. Family A might have dinner between 6-7pm, while another family has dinner between 7-8pm. On some nights people will be home, etc. It works now. It will work with the NBN.

            The network goes into predictable patterns, but these patterns bare no relation to the allocated quota, in other-words they aren’t uniform. From day to day, week to week, or even month-to-month the peak usage will change. User A won’t always use the network between 6 and 7.

            Actually it does, he makes a false assumption about end users fully loading their network during peak. This is simply not the case, except for very high quota users and not the case at all for fast speeds. This is not the case. If you move from 5TB to a more realistic quota for today of 500GB and that becomes 2.5 minutes. Not very much time at all.

            Actually, you make that assumption by continually translating quota into “time”. It doesn’t work like that. The internet can be slip into two main groups of traffic: Real time and non real time.

            Real time traffic uses a consistent amount of bandwith within a margin of error, however this bandwidth is usally less than the available bandwidth. Gaming for example only uses a few kbps, video calling maybe 1-2mbps.

            Non real time traffic is designed to minimise the transaction time to as short as practically possible. Loading a website only takes a few seconds at most, downloading a video on YouTube is buffered, etc.

            A users total transaction time might be only a few minutes but the time they are on the network can be for many hours. As you have more users the chance of collision is high, because there are hundreds of active users at one time.

            With Telstra HFC networks, the chance of collision is so high that the peak bandwidth available suffers. Telstra can’t fix this problem by reducing everyone’s quotas. Most, as above, probably won’t even be affected if they do that.

            Except that NBNCo is selling end-points on a network which have very different characteristics to backhaul.

            No they don’t. Your NTU is still an aggregation node.

            WRONG. Telstra moved because competition had already marched too far ahead and they were loosing customers. NBNCo don’t have to worry about that. What you may not realise is that the NBNCo numbers are based entirely on delivering 7% ROI. Providing faster speeds won’t dramatically change NBNCo expenses. Higher revenue just means having to cut charges. There is zero real incentive for NBNCo to deliver higher speeds, infact because higher speeds are slightly more demanding there is actually a perverse incentive not to deliver higher speeds

            Just because competition was a contributing factor does not mean it is the only factor. Further, since NBNCo do not have a profit motivation like Telstra, there is no reason to believe they won’t put customer requirements ahead of sustaining profit.

          • > It’s basic knowledge for anyone who’s done any network engineering. Someone with 5TB of quota does not “only use the network for 23 minutes a day”, because they will be bursting. They don’t use their entire connection at once. Or even in discrete chunks. Most users do not actually use anywhere near their allocated quota.

            Quite true, so thanks for reminding everyone that I’m presenting a worst case example.

            > In fact in Australia they only use about 15% on average. Therefore, your continued premise that “problem users” will be shaped before they can significantly contribute to network congestion is false.

            Do you have numbers to support the 15% average? The only published numbers I’ve seen are in v1 of the NBNCo Corporate Plan and they were significantly higher.

            Can you provide any evidence to support your assertion? Someone who uses their entire 1TB quota in a single hit will be shaped in just over 2 hours.

            > People who buy more quota might not actually use more quota.
            > In other words, when a new user is added to the network, their quota will have little relation on the decision by the ISP to provision more backhaul. Only the peak loading determines that.

            I don’t know what you are using for peak loading, but when quota is clearly the limiting factor in usage in Australia. What you are clearly

            > The network goes into predictable patterns, but these patterns bare no relation to the allocated quota, in other-words they aren’t uniform. From day to day, week to week, or even month-to-month the peak usage will change. User A won’t always use the network between 6 and 7.

            Yes, but what about user A1, A2, ….. A10,000? Consider tossing a coin. It could land either heads or tails. If you toss 10 coins, then it is likely close to 5 will be heads and 5 will be tails. If you toss 10,000 coins it even more likely close 5000 will be heads and 5000 will be tails. The more coins you toss the more likely that half will be heads and half will be tails. The same effect occurs with customers using the network. This is the reason that we have seen so much ISP consolidation in the past 3 years) is that as your customer base grows, the usage averages out and there are less sudden peaks.

            > Actually, you make that assumption by continually translating quota into “time”. It doesn’t work like that. The internet can be slip into two main groups of traffic: Real time and non real time.

            It is still individual packets that require time on the network. We could equally assume the 23 minutes was split evenly over the peak period. Remember we are considering only the worst case of people consuming traffic during the evening peak.

            > Real time traffic uses a consistent amount of bandwith within a margin of error, however this bandwidth is usally less than the available bandwidth. Gaming for example only uses a few kbps, video calling maybe 1-2mbps.

            So gamers will appreciate the reduced latency.

            > Non real time traffic is designed to minimise the transaction time to as short as practically possible. Loading a website only takes a few seconds at most, downloading a video on YouTube is buffered, etc.

            Excellent, the user is off the network before the next person starts to use the network, rather than hanging around because of the artificial cap (and contributing to buffer bloat problems).

            > A users total transaction time might be only a few minutes but the time they are on the network can be for many hours. As you have more users the chance of collision is high, because there are hundreds of active users at one time.

            Only if you don’t increase the bandwidth that is available. As I’ve continually pointed out the larger your network, the less bandwidth you need to add to reduce the possibility of collision and provide sufficient headroom. It is a really simple concept to grasp.

          • I have really tried to be reasonable. I even tried to write a point by point rebuttal but it’s clear you have absolutely no idea what you’re talking about, or worse, are completely incapable of reading. It is clear to me that this argument is pointless now.

            I don’t say the above lightly so let me explain with what you have just done in that post which has finally pissed me off:

            – You have the balls to ask me for evidence when I already provided you with a link. You see that underlined part? The one in blue? Take your mouse, and press the left button over it. You’ll notice that another page’ll open.

            – Peak loading is the used bandwidth in peak time, what other definition is there?

            – Giving me a lesson in statistical distribution and averaging does not address my point which was that [the observable] patterns bare no relation to the allocated quota. You seem to have completely missed that, and tried to argue that I don’t understand statistical distribution and averaging, which I do, you know, where I said: The network goes into predictable patterns.

            – You have continually assumed that provision for 1Gbps services, assuming NBNCo didn’t impose speed tiers, your implied solution, will not cost any-more for an RSP than provisioner for a 10Mbps service. You are wrong. Because they have to pay for back-haul. Arguing that once they get there it won’t cost any-more to add more users is completely missing the point of my argument . So it’s a simple concept to grasp, yes, but it doesn’t address the problems I continually try and point out to you. Like the key one, quota doesn’t indicate peak usage.

            – Yes, once they get a pair of 1Gbps customers up they can add more at no extra cost. But to do that requires 2Gbps of backhaul. Are you assuming that the excess demand created by 1Gbps connections will stimulate the backhaul market to lower prices, because I think you’ll find what with all the extra equipment they need to buy that it’ll have the complete opposite effect.

            – You continually seem to imply that the amount of back-haul you need to add per user is an reducing exponential relationship. It isn’t. It’s actually closer to linear, once you get over the initial threshold which according to Simon Hackett is twice the intended burst speed.

            Now if you want to try and repeat the arguments you have been making go ahead, but unless you actually bring anything new to the argument you will not receive any-more replies from me. That includes the two other replies I really wanted to reply to, for the same reasons, with your conversations with Hubert and NPSF.

          • “How about you acknowledge that after 4 years, I might actually have a point.”

            Because after 4 years you’re repeating the same simple, flawed and unresearched points. If you had spent that time developing your argument instead of repeating it ad nausea you might have had more success.

            Fundamentally you think speed is free. This is false, and from this your entire argument collapses.

            “The data is there and you failed to refute it.”

            What data is where? I’ve never seen you show a single piece of data that even remotely supports your position. The best I’ve seen is your ‘math’ which has such laughable assumptions it makes me want to cry.

          • > Fundamentally you think speed is free. This is false, and from this your entire argument collapses.

            Considering that the same fibre and NTU is installed in all homes and it is only a software setting, then yes the speed is free.

            > What data is where? I’ve never seen you show a single piece of data that even remotely supports your position. The best I’ve seen is your ‘math’ which has such laughable assumptions it makes me want to cry.

            So provide some of your own. You’ve had plenty of time.

      • Essentially the cost for RSPs is driven by the amount of data that it is downloaded at, not the speed

        Quota has minimal relation to the costs and technical implementation of a network. If users downloaded at a constant rate continously your argument may be valid, but they don’t, they burst, and they have peak times.

        ISPs have to determine the amount of CVC of backhaul they purchase bassd upon the peak bandwidth usage of their customers, not how much quota they have.

        Until you understand this fundamental principle your arguments are fatally flawed.

        No wonder NPSF decided not to respond to your “data”, it’s like arguing with someone who thinks a lack of traffic lanes causes traffic jams and adding another lane will “fix” it, on the surface he’s partially correct but he completely ignores or fails to understand the principles behind traffic behaviour.

        He’s facing the problem with a solution that works with low traffic volumes but that solution will have diminishing returns. Similarly quota has diminishing returns with faster speeds. This is because content access becomes more bursty.

        Quota then becomes a pure profit driver. At higher speeds the cost per megabit actually goes down as you can assume your users will have faster transactions.

        By introducing a flat speed tier you shoot NBNCo in the foot, because not only have to removed the only method of cross subsidation they can utilise, you reduce the flexibility of the product offering. Effectively you remove the artificial scarcity, which for users is good, but for the network to pay off its debts in full?

        Removing speed tiers will only work if coupled with an injection of a subsidy from the government, rather than bonds.

        • No need to add additional lanes to fix traffic jams, just increase the speed. Simple.

          four times as many cars can go down a lane at 100kms/hr than at 25kms/hr
          traveling at 25kms/hr also introduces extra delays due to the time to stop and start, so may effectively only be 12kms/hr

        • > Quota has minimal relation to the costs and technical implementation of a network. If users downloaded at a constant rate continously your argument may be valid, but they don’t, they burst, and they have peak times.

          Wrong. Speed tiers actually make sense if users download at a constant rate because they provide the constraint on how much impact a user can have on a network.

          > ISPs have to determine the amount of CVC of backhaul they purchase based upon the peak bandwidth usage of their customers, not how much quota they have.

          Correct, except that the amount that customers can download is not limited by speed it is limited by quotas. A 5TB quota provides less than 23 minute of full speed access a day.

          > Until you understand this fundamental principle your arguments are fatally flawed.

          Maybe it might be you who is lacking in understanding.

          > Similarly quota has diminishing returns with faster speeds. This is because content access becomes more bursty.

          What you are ignoring is that when you have a large customer base, this bursty behaviour actually averages out nicely. Exactly the same problem exists on ADSL2+ networks now and RSPs manage with out too many issues.

