AT&T verifies data throttling for top 5% of users | Hacker News Reader

AT&T verifies data throttling for top 5% of users

9to5mac.com

41 pointsFlemlord49 comments

This doesn't really make sense to me because it seems that over time it will reduce the threshold for the entire network. Here is what I'm thinking: suppose that for the first month the top 5% uses 50gb of data per month. Well once the throttling kicks in they will not be able to reach the 50gb per month that they usually hit, and instead they only hit 40gb.

Next month the top 5% of usage rates is 40gb per month, and so the threshold kicks in sooner. If this repeated over the course of months then the threshold would get lower and lower, until it began to infringe on normal usage. I'm willing to assume that AT&T is doing this intelligently, and that the top 5% is a static measurement, but the way it is described makes it sound like a sliding scale that will lower the threshold more and more over time.

And that doesn't even touch the issue of always throttling connections for 5% of their customers...

They're not capping the top 5% each month. They're throttling it. That means that the top 5% could always be a certain range. So if the top 5% is over 50GB/month, then once you hit 50GB you'll be slowed down, but not stopped. You can keep going, and this is why the top 5% can stay the same from month to month.

It's possible that they can hit 50GB next month just as before, unless their behavior changes.

You're right that this plan could bring the top 5% down to 40GB over time, but that's only if the users behavior on aggregate changes. Once you're in the 95th percentile of users, I'm not sure how much extra usage is needed to get to the 99th, or even how much changing your behavior will significantly lower the averages.

When your bandwidth is limited it becomes difficult to get many more gigabytes as that is the whole point of a throttle. So when that throttle is applied to the power users then they physically can't download as many gigabytes as they used to. So my theory is that over time the difference between the 99th percentile and the 95th is going to decrease as the power users who used to raise the bar higher are throttled down to a more normal gigabyte per month usage.

It might lower the ceiling, but I don't see how it could bring the threshold lower. Since the throttling is reset every month and doesn't kick in until you hit the threshold, it will only affect people who go over the threshold to begin with. If you were consistently below the threshold before, I don't see how this would cause the threshold to pass you by unless the top 5 percent just get pissed off and leave altogether.

  unless the top 5 percent just get pissed off and leave altogether.

But this is exactly what will most likely happen. Assuming that at least a larger percentage of this top 5% are 'power downloaders', I can pretty much assure you that those will not take this, and leave (if they can, that is).

My problem with this is that they've advertised "UNLIMITED" data and convinced a lot of people to join their network because of it, and now they are saying that's not actually true.

I know they have footnotes in the contract, but isn't it some kind of false advertising? It's not "unlimited", it's more like "a lot", and they get to decide what a lot means.

If you're not going to really give people unlimited data, and you're going to get upset about it when people actually try using it as if it was unlimited, then how about you stop advertising it as such?

It's illegal to offer unlimited in Australia if it's not by definition, unlimited.

http://www.theage.com.au/business/accc-seeks-unlimited-optus...

http://www.lifehacker.com.au/2011/01/dodo-busted-by-accc-for...

Optus is the second most popular ISP in Australia. Telstra is most popular and even they got slapped for having 'Unlimited' in the past (Now called Liberty plans).

Same with a lot of residential ISPs too. They all advertise internet in terms of download and upload speeds, yet if you used the connection at that speed you'd hit the fine-print data cap in under a day.

To play devil's advocate: does cutting back on the transfer rate constitute a data limit? Users that hit that 5% threshold aren't blocked from sending and receiving data, they just don't get the same transfer speeds.

Granted, it sucks to see this happen, but if transfer rates aren't also specified in the plan/advertising it's hard to argue that this represents going back on an "unlimited data" promise.

True that the transfer rates aren't specifically spelled out, but if I'm advertised the nation's largest 3G network, or something to that tune, then my expectation is unlimited data on that network.

> To play devil's advocate: does cutting back on the transfer rate constitute a data limit?

I would say yes: if you have, say, a 1MB/s line, which you keep constantly saturated, you can transfer (1 month * 1MB/s) = ~2.5TB in that month. Whether you have a data cap or throttling, it translates to such a line getting being able to transfer less than 2.5TB of data per month.

So you're saying absolutely no plan at all can be considered unlimited because with current technology there is some maximum download speed? I guess that's right (and probably not just about current technology: I would think there are theoretical limits to download speeds as well).

No. "Unlimited", to me, means you get the advertised speed every second of the month, no matter the data volume you transfer. As soon as they start slowing you down, it is no longer unlimited.

No, but I think you can put forward that if the transfer speed is reduced in direct response to your usage, that the "unlimited" portion of the advertising clearly no-longer applies.

Taken to the extreme, one could claim "unlimited" service permitting one http connection a day -- open a second connection, and zero bytes go through until midnight.