          > By introducing a flat speed tier you shoot NBNCo in the foot, because not only have to removed the only method of cross subsidation they can utilise, you reduce the flexibility of the product offering. Effectively you remove the artificial scarcity, which for users is good, but for the network to pay off its debts in full?

          So your argument is that is is good for consumers to be ripped off (especially the poor) to make the financial models of the NBN slightly less risky and that we should all be happy with a fibre network where 50% connect at 12/1Mbps? The end result is we have a network that is more vulnerable to competition from wireless and fails to deliver the real benefits to half the population?

          > Removing speed tiers will only work if coupled with an injection of a subsidy from the government, rather than bonds.

          I disagree. Removing the speed tiers will have two effects: more customers will sign up, because the 4G networks are no longer faster than the base 2 speed tiers. Secondly, because people have access to faster speeds many will experience new services, start using them and move to higher quota plans.

          • Wrong. Speed tiers actually make sense if users download at a constant rate because they provide the constraint on how much impact a user can have on a network.

            And Quota’s don’t provide a constraint on how much impact a user can have on the network. I think you need to understand something about network design.

            If you were designing a residential broadband network for me we would likely have this conversation at some point:

            I want you do the following: take quotas, transfer caps, whatever you want to call them, then forget them. Remove them from your vocab entirely. Now once you have done that, design the network, then hand the completely design over to marketing department. They will then impose quotas, and your job as the engineer will be to make sure the network keeps running smoothly, and tell marketing when we’re over or under utilising the network so they can adjust the quotas.

            Correct, except that the amount that customers can download is not limited by speed it is limited by quotas. A 5TB quota provides less than 23 minute of full speed access a day.

            Artificially. Quotas are imposed after the fact.In reality the amount of data a single user can download is determined by how much the network is being utilised at any given moment. A high quota user who only uses their connection in off-peak times costs less than a high quota user who uses their connection only in peak times. The former can just be tacked on with no need for backhaul provision, the later requires the RSP to buy more backhaul.

            Maybe it might be you who is lacking in understanding

            No, I understand contention, peak loading, the difference between PIR and CIR perfectly. I also understand that how much someone has downloaded is a useless metric for determining how much a problem a user is. It is that last point you seem to have trouble with.

            What you are ignoring is that when you have a large customer base, this bursty behaviour actually averages out nicely. Exactly the same problem exists on ADSL2+ networks now and RSPs manage with out too many issues.

            No I’m not ignoring that. At all. And yes, RSPs do manage find without to many issues, what makes you think that adding speed tiers will create an issue?

            So your argument is that is is good for consumers to be ripped off (especially the poor) to make the financial models of the NBN slightly less risky and that we should all be happy with a fibre network where 50% connect at 12/1Mbps?

            Yes. That is exactly it. Except the part where only 50% connect at 12/1Mbps and the part where people get ripped off. There is already ample evidence that the estimates made by NBNCo on usage were actually very conservative.

            The end result is we have a network that is more vulnerable to competition from wireless and fails to deliver the real benefits to half the population?

            Depends on what you consider the benefits. You seem obsessed with the headline speed. There are other benefits. I have pointed those out to you before.

            Okay Mathew, food for thought here:

            My issue with you is not that you have a problem with the fact the minimum speed is being set at 12Mbps and that there are speed tiers.

            My issue with you is that you have provided no viable solution to the problem. All of your posts, and I mean all of them, highlight the issues with the NBN, and only give a small teaser of a possible solution.

            It’s all well and good saying that what NBNCo are doing are bad, but then you try and justify why there decision is bad with a fundamental misunderstanding of how running a large scale network actually works in reality. This fundamental misunderstand causes myself and others to get very frustrated at you, but doesn’t change the fact you’re right:

            – Speed tiers, in particular 12Mbps, are kinda slow and don’t allow you take full advantage of the network.

            Okay, with that out of the way. NBNCo needs to introduce them as a method of cross subsidisation. Until you provide another method, or a way to reduce the required ARPU to pay off the debts NBNCo will be amassing, and let me remind you NBNCo needs an ARPU of around $100 by the time the network is complete, your argument at best is pointing out readily observable flaws, and at worst is flat out bitching.

          • > And Quota’s don’t provide a constraint on how much impact a user can have on the network. I think you need to understand something about network design.

            Sure for a short period of time a person can have an impact on the network, but then shaping kicks in and the user is no longer a problem.

            > If you were designing a residential broadband network for me we would likely have this conversation at some point:
            > I want you do the following: take quotas, transfer caps, whatever you want to call them, then forget them.
            > Remove them from your vocab entirely. Now once you have done that, design the network, then hand the completely design over to marketing department. They will then impose quotas, and your job as the engineer will be to make sure the network keeps running smoothly, and tell marketing when we’re over or under utilising the network so they can adjust the quotas.

            Interesting idea, but it doesn’t work that way. Throughput is a fundamental part of network design. A single user on 1Gbps without quota can download 324TB/month. To design a network with that assumption would be to over-engineer it to a point where it would be uneconomic.

            > Artificially. Quotas are imposed after the fact.In reality the amount of data a single user can download is determined by how much the network is being utilised at any given moment. A high quota user who only uses their connection in off-peak times costs less than a high quota user who uses their connection only in peak times. The former can just be tacked on with no need for backhaul provision, the later requires the RSP to buy more backhaul.

            > No, I understand contention, peak loading, the difference between PIR and CIR perfectly. I also understand that how much someone has downloaded is a useless metric for determining how much a problem a user is. It is that last point you seem to have trouble with.

            Internode had a plan called flat rate that attempted to address this. It failed because not enough ‘light’ users saw the value.
            However in terms of network design, one really only considers peak demand. You will note that elsewhere I’ve pointed out that off-peak quotas could grow substantially because of the extra bandwidth available to support peak demand.

            > No I’m not ignoring that. At all. And yes, RSPs do manage find without to many issues, what makes you think that adding speed tiers will create an issue?

            So your argument is that is is good for consumers to be ripped off (especially the poor) to make the financial models of the NBN slightly less risky and that we should all be happy with a fibre network where 50% connect at 12/1Mbps?

            > Yes. That is exactly it. Except the part where only 50% connect at 12/1Mbps and the part where people get ripped off. There is already ample evidence that the estimates made by NBNCo on usage were actually very conservative.

            It doesn’t matter if the estimates are conservative or not … it is the failure to

            > The end result is we have a network that is more vulnerable to competition from wireless and fails to deliver the real benefits to half the population?
            Depends on what you consider the benefits. You seem obsessed with the headline speed. There are other benefits. I have pointed those out to you before.

            To an average person with low demands, it is just bits being transferred. As long as it works most of the time that will be fine.

            > My issue with you is not that you have a problem with the fact the minimum speed is being set at 12Mbps and that there are speed tiers.
            > My issue with you is that you have provided no viable solution to the problem. All of your posts, and I mean all of them, highlight the issues with the NBN, and only give a small teaser of a possible solution.

            WRONG. I have always proposed adjusting the numbers so that AVC is a smaller proportion of fees and CVC is a higher component. NBNCo intend to head this way in the future. That information is already in posts in this thread, and earlier posts.

            Heavy downloaders will initially have less peak quota, but most likely more off-peak quota. Those people who really need to download TBs of data can schedule it for off-peak.

            > Okay, with that out of the way. NBNCo needs to introduce them as a method of cross subsidisation.

            You keep talking about cross-subsidisation. However I don’t see that it is fair that people who put strain on the network are subsidised by those people who want a fast connection occasionally.

            That $100 ARPU figure is what 4 times the current ARPU do you have a reference for it?

          • Sure for a short period of time a person can have an impact on the network, but then shaping kicks in and the user is no longer a problem.

            Considering most users only use about 15% of their quota this isn’t the case. See my other earlier reply.

            Interesting idea, but it doesn’t work that way. Throughput is a fundamental part of network design. A single user on 1Gbps without quota can download 324TB/month. To design a network with that assumption would be to over-engineer it to a point where it would be uneconomic.

            No where in that statement did I imply you should provision 1:1 connections. At all. Not once.

            I implied that you, the network designer, should not be concerned with their usage amount, but instead their usage patterns. You design the network around usage patterns, not usage amount.

            Internode had a plan called flat rate that attempted to address this. It failed because not enough ‘light’ users saw the value

            A flat rate per gigabyte does not address this issue.

            However in terms of network design, one really only considers peak demand. You will note that elsewhere I’ve pointed out that off-peak quotas could grow substantially because of the extra bandwidth available to support peak demand.

            This is exactly what I have been trying to tell you. Peak demand is not related to the purchased quota is the part you seem to be failing to grasp. Repetitively.

            To an average person with low demands, it is just bits being transferred. As long as it works most of the time that will be fine.

            With that argument the fact that quotas are imposed on mobile connections about 1/10th the amount on the NBN will trump. 4G and 3G connections have similar characteristics to ADSL2 in terms of peak speed performance, and even, especially with 4G, latency, and yet, despite this, ADSL2 growth is still growing. This is despite the fact they probably won’t use all their quota, as per that ZDnet article I linked.

          • WRONG. I have always proposed adjusting the numbers so that AVC is a smaller proportion of fees and CVC is a higher component. NBNCo intend to head this way in the future. That information is already in posts in this thread, and earlier posts.

            But that proposal doesn’t address any of the issues presented to you adequately. Like our concern that it will adversely affect low income families. Your respond to that has been “but it won’t raise prices”, with no further discussion. Instead you turn it round and try to explain to us again why you think the NBNs current tiered structure is flawed, using flawed logic, and bad assumptions.

            Heavy downloaders will initially have less peak quota, but most likely more off-peak quota. Those people who really need to download TBs of data can schedule it for off-peak.

            Like Exetel’s “Unlimited off-peak” offering. It seems the market has already provided what you propose.

            You keep talking about cross-subsidisation. However I don’t see that it is fair that people who put strain on the network are subsidised by those people who want a fast connection occasionally.

            I honestly don’t care what you think is “fair”, while NBNCo needs to cross-subsidise, there is going to people who miss out because of this.

            Do you think it is “fair” for a regional ADSL connection to cost the same as a metro one despite the fact the regional one costs many times more to install?

            Do you think it is “fair” that a considerate user who limits his bandwidth during peak times and ensures his connection doesn’t adversely affect the network should pay more than a user who blindly downloads at all hours of the day and expects to get the full line speed of his connection?

            Cross-subsidisation is “unfair”. The point is to distribute costs between different products.

            Now, unless you can present a way to cross-subsidise that you consider “fair”, can practically be implemented, and address any concerns we might have with it instead of complaining about how “unfair” the current method is, then my advise to you is:

            Build a bridge and get over it.

          • > But that proposal doesn’t address any of the issues presented to you adequately. Like our concern that it will adversely affect low income families. Your respond to that has been “but it won’t raise prices”, with no further discussion. Instead you turn it round and try to explain to us again why you think the NBNs current tiered structure is flawed, using flawed logic, and bad assumptions.

            We know for a fact that FTTP will provide zero benefit at 12/1Mbps when compared with competing technologies. We know that only the severely budget constrained would opt for the 12/1Mbps when $5/month will double the speed.

            > Heavy downloaders will initially have less peak quota, but most likely more off-peak quota. Those people who really need to download TBs of data can schedule it for off-peak.
            > Like Exetel’s “Unlimited off-peak” offering. It seems the market has already provided what you propose.

            Except that the market cannot move beyond NBNCo’s pricing constraints. Exetel’s unlimited off-peak is still limited by the selected speed. The jump from 12/1Mbps to 100/40Mbps is $30.

            > Do you think it is “fair” for a regional ADSL connection to cost the same as a metro one despite the fact the regional one costs many times more to install?

            Does it really cost more to install or are you talking about backhaul costs?
            This is cross-subsidisation to ensure all are given access to the same service regardless of location. It is not the poor subsidising a network that only the rich will fully benefit from.

            > Do you think it is “fair” that a considerate user who limits his bandwidth during peak times and ensures his connection doesn’t adversely affect the network should pay more than a user who blindly downloads at all hours of the day and expects to get the full line speed of his connection?

            I don’t see where speed factors into this. The considerate user has the option of a plan with a small peak quota and saving money. The “inconsiderate” user will have to pay for a larger quota and will be shaped. Perfectly fair.

            > Cross-subsidisation is “unfair”. The point is to distribute costs between different products.

            The point is that you want your cross-subsidisation to encourage the desired outcome. For the NBN that can only be fast speeds. Everything else is achievable with other technologies. Secondly the game changing uses of the network only occur at 100/40Mbps and faster.

            > Now, unless you can present a way to cross-subsidise that you consider “fair”, can practically be implemented, and address any concerns we might have with it instead of complaining about how “unfair” the current method is, then my advise to you is:

            I’ve already pointed it out, but you continue to ignore it. Bring forward CVC making an increased part of the revenue. It really is simple. A bit more risky, but very simple.

          • That $100 ARPU figure is what 4 times the current ARPU do you have a reference for it?

            I thought you said you READ the business plan? Let me find it for you. Ahh here we go, page 69.

            They expect an ARPU of about $40 in 2015 raising to around $100 in 2039.

      • “The ironic fact is that by introducing speed tiers, FTTN, HFC, 4G and even approaching half of ADSL2+ connections will be faster than half of NBNCo fibre connections according to the corporate plan”

        False. According to the product roadmap more than half have chosen speeds beyond what FttN, HFC, 4G and ADSL2+ connections can achieve so you must be reading required minimums not fact based statistics.

        • > False. According to the product roadmap more than half have chosen speeds beyond what FttN, HFC, 4G and ADSL2+ connections can achieve so you must be reading required minimums not fact based statistics.

          I was pretty sure that HFC offered 100Mbps down. NBN upload speed might be faster but that isn’t a huge concern for people who predominately consume content.

          Still holding on to the early adopter figures I see. As you are well aware NBNCo did not adjust the percentage of users predicted to connect at 12/1Mbps on fibre when they updated the Corporate Plan. NBNCo did adjust upwards the number of people connected at 100Mbps. This suggests that NBNCo still put significant weight on their analysis and think it is correct.

          • “I was pretty sure that HFC offered 100Mbps down. NBN upload speed might be faster but that isn’t a huge concern for people who predominately consume content.”

            Right, so people with slower upload speeds won’t be disadvantaged then? You cant have it both ways…

            “Still holding on to the early adopter figures I see.”

            Those are real world figures, not required minimums which look like are not going to happen. That’s a good thing btw. It means your fears were unfounded. Imagine that. You lost. NBNco won.

            “As you are well aware NBNCo did not adjust the percentage of users predicted to connect at 12/1Mbps on fibre when they updated the Corporate Plan. ”

            We’ve been over this. I told you long before the revised plan was even released that not much would change so the big question now is do you understand why? Yes/No?

            “This suggests that NBNCo still put significant weight on their analysis and think it is correct.”

            This statement suggests you don’t have a fucking clue what you are talking about. The numbers in the corporate plan are required minimums and conservative estimates, not a divine prophecy of things to come. Please try to use what little brains you have. This is not a difficult concept to grasp even for you. Try it, you might surprise yourself.

          • > Those are real world figures, not required minimums which look like are not going to happen. That’s a good thing btw. It means your fears were unfounded. Imagine that. You lost. NBNco won.

            Real world figures, doesn’t change the fact that many people in the areas still haven’t signed up and one would expect those people to either not connect or choose the cheaper plans.

            The people who are loosing are those who are going to be stuck on a 12/1Mbps connection helping to pay for people with 1Gbps.

            > We’ve been over this. I told you long before the revised plan was even released that not much would change so the big question now is do you understand why? Yes/No?

            They significantly revised the number of customers on 100/40Mbps, so if they believe the number of 12/1Mbps is going to be lower it would have been changed to improve the financials.

            > This statement suggests you don’t have a fucking clue what you are talking about. The numbers in the corporate plan are required minimums and conservative estimates, not a divine prophecy of things to come. Please try to use what little brains you have. This is not a difficult concept to grasp even for you. Try it, you might surprise yourself.

            Time and time again the numbers in the NBNCo Corporate Plan have shown to not be conservative estimates. The roll out is years behind the original schedule.

  2. That is not entirely true the other option is to have the same max speed but pay more for priority so you can pay less but your speed is up to 1Gbps and if you pay more you get a guaranteed speed tier 100Mbps or above and less for 12Mbps – 100Mbps

    • Introducing consumers to the concept of CIR will be difficult to impossible, and imposing priorities like that is very hard to achieve seamlessly.

      • And totally irrelevent. Consumers don’t care it they have a CIR of 100mb/s, because they are MOSTLY only going to be using their bandwidth for a fraction of a second.

        When they download data, they will most often be limited to the speed of the source, not the size of the pipe into their home.

  3. I do agree with the fact that it doesnt cost the NBNco significantly more to provide 100Mbs rather than 25Mbs, they may require more backhaul capacity if 100% of users could access higher speeds but other than that the line costs should be the same.

    If we really wanted to change up the industry lets make all Australian data free and open up 100Mbs as the standard speeds- peering traffic has no real reason to be counted as most ISP’s are part of IX Australia anyhow..

    the question then would be how much would we pay and how do ISP’s compete with each other and diferentiate their products :)

  4. There is also alot of Transitional stages that the both Industry and NBN are taking, so you need to take into account of old plans, really old plans, Telstra and Optus cable plans and alot of people making comparing with their current plan with the new NBN plans.

  5. I don’t disagree with him either but the multiple tiers do allow a lot of differentiation between retailers.

    I think also he underestimates the consumer’s willingness to shift between plans as needs, wants and means change.

  6. They need tiers so they can charge some people more. The NBN has to return on investment at 6-8%. It relies on some people paying more. To pay more you want more. People will pay more to get 100Gb at 100Mbit than at 12Mbit. How do you charge more if everyone wants 100Gb and everyone gets the same speed? Do they all pay the higher price? Some would prefer to pay less for less.

  7. Abolishing speed tiers at the NBN wholesale level would be reasonable. That would allow more flexibility to retailers to compete; some may offer “flat rate” plans without tiers, others may still choose to have tiers but they could be different tiers. Current enforcement of uniform tiers by the wholesaler doesn’t make much sense.

  8. “They need tiers so they can charge some people more. ”

    And they charge some people more so that they can charge some people less (it’s a nation building project, after all) – if one tier were in place the wholesale price would have to be (Hackett?) $9 more compared to the $24 lowest now… and as you said “Some would prefer to pay less for less.”…

    • > And they charge some people more so that they can charge some people less (it’s a nation building project, after all)

      If it is a nation building project, then you need to consider the social policy implications of speed tiers on those at the lower end of the socio-economic scale. Many people in this bracket (students, pensioners, etc.) will simply never be able to justify the costs of higher speed tiers, but they could benefit from the functionality.
      – Distance Education and Training
      – E-health
      – Delivery of social services
      – Social Interaction via Video Conferencing for mobility challenged persons

      There are a huge number of innovative services that could be delivered that could improve people’s lives. For example, Grandma having a 15 minute video conference with the Grandchildren. Video works because pre-school kids are visual and can show Grandma what they are doing. Everyone benefits and the parents are able to see Grandma and be re-assured about her health. If Grandma only has a 12/1Mbps the picture quality is poor and the experience is lacking.

      > if one tier were in place the wholesale price would have to be (Hackett?) $9 more compared to the $24 lowest now… and as you said “Some would prefer to pay less for less.”…

      The wholesale rates are numbers that came out of a spreadsheet based on estimates of take-up. One could easily argue that if you charged a minimal access fee (e.g. give away the razor blade handle) and charged for data (blades) that take-up would be higher. Increasing take-up from 70% to 80% would significantly improve the return. The reality is that those people who want to pay less will opt for cheaper mobile wireless, especially when combined with their mobile phone plan.

      If people have access to a service, then they can experience the benefits (e.g. video streaming, video conferencing, eHealth, etc.) and in all likelihood will increase their usage. However if someone is stuck on 12/1Mbps then many of these services are simply not possible, as explained by Quigley: “You certainly can’t do high-definition video service on a 1 megabits per second upstream — it’s impossible,” in Low-income users denied NBN benefits.

      • That’s all well and good, but what is your alternative means of cross subsidation? Quota? No, high bandwidth technology tends to be high usage.

        Arguing that the benefits of the technology is reduced and that it is all artificial scarcity doesn’t change the fact that NBNCo in its current form requires some kind of cross subsidation to be financially viable.

        The best we can do RIGHT NOW is to raise the minimum.

        • > That’s all well and good, but what is your alternative means of cross subsidation? Quota? No, high bandwidth technology tends to be high usage.

          Yes, higher speeds do mean people are likely to download more. However it doesn’t mean that everyone will download more and those people who only use email and the occasional video conference will do fine on 50GB.

          > Arguing that the benefits of the technology is reduced and that it is all artificial scarcity doesn’t change the fact that NBNCo in its current form requires some kind of cross subsidation to be financially viable.

          The current cross subsidation approach delivers the worst to everyone. Have you considered that off-peak quotas could grow substantially because of the increased CVC requirements during peak?

          The NBNCo Corporate Plan is at significant risk now that many people predicted to connect at 12/1Mbps will simply ditch their fixed connection and opt for faster 4G.

          > The best we can do RIGHT NOW is to raise the minimum.

          So we double the minimum to 25Mbps, and move from 1.2% to 2.5% of the top speed. Big improvement … not! Being happy with doubling the minimum would be stupidity.