You'd say "that's absurd; it's not high speed internet. You've effectively limited my speed." They advertise _unlimited_ and _high speed_. If they throttle, then one may have a case depending on the reasonable expectations one might have about how fast (total throughput) a service labeled as "high speed" should be.

(IANAL - this is not legal advice)

On the surface it seems like a fair and balanced approach to the problem. I wish there was an actual number set, rather than the top 5% of each month's users which means that they're promising to always throttle speeds of 5% of their customers. Perhaps they're hoping this will cause a race among the customers who use the most data to reduce their data usage every month?

I also find great humor in the throw-away line concerning the necessity of the acquisition of T-Mobile at the end.

I also find great humor in the throw-away line concerning the necessity of the acquisition of T-Mobile at the end.

Apparently AT&T has hired Cato the Elder to do their PR. This is bad news for Carthage.

We shall lay low their towers and sell their engineers to the Shanzhai.

Preface: What I am about to say is a complete and utter assumption.

One would assume that ATT did some analysis on bandwidth usage patterns amongst heavy users. They probably came to the conclusion that the top 5% of bandwidth hogs account for X% of their entire bandwidth throughput.

ATT's approach doesnt really solve their problem very much though. If someone has extremely sporadic bandwidth usage but uses gigs at a time, limiting these users won't do much for solving whatever problem they are trying to solve. Maybe their data modeling/forecasting is so accurate that they are able to segment those 5% into various segments and limit that way.

That's not amusing at all. And in fact, what is more likely: competition on GSM will bring better service and more data cheaper to users, or a monopolistic single service on GSM will bring better service and more data cheaper to users.

If I was at the FTC or in Congress, or at Sprint, I would definitely be waving this AT&T policy around, and use it to stop the merger.

What AT&T needs is more competition, not less.

A couple of changes would make this into an actually reasonable policy that I wouldn't mind other ISPs adopting: 1. Define a usage threshold instead of always throttling 1/20th of your customers, and 2. Allow customers to pay a reasonable fee to boost their speeds back up after they hit the throttling.

Isn't #2 pretty close to their tiered data plans?

One new measure is a step that may reduce the data throughput speed experienced by a very small minority of smartphone customers who are on unlimited plans – those whose extraordinary level of data usage puts them in the top 5 percent of our heaviest data users in a billing period.

If 1 out of 20 people qualifies, then it's by no means "extraordinary". "Extraordinary" should be something like 1 in 10,000 or 1 in 100,000. If 3 people on your block fit the bill, then that's just called "ordinary".

Stop pretending that you're not a dumb pipe and bill by usage. The sooner information is treated like the basic utility it is, like fuel, water or electricity, the better off we'll all be. It's painful to watch every last data provider dance around the subject like this. You're not fooling anybody. All right, you're fooling a lot of people, but not me, and not anybody else for very long.

People don't want to pay by usage, though. They want unlimited bandwidth and transfer for $30/month, and they aren't going to be happy with settling for $15/GB.

Customers always want better products for free. The problem is that ISPs are responding to the desire for unlimited bandwidth by providing "unlimited" with tons of caveats. Everybody would be better off if they would just cut to the chase and charge by usage. Advertising "unlimited" should qualify as false advertising if it's not truly unlimited, which would push everybody towards more sane behavior.

Everybody would be better off except for the ISP's who would now be exposing the ridiculously high dollar per gigabyte rate they are charging.

I agree that they should not (and should not be able to) call a limited plan "unlimited". However, I don't think that going to a true pay-for-usage plan is realistic. Most people don't use much bandwidth. Those people are effectively subsidizing the high-usage clients.

Let's assume that the average data usage is 1GB (the actual amount doesn't matter). Right now, everyone pays $30 whether they use 1kB or 1TB. Imagine if instead we go to a pay per MB plan. A MB currently costs $0.03 (on average, and assuming 1000MB to the GB). We know that the top 5% are using 12x as much data as the average user, so the top users are around 12GB, while the average user is around 420MB. So if we keep the usage price the same, the average non-top-5% user will be paying only $12.60. The top users are paying $360.00. Of course, the top users aren't going to continue this kind of payment. They'll leave for another provider or else drastically cut their usage. So now AT&T is serving less data (yay), but they're also bringing in about 42% as much revenue.

And bear in mind that this same calculation could be done for the next 5% and the next. The person using 1MB today paying for the 999MB they don't use so that someone else can use it. After the new plan, their fee goes down to $0.03, and someone else's goes up by $29.97. A lot of people would simply stop using any significant bandwidth and force AT&T to raise the price per MB drastically.

I think tiered plans make more sense, where instead of a true pay-for-usage, you basically have pay-for-max-usage. I think this might be where we end up. And the reality is that what we'll probably see will be $30 for "average" or slightly-above-average usage (1GB in my example), and other tiers for "exceptional" usage. Maybe $45 for 2GB and $60 for 4GB. AT&T isn't going to take a big pay cut on average users, because exceptional users by and large will not make up the difference. They'll just stop being exceptional users.