          • Yes, higher speeds do mean people are likely to download more. However it doesn’t mean that everyone will download more and those people who only use email and the occasional video conference will do fine on 50GB.

            I’ve been over why quotas are a terrible metric. But there’s another problem : no retailers will be happy with being charged per gigabyte. Where is there room to differentiate products? How can they engage in competition? They can’t. You’ll do far more damage.

            The current cross subsidation approach delivers the worst to everyone. Have you considered that off-peak quotas could grow substantially because of the increased CVC requirements during peak?

            To the point that provides like Exetel are offering unlimited off peak? Because. I hadn’t considered that. /sarcasm

            The NBNCo Corporate Plan is at significant risk now that many people predicted to connect at 12/1Mbps will simply ditch their fixed connection and opt for faster 4G.

            Only insanely low usage users will do that due to tiny quotas, and you further assume that congestion on mobile 4G will not increase or have any affect on performance. My 4G connection dropped from an average of 35Mbps to 15Mbps in the past 6 months as Telstra signed more people.

            NBNCo have predicted this movement as well remember. That was all nicely wrapped up in their corporate plan too. The one you have claimed to read. You can’t claim that evidence that points to one statistic being conservative, that is the number of 12Mbps users and call that bullshit and then turn around and say, with evidence to the contrary, that the estimate for the number of wireless only households was also conservative.

            Either you acknowledge that the estimates are conservative and rely on real world data, or you stick steadfast to the estimates until you have more compelling evidence for or against. There is no middle ground here where you get to pick what aspects of the evidence you ignore in order to cherry pick a position.

            I’ve seen that tactic before in people who try to argue that wireless is taking over the market. It never ends well for their creditability.

            2.5% of the top speed. Big improvement … not! Being happy with doubling the minimum would be stupidity.

            It is actually, it’s a thrice fold improvement over the current average Internet speed in Australia as opposed to the current 1.5x. Just because I doesn’t use the technology too the full doesn’t mean it won’t have improvements. You seem to be assuming it’s a) all of nothing and b) that headline speed is the only benefit.

          • > I’ve been over why quotas are a terrible metric. But there’s another problem : no retailers will be happy with being charged per gigabyte. Where is there room to differentiate products? How can they engage in competition? They can’t. You’ll do far more damage.

            There is vastly more room to engage in competition. Plans like Internode’s flatrate become plausible.

            The current cross subsidation approach delivers the worst to everyone. Have you considered that off-peak quotas could grow substantially because of the increased CVC requirements during peak?
            To the point that provides like Exetel are offering unlimited off peak? Because. I hadn’t considered that. /sarcasm
            > Only insanely low usage users will do that due to tiny quotas, and you further assume that congestion on mobile 4G will not increase or have any affect on performance. My 4G connection dropped from an average of 35Mbps to 15Mbps in the past 6 months as Telstra signed more people.

            6GB month is now easily on 4G and that excludes social media traffic.

            > Either you acknowledge that the estimates are conservative and rely on real world data, or you stick steadfast to the estimates until you have more compelling evidence for or against. There is no middle ground here where you get to pick what aspects of the evidence you ignore in order to cherry pick a position.

            I think the NBNCo estimates are surprisingly accurate for a corporate plan. I do question if the numbers add up for the 12/1Mbps users given the pricing for 4G.

            > I’ve seen that tactic before in people who try to argue that wireless is taking over the market. It never ends well for their creditability.

            I’m not arguing that wireless will take over the market. Just that there is a significant number of low usage people in Australia (50% connecting at 12/1Mbps on fibre) who will struggle to see any benefit in an NBN which is slower than their 4G connection. What you need to consider is that these people already have a mobile phone and the fixed line is an additional expense. $50+ month buys you a lot of extra quota.

            > Just because I doesn’t use the technology too the full doesn’t mean it won’t have improvements. You seem to be assuming it’s a) all of nothing and b) that headline speed is the only benefit.

            The revolutionary applications all require 100Mbps or faster to work. Below that speed you could use other technologies to provide the service.

          • Oh god, did you seriously did suggest that charging retailers per gigabyte will somehow be better for competition? You… you’re… oh god, I can’t say any-more without breaking the comments policy.

            Anyway:

            6GB month is now easily on 4G and that excludes social media traffic.

            That doesn’t address my point at all. Which was “you further assume that congestion on mobile 4G will not increase or have any affect on performance. My 4G connection dropped from an average of 35Mbps to 15Mbps in the past 6 months as Telstra signed more people.

            I’m not arguing that wireless will take over the market.

            No you’re not but you ARE overstating the important of wireless by ignoring the fact that the amount of quota you can get on a 4G connection is about 1/10th of what you can get on the NBN for the same money. Even accounting for the fact that people don’t use all of their quota, that’s still a significant difference. You’re ignoring the performance characteristics of 4G which will make it more undesirables over time.You’re ignoring that the early adopters of the NBN have opted for a faster speed tiers than NBNCo had predicted.

            The revolutionary applications all require 100Mbps or faster to work. Below that speed you could use other technologies to provide the service.

            Once again: You seem to be assuming it’s a) all of nothing and b) that headline speed is the only benefit.. What about reduced latency? What about more reliable consistent performance? Are these benefits irrelevant compared to the headline speed? No.

  9. interesting idea.
    Probably worth a thought, imaging getting charged for actual usage only, and throw on peak/off peak rates into that mix as well…

  10. If De Ridder beleived this was truly workable, why didnt he get Telstra to do the same thing when he was there? Telstra is the master of artificial limits. Had he got rid of them in his Telstra days, customers would have flocked to Telstra – following his logic. That is of course unless it doesnt quite work that way ;)

    • > If De Ridder beleived this was truly workable, why didnt he get Telstra to do the same thing when he was there? Telstra is the master of artificial limits. Had he got rid of them in his Telstra days, customers would have flocked to Telstra – following his logic. That is of course unless it doesnt quite work that way ;)

      There are a number of reasons why Telstra created ADSL speed tiers:
      – ADSL quota was initially unlimited (but not for long) and then it went to 3GB
      – stopped cannibalisation of other services (e.g. frame relay, foxtel, etc.)
      – Other competitors to Telstra didn’t exist

      Finally Telstra didn’t need customers to flock to them. For broadband you were a retail customer or a wholesale customer and in some scenarios Telstra charged more for wholesale than retail.

  11. I agree with what Paul Budde said. As I’ve said before 12/1mbps is pointless on fibre the only reason it exists is because of the wireless and satellite plans but like Budde mentioned since those speeds have doubled now the lowest on fibre should be 25/5mbps.

    I think 100/40mbps will naturally become standard by the time the proper NBN is completed (assuming it is) if not sooner so I doubt the slower plans will even exist by then.

    Just wondering though how would people who think we don’t need more speed respond to this if 100/40mbps was the only speed available? What if they argue they want to pay less for a 50/20mbps plan? Shouldn’t they have that option? Or do we force them to pay for a service they believe they don’t need?

    • As the cheapest NBN plan is about $30 a month, this does make it a little hard for pensioners to pay for.
      As a service that is becoming more and more essential, I would expect plans that cost much less than this to be available. In proportion, they would have to offer less.
      How do you do that with something like an internet only service? Cut the quota to say 5Gb a month? Still won’t be cheap enough. Cut the speed perhaps?

      • > As a service that is becoming more and more essential, I would expect plans that cost much less than this to be available. In proportion, they would have to offer less.

        The cost of the providing the service also falls.

        > How do you do that with something like an internet only service? Cut the quota to say 5Gb a month? Still won’t be cheap enough. Cut the speed perhaps?

        There is a point at which it doesn’t make sense to have a plan with a smaller quota, unless you want to charge for excess data. I personally think plans with excess data should be banned because they prey on the vulnerable. RSPs make an assumption as to the average data usage at each point.

        Google Fibre in Kansas provides an interesting example:- 1Gbps at $70/month no quota or $300 install fee and no ongoing charges for 5/1Mbps. Interestingly Google is installing direct fibre (no GPONs) so it is technically a better model than NBNCo. I think we could do worse than to simply sack NBNCo and ask Google to build the network.

        • Google will say no. The network is too big, the CAPEX too high. There is a huge difference from giving fibre to a single city and giving fibre to 93%.

          Unless we artificially reduce the CAPEX NBNCo has no choice but to cross subsidise somehow.

          • > Google will say no. The network is too big, the CAPEX too high. There is a huge difference from giving fibre to a single city and giving fibre to 93%.

            Are you sure Google would say no? It has been estimated based on the Kansas City rollout that it would cost $120 billion for Google to extend the rollout across the USA. If the government offered Google a $30 billion dollar loan at 7%, are you so sure they wouldn’t be interested?

            > Unless we artificially reduce the CAPEX NBNCo has no choice but to cross subsidise somehow.

            Are you talking about the cross subsidisation between fibre, wireless and satellite or cross subsidising between low and heavy users? If it is low and heavy data users then I don’t see the need for cross subsidisation. Heavy quota users already receive significant discounts for higher quota plans on a $/GB basis. Higher quota users are those who put the strain on the network. Higher quota users will benefit from more people paying a nominal access fee ($20/month) rather than those people on 12/1Mbps connections abandoning fibre for faster 4G.

          • That estimation was based upon the assumption that the costs will not be affected by accelerating the time frame and that there is adequate dark fibre assets throughout America to utilise.

            In other words it was done by an economist, not an engineer. Am engineer would know increasing the rollout time frame adds to costs. An engineer would know that dark fibre assets are limited to major cities.

            And an engineer would also know the situation is much worse here in Australia. Our dark fibre footprint is virtually non-existent in regional areas. NBNCo has been told to maximise rollout schedule to as fast as possible, which they estimate to be at 6000 premises a day at peak which they have to sustain for about 7 years.

            The NBN will cost more per capita because of this. I don’t even know why you’re even suggesting this. Google also has considerable dark fibre assets in America it can take advantage on. It has its own backhaul network. It’s own infrastructure. Rolling out to one populated city in America is one thing, but 100% of Australia?

            They might, if given incentive, do the capitals. But that’s about it. And that’s only if we give them an effective monopoly on fibre access. Because I can’t see them wanting to share their infrastructure.

            I love Google, but I don’t think giving them that sort of market control over Australia is wise.

  12. It is all very well for us techies with enough Edith/David portraits to pay whatever the bill is for our wild use of the ‘net.

    OTOH for a single-parent (or even two-parent) family struggling a bit with a couple of school kids it’s pretty hard to cop a bill that wipes out food for a week.

    An entry level price with decent speeds and shaping is a responsible way to provide access for the kids without bankrupty looming if the kids use heaps. They can’t be expected to know what their project research or even some game is costing.

    It’s enough for the seniors too and many of the 80+ gang are hitting the ‘net but not needing to risk their pension on some open-ended charge.

    But then, an economist taking humanity into the calculation? His concern is all about taking more money.
    He says so.
    “”They need to recover their costs efficiently and one way of doing that is by price discrimination, but they’ve got the wrong idea. People are not going to be prepared to pay more for speed. They don’t do it now,” he argued.”

    Oh yes they do!

    I pay close to $100 a month now but I’d hate to find that my bill suddenly became twice ot thrice that because someone sent a huge bunch of data every day while I was away for a couple of weeks.

  13. I wonder if this guy thinks the ‘no speed tiers’ argument will still be applicable when the network switches on 1 gbps. Does he think a few dozen 1 gbps connections sharing 2.44 gbps of bandwidth is a good idea?

    • Yes, the GPON is the limiting factor. So, there needs to be a limit (speed tiers factored by cost), to prevent over utilisation of the GPON.

      Understandably, when NBNCo make 1Gb connections available (10Gbps XG-PON?), and later with symmetrical speeds, they will have to replace the hardware at the end of the fibre. Resultant costs would be passed to the consumer through the higher cost of these new speed tiers.

      • > Yes, the GPON is the limiting factor. So, there needs to be a limit (speed tiers factored by cost), to prevent over utilisation of the GPON.

        The false assumption that you are making is that NBNCo is selling committed information rate (CIR) AVC as standard. In fact NBNCo are selling AVC as peak information rate (PIR) and business may purchase CIR AVC at an additional cost. For Traffic class 1, AVC CIR 5/5Mbps will cost $334/month + AVC PIR charge (minimum of $27).

        The reality is that the network is shared between all the users because not all users will be online at the same time. All RSPs base their entire networks on this. Premium RSPs ensure that sufficient bandwidth is available to avoid congestion, while budget RSPs have congestion on their networks.

        > Understandably, when NBNCo make 1Gb connections available (10Gbps XG-PON?), and later with symmetrical speeds, they will have to replace the hardware at the end of the fibre. Resultant costs would be passed to the consumer through the higher cost of these new speed tiers.

        GPON speed (2.5Gbps) is the limiting factor, however there is an upgrade path to GPON10 and beyond. NBNCo are already paying for this. I would be surprised if NBNCo don’t start installing GPON10 as standard in a few years time when prices fall.

        The second factor is that if sufficient data is being downloaded to flood the GPON and cause congestion, then NBNCo will be rolling in cash from CVC fees and the upgrade to GPON10 will be trivial to justify.

    • this GPON/Speed tiers is nothing but fear mongering. People will actually use their connection for LESS time at the higher speed to download the same amount.

      The limiting factor will be at the server end anyway.

      Streaming video (for example) isn’t going to saturate a 100mb/s or even a 12mb/s link.

      Very few applications are going to be downloading at full speed for any considerable length of time.

      Strangely since Telstra and Optus started providing 100MB/S cable connections, at no additional change, we haven’t seen their network deteriorate (any more than it had before the higher “default” speeds were introduced.

      • Actually you’re both wrong, the limiting factor is backhaul. If NBNCo “unleashed” their network and charged retailers a flat rate per active connection plus POI space retail and port connectivity, effectively the bare minimum, the backhaul that the RSP choose to provide would be the limiting factor.

        Unfortunately NBN will never do this, they will always introduce some from or artificial scarcity into network. Abolishing speed tiers will only place the scarcity purely on “priority” rather than CVC.

        • > Actually you’re both wrong, the limiting factor is backhaul. If NBNCo “unleashed” their network and charged retailers a flat rate per active connection plus POI space retail and port connectivity, effectively the bare minimum, the backhaul that the RSP choose to provide would be the limiting factor.

          That pretty much describes ADSL2+ where providers have installed their own DSLAMs, it was also Simon Hackett’s preferred position at one point in time (it may still be). The limiti

          > Unfortunately NBN will never do this, they will always introduce some from or artificial scarcity into network. Abolishing speed tiers will only place the scarcity purely on “priority” rather than CVC.

          I don’t see why bandwidth (CVC) is artificial scarcity.

          • That model worked because the ULL price paid effectively only represented the copper maintenance cost. With NBNCo’s model you can’t do that because NBNCo’s aren’t providing dark fibre.

            It’s artificial because the fibres are capable of far more than the CVC charges imply. It’s about 10x what an RSP would pay for backhaul now.

  14. So the fact I’m paying more for ADSL1 than I would for ADSL2+ due to my RIM not being upgraded is not down to artificial price manipulation by Telstra.

    News to me.

  15. It would be workable at the wholesale level but you will still end up with limits and speed tiers at the retail level. This artificial limits are in place because people understand them, try explaining contention ratios and the like to average Joe and all they know is “my internet is slow at 6pm”, and “web-browsing is slow at lunch at work when everyone facebooks”. It will likely reduce the cost of CVC charge as ISP will need to buy more which will make real business connection on the NBN cheaper. Net result will be more expensive cheap plans and cheaper expensive plans.

    • > It would be workable at the wholesale level but you will still end up with limits and speed tiers at the retail level.

      I don’t see any reason for speed tiers at the retail level. We would likely have peak / off-peak and speed limiting but we have that with speed tiers anyway.

      There might be an opportunity for plans like Internode’s flatrate to be re-introduced. However I’m doubtful because any plan approaching “unlimited” fails to attract sufficient low usage customers to make it viable for the average user. In fact for budget RSPs the absence of speed tiers will be a benefit because they won’t have issues with selling a 100Mbps service that doesn’t perform during peak time.

      > This artificial limits are in place because people understand them, try explaining contention ratios and the like to average Joe and all they know is “my internet is slow at 6pm”, and “web-browsing is slow at lunch at work when everyone facebooks”.

      Still going to happen on the cheaper providers with speed tiers. In fact it will be worse because people will be paying for 100/40Mbps and still suffering from congestion.

      > It will likely reduce the cost of CVC charge as ISP will need to buy more which will make real business connection on the NBN cheaper. Net result will be more expensive cheap plans and cheaper expensive plans.

      I don’t think there will be much change at the bottom end. Data will be slightly more expensive because the drop in CVC $/Mbps won’t totally compensate for the extra CVC required. Unlimited plans will struggle to be viable (324TB/month is a lot of data) but 12/1Mbps is only 4TB of data a month anyway. However people with faster connections will download more and that will reduce CVC prices more quickly. Plus if access fees are cheaper then more people will connect.

  16. This is wrong. From M Janda’s article
    “Mr de Ridder says NBN Co …
    It would be much more simple to basically charge per gigabyte of data delivered over the NBN network, just as you do with gas and water. It is a utility now. It’s a public service,”
    One example of fixed charges + usage charges (Queensland)
    http://www.urbanutilities.com.au/Your_account/Water_and_sewerage_charges/
    Fixed access charge + usage charges
    “Water Access Charge
    This is a fixed charge that contributes to the cost of transporting and testing the water as well as the maintenance of the distribution network that delivers water to
    your home. $41.79 per quarter”

    By what logic does a network with almost all fixed costs charge solely on usage? You can do it, but you expose to serious volume risk if your forecast even gets a bit out. And then there is question setting the appropriate charge per Gb based on the forecast – see the electricity for this. You understate the forecast usage ==> bumps up rate ==> rake in excessive profit. Wholesale/commercial gas & electricity charging more often than not has capacity charges with the commodity[usage] charge making a relatively small contribution. Is there a breakdown of the cost of network infrastructure versus incremental data costs?

    [Next bit on shakier ground on the technicalities – not my field]
    This rests on the assumption that there is sufficient capacity to service everyone simultaneously at the advertised speed. If people are told it is 100Mb/s down, they will expect it. All of them, at 4-10pm at once. From the USA, Europe & Asia. Across finite capacity intercontinental cables.

    • > By what logic does a network with almost all fixed costs charge solely on usage? You can do it, but you expose to serious volume risk if your forecast even gets a bit out. And then there is question setting the appropriate charge per Gb based on the forecast – see the electricity for this. You understate the forecast usage ==> bumps up rate ==> rake in excessive profit. Wholesale/commercial gas & electricity charging more often than not has capacity charges with the commodity[usage] charge making a relatively small contribution. Is there a breakdown of the cost of network infrastructure versus incremental data costs?

      The reality is that data costs are a small component. However if data was free then access prices would be higher and a small percentage would simply download constantly even if it was saved to /dev/null.

      The volume risk is where it comes back to a question of “nation building or not”. NBNCo have the backing of the government which means they can take a riskier approach (already seen by the lower ROI than commercially acceptable) because the government can and is supporting NBNCo through the initial network build years.

      Very few people would dispute that data usage will grow, especially with faster speeds enabling people to do more. The key risk for NBNCo is take-up. The lower the access price, the more people will take up access and once they are on the network experiencing the benefits of 1Gbps, the people will download more. If people don’t connect then NBNCo’s income is zero, but the bulk of the cost has already been spent in laying the fibre to the premise.

      • Lower access price also means lower ARPU.

        Lower ARPU results in long payback time frame.

        Longer time frame is higher risk and lower ROI.

        We’re already at 7% ROI and 20 years. This is very risky. As I have said before, unless the government is willing to give a subsidy injection we don’t have many options.

  17. Actually, the premise that the speed tiers are completely artificial is false.
    They are there to prevent people from ordering speeds that over utilise the capacity of the GPON.
    Considering the GPON only provides 2.4Gbit to be shared over the 32 connections, you can’t just let the (initial 16) premises connect at max speed, and hope there’s enough bandwidth for all.

      • But what is a reasonable alternative?

        First in best dressed? “I’m sorry, but we’ve already allocated 2.4 out of the 2.44Gbps on your local connection. We only have 40Mbps available for you. Is that fast enough for you sir?” Then you look outside, and realise the old granny across the street has a connection already, and she only uses her 100Mbps for her email.

        I’m all ears for a better way, but IMO artificial barriers are the only way to keep enough bandwidth for those who REALLY want it, and leave less for those who don’t think they need it.

        • > First in best dressed? “I’m sorry, but we’ve already allocated 2.4 out of the 2.44Gbps on your local connection. We only have 40Mbps available for you. Is that fast enough for you sir?” Then you look outside, and realise the old granny across the street has a connection already, and she only uses her 100Mbps for her email.

          That simply isn’t the way it works. You are suffering from the common misconception that NBN AVC is committed information rate (CIR) but it is actually peak information rate (PIR). For example AVC CIR 5/5Mbps will cost $334/month + AVC PIR charge (minimum of $27). More than double the price of 1Gbps AVC PIR.

          The network already has contention. This is standard practice. If Grandma has 1Gbps and 5GB of quota then her effect on the network won’t be noticeable. However you should be more worried about people with 5TB quotas.

          > I’m all ears for a better way, but IMO artificial barriers are the only way to keep enough bandwidth for those who REALLY want it, and leave less for those who don’t think they need it.