I think you numbers are way off. The average user is using less than 100MB a month. If your assumption is correct that the top users are using 12x as much data (don't know where you got this info) - it seems like they'll be throttling anyone that uses more than 1.2GB of data a month...

It doesn't matter if the number for average usage is way off. The math stays the same. Light users are bearing the cost for heavy users. In fact light users are by and large carrying the cost for even average users.

The 12x number came from the article.

The thing that I think is wrong in your calculation is your thoughts that data transfer forms the highest cost for these providers. I sincerely doubt that.

Anyway, a total fee of $0.03 is ridiculous, as it doesn't include fixed costs for the connection.

Directly converting current fees into dollars per megabyte isn't going to work. A person who does nothing but check e-mail will be paying pennies per month, which is not going to pay for the ISP's costs. Instead, I would expect there to be a monthly fee just to have service at all, and then usage charges on top of that. I'd expect that roughly 50% of the current typical monthly charge would be to have service, and then the rest would be captured by usage charges.

There's also nothing preventing ISPs from providing volume discounts. Indeed, they should, as their own cost per byte per customer should go down as the customer uses more.

I'm not sure why you see it as a good thing that average users subsidize top users. Surely top users should pay in proportion to their additional use? After all, I don't pay the same amount for my electricity as somebody who cools his house to 60 degrees and leaves his lights on 24/7. Structure usage so that the average cost per month stays the same (and thus total revenue stays the same) but heavy users get charged more. It's more fair and is the model used in virtually every other service and industry in existence.

> Directly converting current fees into dollars per megabyte isn't going to work.

That was kind of my point. A straight pay-for-usage plan is unlikely to work.

> I'm not sure why you see it as a good thing that average users subsidize top users. Surely top users should pay in proportion to their additional use?

What exactly do you think the monthly fee is? Do you really think it costs AT&T $15 (your suggested fee) to provide someone with the ability to connect to their network? This is still low-usage customers subsidizing the high-usage customers.

> Structure usage so that the average cost per month stays the same (and thus total revenue stays the same) but heavy users get charged more.

That doesn't work, for the reasons I pointed out. Some heavy users will pay a little more, but some will not. They'll simply cut back or leave. Very few would be willing to pay the massive fees that they would incur. So the net result is a massive drop in revenue. You can't just reprice everything and expect usage patterns to remain the same.

> It's more fair and is the model used in virtually every other service and industry in existence.

Yet everyone seems to argue against this model when we're talking about Comcast.

> Yet everyone seems to argue against this model when we're talking about Comcast.

That's because people have become used to receiving enormous subsidies from the clueless unwashed masses, and that kind of self-interest almost always takes priority over recognizing that paying $50/month for downloading 500GB of porn while your neighbor pays $50/month for downloading e-mail just isn't fair.

> That doesn't work, for the reasons I pointed out. Some heavy users will pay a little more, but some will not. They'll simply cut back or leave. Very few would be willing to pay the massive fees that they would incur. So the net result is a massive drop in revenue. You can't just reprice everything and expect usage patterns to remain the same.

I simply don't believe you when you say it doesn't work. Other industries are able to charge for use without a massive collapse in revenues as you state. I don't see why a massive drop in revenue needs to happen. I don't expect usage patterns to stay the same at all, I merely expect the new usage patterns, when applied to sanely structured new prices, to result in similar revenue.

> That's because people have become used to receiving enormous subsidies from the clueless unwashed masses, and that kind of self-interest almost always takes priority over recognizing that paying $50/month for downloading 500GB of porn while your neighbor pays $50/month for downloading e-mail just isn't fair.

Can I assume that you are in favor of tightly-capped broadband, then?

I don't think that scenario is unfair at all, either. If you and I rent town-homes side-by-side, is it unfair that I live in mine year-round and have guests over regularly, while you are rarely home? Of course not, because you're not being charged for usage. You're being charged for capacity. Whether you use that capacity or not is your business. What about a gym membership? Is it unfair if you pay the same as me but only use the gym once/week while I'm there every day?

> I simply don't believe you when you say it doesn't work. Other industries are able to charge for use without a massive collapse in revenues as you state. I don't see why a massive drop in revenue needs to happen. I don't expect usage patterns to stay the same at all, I merely expect the new usage patterns, when applied to sanely structured new prices, to result in similar revenue.

Most industries don't charge for bandwidth. You don't pay for water service based on how many gallons per minute you've got on the line. You don't pay for electric based on how many amps are flowing to your house. Innovation and progress in the networking arena is driven by bandwidth, though. No one compares America's price per GB to South Korea's price per GB. They compare bandwidth. The fact that worldwide the industry works this way indicates that it's not just some arbitrary American telco thing. Increases in bandwidth provide an increase in capability and quality of service (arguably an increase in quality of life). It makes sense to incentivize this and have companies compete on the basis of speed.