          Quotas are the barrier. Speed barriers aren’t needed. Quotas enable people to purchase the data they need.

          • Just because quota is viable in ADSL does nor mean it’s viable in GPON.

            Mathew, you have been commenting here for a while, so let me be frank, until you can address this point, or provide an alternative means of cross subsidation, there is very little we can do.

            We can raise the minimum, but the market seems to be doing that for us.

          • > Mathew, you have been commenting here for a while, so let me be frank, until you can address this point, or provide an alternative means of cross subsidation, there is very little we can do.

            Why should those people who want faster speeds but small quotas have to subsidise those who want larger quotas?

            For example someone who wants to video conference for 30 minutes a month doesn’t require a large quota, but they do require fast speeds.

            It is the people with large quotas that put the strain on the network, therefore they should pay. I was looking at my electricity bill and 10% of the cost is access, while 90% is usage.

          • BECAUSE FASTER SPEEDS ARE MORE EXPENSIVE. Not quota. To provide higher burst capacity you need more CVC. To allow someone to download more you might require more CVC but often you don’t.

            I’ve said it plenty of times, quota is not directly proportional to backhaul or CVC.

          • I’ve said it plenty of times, quota is not directly proportional to backhaul or CVC.
            That seems to contradict the experience of those that have been managing networks in Austrralia for the last decade as far as I know.

          • No it doesn’t. I specifically said Directly Proportional. As in a linear relationship with a set constant. There is a proportional relationship but it isn’t linear.

          • I wish I was living in the world you guys are. Apparently if we have faster local speeds, somehow all the servers in the rest of the world are going to magically be able to deliver content to us at full speed.

          • *facepalm*

            Even if your users can’t use it all the time, buying enough backhaul lo provision for a 1Gbps connection is still likely going to cost more than provisioning for a 100Mbps connection. There is more to a network than the client and server you know right?

          • > Even if your users can’t use it all the time, buying enough backhaul lo provision for a 1Gbps connection is still likely going to cost more than provisioning for a 100Mbps connection. There is more to a network than the client and server you know right?

            True, but once you have brought 2Gbps because you have 2 customers with 1Gbps speeds, you can increase the speed of your low quota customers to 1Gbps, because they are on the network for such a tiny amount of time, that it doesn’t have any noticeable effect.

            The big cost is in jumping up to 1Gbps, but because the biggest cost is in laying the fibre, upgrading the endpoints from 1Gbps to 10Gbps doesn’t require charging 10x as much. Look at how international bandwidth prices continue to fall as capacity increases.

          • Yes, this is true, however, isn’t this exactly what NBNCo assumes in their business plan? They drop the AVC price over time under the assumption more and more people will be accessing a higher speed tier.

  18. I’m by no means an expert, but surely over-subscription is still a potential issue with a FTTH network? I’d imagine that in the future it’ll become less of an issue, but network congestion is still a possibility (however remote). If that’s the case, then speed tiers make logical sense, however an argument could be made that those tiers should be determined by retailers, not the wholesaler. At the very least, it will allow retailers greater flexibility in the services they offer.

    • > I’m by no means an expert, but surely over-subscription is still a potential issue with a FTTH network? I’d imagine that in the future it’ll become less of an issue, but network congestion is still a possibility (however remote).

      Congestion is theoretically possible, however NBNCo are already planning for the upgrade to GPON10 to resolve this. The revenue from increased CVC revenue will more than pay for the costs of upgrading to GPON10.

      > If that’s the case, then speed tiers make logical sense, however an argument could be made that those tiers should be determined by retailers, not the wholesaler. At the very least, it will allow retailers greater flexibility in the services they offer.

  19. Agree 100%. Speed tiers are a carryover from gouging telco’s like Telstra. There is no technical reason to charge less for a slower speed, or more for a faster speed.

    Bean counters have decided they can make more money this way. Isn’t this supposed to be a GOVERNMENT initiative to provide HIGH SPEED broadband to all?

    No doubt, this will go away eventually, like it has on HFC cable plans where telco’s now offer the higher speeds FOR FREE as an enticement to take a plan with them.

    • > Agree 100%. Speed tiers are a carryover from gouging telco’s like Telstra. There is no technical reason to charge less for a slower speed, or more for a faster speed.

      It is also a carryover from other countries where usage is unlimited. We have quotas in Australia and because of the data that can be downloaded over 1Gbps combined with the Tragedy of the Commons, it is extremely unlikely that quotas will ever be removed. The best hope is that as people download more the $/MB will drop.

      > Bean counters have decided they can make more money this way. Isn’t this supposed to be a GOVERNMENT initiative to provide HIGH SPEED broadband to all?

      It is more that bean counters are taking the “safe” approach in their modelling. A revenue model which is based primarily on usage has much higher sensitivity to changes in usage patterns. However almost everyone would argue (as do NBNCo) that usage will continue to grow strongly.

      > No doubt, this will go away eventually, like it has on HFC cable plans where telco’s now offer the higher speeds FOR FREE as an enticement to take a plan with them.

      Except that NBNCo are a monopoly provider that won’t face competition. The only reason speed tiers were removed on ADSL was that Internode constructed their own DSLAMs in exchanges and released plans without speed tiers. It took many years for Telstra to enable ADSL2 speeds and finally remove speed tiers from ADSL.

      NBNCo won’t have any incentive to remove the speed tiers and if take-up is lower than forecast, then usage will also be lower. At that point it would take an extremely brave CEO to move to a riskier volume based pricing model.

      • It is also a carryover from other countries where usage is unlimited. We have quotas in Australia and because of the data that can be downloaded over 1Gbps combined with the Tragedy of the Commons, it is extremely unlikely that quotas will ever be removed. The best hope is that as people download more the $/MB will drop.

        Charging per transfer is arguably more profit motivated than per PIR since only the later accurately represents costs. ISPs as they are getting more entrenched in the American market are introducing quotas in order to maximise profit.

        It is more that bean counters are taking the “safe” approach in their modelling. A revenue model which is based primarily on usage has much higher sensitivity to changes in usage patterns. However almost everyone would argue (as do NBNCo) that usage will continue to grow strongly.

        Can you blame them for being conservative? It’s a multi billion dollar project. High enough risk as it is. Reducing risk is important to them. They have even reverted to using older field tested technology where possible. I’d like give you $27 billon dollars and see you take a high risk investment profile with it, see how long you can sleep at night for…

        Now, if the conservative profile performs better than expected I see no reason they won’t take on more risk by dropping AVC prices or removing the lower speed tiers.

        Except that NBNCo are a monopoly provider that won’t face competition. The only reason speed tiers were removed on ADSL was that Internode constructed their own DSLAMs in exchanges and released plans without speed tiers. It took many years for Telstra to enable ADSL2 speeds and finally remove speed tiers from ADSL.

        Do you have any evidence that suggests competition was the only contributing factor?

        Further you seem to forget that the ACCC allowed competition between mandating ULL, without which Internode could have never competed. The ACCC don’t have that option with NBNCo, so they will have to go for other options, like enforcing price changes, like they have been with ULL and TW prices for the past decade.

        NBNCo won’t have any incentive to remove the speed tiers and if take-up is lower than forecast, then usage will also be lower. At that point it would take an extremely brave CEO to move to a riskier volume based pricing model.

        If by volunteers you mean per Gigabyte, yes, very brave. Let’s introduce a pricing model that effectively destorys competition.

        And if you don’t mean that, you seem to forget NBNCo are not profit motivated for as long as they remain a GBE.

    • “Agree 100%. Speed tiers are a carryover from gouging telco’s like Telstra.”
      Actually, charging everyone for full speed would be price gouging. Why charge light users for speeds they aren’t going to use? To help make prices cheaper for the heavy users? How is that fair?

      “There is no technical reason to charge less for a slower speed, or more for a faster speed.”
      Yes there is, backhaul provision. Just because it’s fibre doesn’t mean there is infinite bandwidth. In fact quite the opposite, higher bandwidth in the last mile means there is more pressure on the backbone which is already fibre. ISPs have to accommodate peak usage, giving everybody the fastest speed even if they don’t want it needlessly increases the amount of backhaul they have to pay for.

      “Bean counters have decided they can make more money this way. Isn’t this supposed to be a GOVERNMENT initiative to provide HIGH SPEED broadband to all?”
      No, it’s a government initiative to make affordable high speed broadband available to all. They aren’t giving it away, that would be completely irresponsible and unnecessary.

      “No doubt, this will go away eventually, like it has on HFC cable plans where telco’s now offer the higher speeds FOR FREE as an enticement to take a plan with them.”
      Comparing apples and oranges. HFC operators can’t charge more for higher speeds because users can’t be guaranteed higher speeds. So they charge everyone the same and you get what you get depending on line quality, time of day, weather etc. Sound familiar? Oh wait yes it’s the same as ADSL2+, just faster some of the time.

      • > Actually, charging everyone for full speed would be price gouging. Why charge light users for speeds they aren’t going to use? To help make prices cheaper for the heavy users? How is that fair?

        Because it doesn’t actually cost anything for the full speed, and everyone in the country receives a benefit when the end speed goes up.

        > Yes there is, backhaul provision. Just because it’s fibre doesn’t mean there is infinite bandwidth. In fact quite the opposite, higher bandwidth in the last mile means there is more pressure on the backbone which is already fibre. ISPs have to accommodate peak usage, giving everybody the fastest speed even if they don’t want it needlessly increases the amount of backhaul they have to pay for.

        Again you’ve demonstrated a significant lack of understanding. It is quota that limits broadband usage in Australia. Do you seriously think if an ISP has 1000 customers at 12/1Mbps they buy 12,,000Mbps of backhaul and if those same customers connect at 1000/400Mbps, they buy 1,000,000 of backhaul?

        Take the example of someone on a cheap 50GB/month plan, that equates to less than 14 seconds per day at full speed – barely a blip on the network load. Even someone on a 1TB plan has less than 5 minutes a day at full speed. What happens is that in reality is that ISPs buy only a small fraction of the total customer bandwidth and rely on the fact that not everyone is using their connection at the same time. In fact, if you are familiar with buffer bloat you would be well aware that having faster end-points on the network would help alleviate the problem.

        Consider this scenario: It is Microsoft’s patch Tuesday so as people arrive home and turn on their computers windows starts automatically downloading 500GB of updates. Now Microsoft use a content distribution service, so that data can be supplied at 1Gbps. A person connected 12/1Mbps will take about 6 minutes. On a 1000/400Mbps connection it is 4 seconds. The person on the faster connection has finished their download and is no longer consuming bandwidth.

        > No, it’s a government initiative to make affordable high speed broadband available to all. They aren’t giving it away, that would be completely irresponsible and unnecessary.

        Except they aren’t. The fact that NBNCo are predicting 50% of fibre connections will opt for 12/1Mbps means that high speed (100Mbps or faster) means they have failed at that objective. Heck, HFC, FTTN, 4G and approaching half of ADSL2+ connections already offer this.

        The government gives away many things that it could charge for, including education and healthcare. Better access to education and healthcare are promoted unceasingly by the government as outcomes of the NBN. If it could be shown that the NBN reduced costs in these very expensive areas, then it would be easy to justify significant discounting of access to the NBN.

        For example, the government could provide “health services”, “education programs” and “social services” over the NBN and negotiate with RSPs to carry these services for free. I’m not sure that I’m comfortable with the idea of providing the government with a medium propaganda, but alternatively free access to education and healthy living programming is attractive.

        > Comparing apples and oranges. HFC operators can’t charge more for higher speeds because users can’t be guaranteed higher speeds

        Again you are working under false assumptions. Firstly NBNCo is selling AVC PIR (peak not committed) speeds. If you you want 5/5Mbps of CIR, wholesale cost is $334 (+$27 of AVC PIR). Secondly RSPs don’t guarantee end-to-end performance particularly once it leaves their network. Thirdly how many servers do you think can currently sustain 1Gbps?

        If everyone on fibre has access to 1Gbps then service providers will experiment with services that utilise those speeds and international companies are much more likely to explore new technologies here.

        • Because it doesn’t actually cost anything for the full speed, and everyone in the country receives a benefit when the end speed goes up.
          At the last mile it doesn’t, but you are dreaming if you think that the speed at the customer end has no effect on the upstream. The upstream has huge bandwidth issues

          Again you’ve demonstrated a significant lack of understanding. It is quota that limits broadband usage in Australia. Do you seriously think if an ISP has 1000 customers at 12/1Mbps they buy 12,,000Mbps of backhaul and if those same customers connect at 1000/400Mbps, they buy 1,000,000 of backhaul?

          Higher end user speeds
          Will be more expensive for the 12/1 users. (no longer being subsidised)
          Will require more bandwidth across the NBN Network (CVC)
          Will require more international bandwidth.
          Will be cheaper for the 100/40 users (no longer have to subsidise)

          You are advocating significantly higher prices for the already price sensitive people at the bottom of the market.
          Aren’t you doing this to help them get internet? What if your higher prices mean they can’t afford it at all. And advocating lower prices for those that can (arguably) already afford it.

          Thirdly how many servers do you think can currently sustain 1Gbps?

          I would like to introduce you to the concept of Torrents, where 10 thousand people with 1 megabit upload; send data to each other from diverse connections.

          If everyone on fibre has access to 1Gbps then service providers will experiment with services that utilise those speeds and international companies are much more likely to explore new technologies here.

          And significantly higher minimum prices. Which means your plan is 100% worse for more people. (ie going from “Some internet” to “No internet” because it costs too much).

          • > You are advocating significantly higher prices for the already price sensitive people at the bottom of the market.

            No I’m not. There is absolutely no reason for the cheapest plans to change. There may be a short term effect in that data is slightly more expensive for higher quota plans (1TB) but that is only fair as these people are actually putting load on the system. The counter balance to this is that RSPs are likely to have significantly more spare capacity in off-peak times, meaning people who are prepared to schedule their downloads will be able to download more.

            > Will require more bandwidth across the NBN Network (CVC)

            Which means CVC prices will fall dramatically.

            > Will require more international bandwidth.

            Are you sure about that? Most popular content is already hosted here via content distribution networks. An RSP doesn’t need to provide 1Gbps all the way to Europe. In fact faster NBN speeds will encourage more hosting here, because it will deliver a better end user experience.

            If you consider the primary social benefits (education and health) all of that traffic is local.

            > Aren’t you doing this to help them get internet? What if your higher prices mean they can’t afford it at all. And advocating lower prices for those that can (arguably) already afford it.

            NBNCo have said in the corporate plan, that 13% of premises passed by fibre will choose wireless because it is cheaper. The reality is that 4G is cheaper and faster than 3 cheapest speed tiers on the NBN and that is before you consider that many people have data bundled in with their mobile plan.

          • No I’m not. There is absolutely no reason for the cheapest plans to change.

            Actually, in effect you are, but you may not realise it.

            NBNCo needs to achieve an ARPU of about $100. They can’t cross subsidise any other way, unless you got some ideas? Let’s hear them.

            Without NBNCo doing the cross subsidisation we have to rely on RSPs to do it. RSPs, particularly Australian ones, aren’t very good at doing this. They will attempt to maximise profit over all else. This means that they are far more likely to pass the costs directly onto the consumer.

          • > Actually, in effect you are, but you may not realise it.
            > NBNCo needs to achieve an ARPU of about $100. They can’t cross subsidise any other way, unless you got some ideas? Let’s hear them.

            Ouch. I thought they were targeting $30 – $35, but that is immaterial. There is no reason why access cannot make up 10% of the bill and data 90%. That is how my electricity bill is structured.

            > Without NBNCo doing the cross subsidisation we have to rely on RSPs to do it. RSPs, particularly Australian ones, aren’t very good at doing this. They will attempt to maximise profit over all else. This means that they are far more likely to pass the costs directly onto the consumer.

            I don’t see why you think this. The RSP level is very competitive, so costs will not be unfairly passed onto consumers.

          • Ouch. I thought they were targeting $30 – $35, but that is immaterial

            Initially, when AVC is more significant that CVC, there are targeting $35 APRU. This is only because initially there is not enough users on a POI to require many ISPs to significantly dip beyond the “free 150Mbps” they are getting. As demand increases, CVC will become more significant than AVC, especially as business and multi-casting products will be introduced.

            And it’s not immaterial. It is extremely relevant. While NBNCo have a huge debt to pay off they have to do everything in their power to make their 7% target. It is extremely difficult to make adjustments while maintaining the low entry level price.

            I don’t see why you think this. The RSP level is very competitive, so costs will not be unfairly passed onto consumers.

            I don’t know where you’ve been but the only RSP in Australia currently that gives a toss about making sure there is a cheap entry level product for low income consumers is Telstra, and that is because they are obligated too. Every other RSP will only remain where they can sustain profit. I’m sorry, it may be cynical, but if you raise the entry level wholesale price, the entry level retail price will raise as well. And by removing speed tiers NBNCo are required to do just that.

          • Ouch. I thought they were targeting $30 – $35, but that is immaterial. There is no reason why access cannot make up 10% of the bill and data 90%. That is how my electricity bill is structured.
            For that to work the CVC charging would need to changed. Simon Hackett has previously floated the idea of “Burstable Billing” http://en.wikipedia.org/wiki/Burstable_billing

            Currently CVC has to be predicted as it has to ordered in advance and can’t be adjusted mid month. The option for burstable billing may help and facilitate something that comes closer to “user pays” model.

            At the very least it is an interesting conversation to be had, I wonder if the both types of billing could coexist?

          • Link http://blog.internode.on.net/2011/07/21/nbn-retail-pricing-pressure-points/

            2) Provide technical burst capacity in each CVC and use 95th percentile charging for actual utilisation

            This alternative approach is to charge for CVC usage based on something called ‘95th percentile‘ charging – in effect, to charge on the average real world usage in the CVC by customers, and to provide as much CVC size as each RSP technically requires above that average for adequate performance (which winds up being… 200 megabits minimum and then at least 100 megabits of burst above the observed usage peak assessed).

            There are other approaches too, but by far the simplest is the provision of the first 200 megabits per CVC free of charge (on an ongoing basis) by NBNCo.

            This eliminates the punitive overhead (per CVC and per POI) imposed on RSPs by the current model.

            This then delivers a genuine level playing field in terms of effective per-user wholesale access cost for RSPs by removing the punitive long-term cost overhead that will otherwise exist.

            And it does this without making things more complicated for anyone concerned – its the same cost structure, but with a specific change to make it financially rational during the initial decade of the NBN build.

            It would remove a windfall cost component that has a plethora of negative impacts on RSP participation and that will place substantial uppward pressure on retail prices from providers.

            … and its got to be fixed, if we are to expect rational retail service provider participation in new development areas (greenfields) as well as in the major areas of existing housing (brownfields) that the NBN will spend the coming ten years overbuilding with fibre.

          • Which is why NBN are now offering the first 150Mbps of CVC free, see here.

            However, getting RSPs to agree to a 95th percentile model is going to be difficult. Simon may have suggested it, but most ISPs like nice, predictable costs. Exetel for example is known for kicking high users off their network.

        • “Take the example of someone on a cheap 50GB/month plan, that equates to less than 14 seconds per day at full speed – barely a blip on the network load. Even someone on a 1TB plan has less than 5 minutes a day at full speed.”
          You fail to understand usage patterns. Sure, people have have limited ‘time at maximum speed’, but they all use this time at the same time of day. Hence peak load, you conveniently skipped over that point in your reply to me.

          “What happens is that in reality is that ISPs buy only a small fraction of the total customer bandwidth”
          Wow really? What a shocker! And when total customer bandwidth goes up, you think perhaps that fraction becomes a larger number? One tenth of a big number is still bigger than one tenth of a small number you know.

          “The fact that NBNCo are predicting 50% of fibre connections will opt for 12/1Mbps means that high speed (100Mbps or faster) means they have failed at that objective.”
          Making it available doesn’t mean people have to get it. Those who want it can get it, objective achieved.

          “Heck, HFC,”
          not reliably
          “FTTN,”
          there is no FttN in Aus, so no it doesn’t
          “4G”
          not reliably
          “and approaching half of ADSL2+ connections already offer this.”
          lol what fantasy world are you living in?

          “The government gives away many things that it could charge for, including education and healthcare.”
          Yes it does, because there is a case for giving them away. There is no case for giving away broadband because the NBN is doing it for ‘free’ (from a gov’t perspective). The parallel with Aus Post is remarkable, they charge more to deliver larger packages, do you think the government should step in and spend taxpayers’ money to make prices equal for all deliveries? I mean, that would truly mean the benefits of the postal system are available to everyone right?

          “Firstly NBNCo is selling AVC PIR (peak not committed) speeds. If you you want 5/5Mbps of CIR, wholesale cost is $334 (+$27 of AVC PIR). Secondly RSPs don’t guarantee end-to-end performance particularly once it leaves their network.”
          Apples and oranges again. There’s a big difference between charging for more bandwidth that they might not be able to use occasionally due to contention, and charging for more that the simply can’t use due to last mile speed limits.

          “Thirdly how many servers do you think can currently sustain 1Gbps?”
          Very few. Did you know it’s possible to connect to more than one server at a time? We aren’t all on one big token ring, we have these things called routers and switches everywhere!

          “If everyone on fibre has access to 1Gbps”
          They will have access to 1Gbps. A lot more people will not connect therefore not have access to 1Gbps or any other speed if you force minimum prices up by not offering slower access.

          • “Take the example of someone on a cheap 50GB/month plan, that equates to less than 14 seconds per day at full speed – barely a blip on the network load. Even someone on a 1TB plan has less than 5 minutes a day at full speed.”
            > You fail to understand usage patterns. Sure, people have have limited ‘time at maximum speed’, but they all use this time at the same time of day. Hence peak load, you conveniently skipped over that point in your reply to me.

            So you are suggesting that everyone will access the internet for that same 5 minutes? Highly unlikely.

            > Making it available doesn’t mean people have to get it. Those who want it can get it, objective achieved.

            The NBN is not about those who want it. It is about providing a fast network for all Australians. If it was about those who want it, then

            “Heck, HFC,”
            > not reliably
            “FTTN,”
            > there is no FttN in Aus, so no it doesn’t
            “4G”
            > not reliably
            “and approaching half of ADSL2+ connections already offer this.”
            > lol what fantasy world are you living in?

            The land of hard facts … The average ADSL2+ speed in Australia is over 10Mbps. That doesn’t mean everyone has over 10Mbps, but 50% do. What I’ve pointed out is that other technology is easily able to exceed 12/1Mbps.

            “The government gives away many things that it could charge for, including education and healthcare.”
            > Yes it does, because there is a case for giving them away. There is no case for giving away broadband because the NBN is doing it for ‘free’ (from a gov’t perspective).

            If you can generate a social benefit (e.g. reduction in health care costs or better educated population) then there is a justification for giving it away for free.

            “Thirdly how many servers do you think can currently sustain 1Gbps?”
            > Very few. Did you know it’s possible to connect to more than one server at a time? We aren’t all on one big token ring, we have these things called routers and switches everywhere!

            You conveniently miss the point that a 1TB plan provides only 5 minutes a day at full speed. What i was pointing out is that the actual need for 1Gbps at the moment is small, but by having it available it enables people to build innovative services.

            Sure someone could do it, but with current quotas they would be quickly shaped. Someone who wants to download large amounts of traffic should pay for the load they are placing on the network. This will fund upgrading to GPON10 and also lead to a reduction in CVC prices.

          • “So you are suggesting that everyone will access the internet for that same 5 minutes? Highly unlikely.”
            No, but most people will use their 5 minutes around the same time each day, concentrating their usage into a peak time. When there is peak hour traffic on the road is that because every single person tries to commute at the exact same time? No.

            “The land of hard facts … The average ADSL2+ speed in Australia is over 10Mbps. That doesn’t mean everyone has over 10Mbps, but 50% do. What I’ve pointed out is that other technology is easily able to exceed 12/1Mbps.”
            You shouldn’t make up numbers out of thin air, it makes you look ignorant. http://www.news.com.au/technology/state-of-the-internet-australia-web-speeds-ranking-dwindles-to-40th-place-globally/story-e6frfro0-1226560992748

            “You conveniently miss the point that a 1TB plan provides only 5 minutes a day at full speed. What i was pointing out is that the actual need for 1Gbps at the moment is small, but by having it available it enables people to build innovative services.”
            So you’re saying people can switch to a higher quota plan to take advantage of future innovative services but they can’t switch to a higher speed plan?

          • > No, but most people will use their 5 minutes around the same time each day, concentrating their usage into a peak time.

            Most people will use their 5 minutes spread out over the evening. It isn’t that everybody uses the internet at 7:30pm. The larger your customer base the more spread out it is.

            >> “The land of hard facts … The average ADSL2+ speed in Australia is over 10Mbps. That doesn’t mean everyone has over 10Mbps, but 50% do. What I’ve pointed out is that other technology is easily able to exceed 12/1Mbps.”
            > You shouldn’t make up numbers out of thin air, it makes you look ignorant. http://www.news.com.au/technology/state-of-the-internet-australia-web-speeds-ranking-dwindles-to-40th-place-globally/story-e6frfro0-1226560992748

            You shouldn’t quote meaningless figures. What you have quoted is the speed of all internet services in Australia, whereas I was talking specifically about ADSL2+ connections. Internode && iiNet published information based on DSLAM speeds. The quickest link I could find is this one: http://www.ingenium.net.au/clubsite/files/2/File/KML/iiHeat.asp, but it should also be on the Internode Blog.

            >> “You conveniently miss the point that a 1TB plan provides only 5 minutes a day at full speed. What i was pointing out is that the actual need for 1Gbps at the moment is small, but by having it available it enables people to build innovative services.”
            > So you’re saying people can switch to a higher quota plan to take advantage of future innovative services but they can’t switch to a higher speed plan?

            What I’m saying is that someone can try an innovative service if they have 1Gbps until their quota runs out. If the service provides something of value, then they can increase their quota. However NBNCo don’t provide the option for “burst” speeds to trial a service. See the difference?

  20. > Lower access price also means lower ARPU. Lower ARPU results in long payback time frame.

    Wrong. Lower AVC means more people connected. Higher speeds means more income from CVC. The NBNCo Corporate Plan talks about reducing the proportion that AVC contributes to revenue and increasing the proportion that CVC contributes. All that I’m proposing is bringing forward that to encourage take-up and deliver a faster network for consumers.

    > We’re already at 7% ROI and 20 years. This is very risky. As I have said before, unless the government is willing to give a subsidy injection we don’t have many options.

    All that is required is to modify the revenue base to include more revenue from CVC and less from AVC. I agree this will be more risky in the short-term, but few would debate that data usage is not going to grow and that higher speeds increase the rate at which data usage grows. In fact many would argue based on history that data usage will continue to grow faster than we anticipate.

    • Wrong. Lower AVC means more people connected. Higher speeds means more income from CVC. The NBNCo Corporate Plan talks about reducing the proportion that AVC contributes to revenue and increasing the proportion that CVC contributes. All that I’m proposing is bringing forward that to encourage take-up and deliver a faster network for consumers.

      Learn to read. Did you notice how I was not talking about AVC, but in fact access price, not AVC. If you talk about AVC in isolation you’re right. You seem to misunderstand that CVC is the mechnism for artifical scarcity, not AVC. They could remove AVC completely and all that would cause providers to do is artifically shape their connections. The difference between 12Mbps and 100Mbps in AVC cost is negligible compared to the CVC costs providers will have to pay.

      All that is required is to modify the revenue base to include more revenue from CVC and less from AVC. I agree this will be more risky in the short-term, but few would debate that data usage is not going to grow and that higher speeds increase the rate at which data usage grows. In fact many would argue based on history that data usage will continue to grow faster than we anticipate.

      The prices are already geared towards more revenue from CVC, so this adjustment is not required.

  21. > The difference between 12Mbps and 100Mbps in AVC cost is negligible compared to the CVC costs providers will have to pay.

    Totally wrong, because once you have a couple of customers at the faster speed, everyone else can come along for the ride if they are low quota users. Do you not comprehend the implications of a 1Gbps connection with a 5TB quota means less than 23 minutes of full speed a day.

    If you look at RSP plans, what you find is a fixed increase for each speed tier and a fixed price increase for each quota increase. iiNet conveniently ask people to choose quota and speed.

    > The prices are already geared towards more revenue from CVC, so this adjustment is not required.

    When considering the cheapest plan, NBNCo have set the pricing structure so that CVC contributes < $1 and AVC $24. Contrast this with NBNCo's stated goal of CVC contributing 50% of revenue. I suggest you actually take the time to read the NBNCo Corporate Plan and then come back with an informed opinion. Like most people in this debate you don't actually understand what NBNCo are planning.

    • That initial problem only applies to the initial users. Once you get past the initial users you’ll add CVC at a pretty much consistent ratio to the desired PIR, regardless of their quota. In other words what determines CVC is more strongly related to PIR than quota.

  22. > That initial problem only applies to the initial users. Once you get past the initial users you’ll add CVC at a pretty much consistent ratio to the desired PIR, regardless of their quota. In other words what determines CVC is more strongly related to PIR than quota.

    What maths did you use to work that out? The more customers you have the less CVC you need per customer, because the peaks are smoother. This is the exact reason that NBNCo are currently providing 150Mbps of CVC free.

    I guess once you have a large enough customer base it is possible that adding a customer would require purchasing additional CVC, but it would be a small fraction of the actual PIR.

    • Umm. What? You just corrected me with my own explaination reworded.

      That’s exactly my point. Once you get past the “initial requirement” of CVC, i.e. the inital users, you add CVC at a consistent rate in order to maintain your desired contention ratio.

      • > That’s exactly my point. Once you get past the “initial requirement” of CVC, i.e. the inital users, you add CVC at a consistent rate in order to maintain your desired contention ratio.

        Contention ratios are based on the amount of data being transferred through each link in the network. It has little relationship with end-point speeds for a large enough customer base where each link is close to the contention point during peak.

        The CVC added is not consistent. The first customer requires that CVC = end point speed. The second customer also requires that CVC = end point speed, because the risk of collision is too high. The 3rd customer requires CVC = 50%, The 4th customer requires 25%, the 5th customer requires 15%, etc.

        When your customers are utilising their connection for only a small part of the peak period, then the amount of additional CVC required becomes very small for a new customer. Moving a low quota user (50GB) from 12/1Mbps to 1Gbps will have no effect.

        • Contention ratios are based on the amount of data being transferred through each link in the network. It has little relationship with end-point speeds for a large enough customer base where each link is close to the contention point during peak.

          The CVC added is not consistent. The first customer requires that CVC = end point speed. The second customer also requires that CVC = end point speed, because the risk of collision is too high. The 3rd customer requires CVC = 50%, The 4th customer requires 25%, the 5th customer requires 15%, etc.

          When your customers are utilising their connection for only a small part of the peak period, then the amount of additional CVC required becomes very small for a new customer. Moving a low quota user (50GB) from 12/1Mbps to 1Gbps will have no effect.

          That is what I’m talking about. Seriously. I don’t understand how you can repeat back to me the same thing reworded so many times.

          The initial requirement the the first dozen or so users, once you get past that the relationship to users/CVC becomes practically linear.

          • > The initial requirement the the first dozen or so users, once you get past that the relationship to users/CVC becomes practically linear.

            You are working at the wrong level of scale. Try thinking about it in terms of 10,000 users. That is the minimum numbers for the NBN for the numbers to add up.

            Firstly it is not linear. It is a curve that tends towards a flat line.

            Secondly relationship between users/CVC is linked with quota. If you add someone with a 5TB quota that is going to have a different impact from someone with a 50GB quota.

          • That is what I’m talking about. Seriously. I don’t understand how you can repeat back to me the same thing reworded so many times.

            The initial requirement the the first dozen or so users, once you get past that the relationship to users/CVC becomes practically linear.

  23. The tiers are there to allow price discrimination. An economist should understand that.

  24. If 85% of people only want the top tiers, why can’t they offer only the higher tiers and make up the difference in volume?

Comments are closed.