As for why a massive drop in revenue has to happen, ask yourself if you (a hypothetical high-bandwidth user) would continue your service if your bill went up 10x. My guess is that you would not happily pay the increase. Instead you would drop the service entirely or at least drastically reduce your usage. In that case, the extra bandwidth is no longer being utilized, so you're not paying for it and ATT isn't getting the revenue. So revenue drops unless they crank up the price/GB until everyone's paying the same $30 they were paying before. Maybe that's "fair", but it's not especially useful, because now everyone's paying the same price but basically capped at the previous average amount.

Why would you assume I'm in favor of tightly-capped broadband? I would have thought it was extremely clear that I'm in favor of paying per use, which would then eliminate the need for any caps. I can't comprehend why ISPs like Comcast are instituting hard caps rather than simply charging for usage above the limit. If I use more than 250GB/month, Comcast will fire me as a customer. Why don't they just send me a bill for the excess instead? I get to keep my service, they get more money, everybody wins.

I don't see how your hypothetical demonstrates anything. You simply assume that my costs would go up by a factor of ten, when I see no reason why it would be anything near that kind of increase.

What you're talking about isn't pay-for-usage. It's not pay-per usage when Comcast charges a flat fee up to 250GB and then charges extra beyond that. That's a soft cap or a tiered plan. I agree that Comcast seems stupid for cutting people off at 250GB, but I would not call it pay-for-usage to extend beyond that.

I assume you would see a 10x jump in cost because AT&T says that the top users are consuming 12x as much as the average. If users are paying their "fair share", then people using 12x as much bandwidth would pay about 12x as much. Otherwise it's not "fair" and it's not "pay-per-usage". (There might be some "volume" discount, but the overall cost would still be drastically higher.)

I'm not strictly opposed to a tiered plan, but that's not pay-per-usage. It's no more "fair" than the current system. It just means that heavy users have the option to pay extra rather than getting capped. Everyone else would still pay about the same amount, and the guy using 1MB/month is still subsidizing the guy using 250GB/month. Assuming those top 5% aren't going to pay drastically more, everyone else would have to pay about the same on average to keep revenues constant.

Perhaps you have a different idea in mind for how the plan should work, and I'm not understanding your idea. If so, please clarify how you think AT&T should charge, and how the new plan would maintain revenue.

The idea itself is simple: charge a fixed amount for the connection itself, to cover things like physical line maintenance, billing, customer service, etc. Charge a variable amount for use of the connection. That variable amount does not have to be a fixed dollars-per-byte rate.

I'd imagine that the casual e-mail users would pay on the order of $20-30/month (Comcast already offers such a plan in my area, making it cheap by limiting the speed rather than the transfer quantity), average users would pay about what they pay now ($50-60), and heavy users might pay $100-200.

Sorry, but can you clarify how much you expect the fixed connection to cost? It sounds fine to charge a fixed connection fee, but I kind of doubt that the revenue works out the same unless the fixed fee is nearly the cost of the current service. You say that the transfer cost doesn't have to be a fixed dollar/byte, but if it's not, what are you envisioning? If you charge $10 for the first GB and $10 for the next ten GB, you're just on a tiered plan.

You say that light users would pay $20-30, average users would pay $50-60, and heavy users would pay $100-200. I can't see a lot of people willing to pay $200 for home broadband service (I also can't see the FCC allowing it). I have a hard time seeing most of them paying $100 for just Internet access, honestly, and even at that rate it'd be hard to argue that light users at $20-$30 aren't still subsidizing the heavy users. 4x the cost for 12x the bandwidth? Unless we're assuming that the fixed connection fee is $20?

Truthfully, I don't think a lot of people will want the fluctuating bill from month-to-month, either, even low-usage customers. It's nice to know that you owe Comcast $X every month so that you can budget.

Comcast's speed-limited plans are entirely different. Telcos are competing based on bandwidth, not transfer amount. Comcast is offering several plans in that space, none of which are actually competing based on usage.

If it was $15/GB, I would be happy. AT&T charges $15/200MB.

As this is in effect a change of the Terms of Service, I wonder if they will let you get out of your 2 year contract if you were on the unlimited plan WITHOUT a early termination fee.

I know that any change to the ToS that involves pricing/coverage is eligible for a termination of contract with no fees, but would this be covered too?

Thoughts?

You might have to use a lawyer to get it done, but probably yes. But that is true for most of the stuff involving termination fees. The same is true for if you bring in your own cellphone to a plan. They make a lot of money betting you will decided it is not worth fighting for a clause in the contract in your favor and will write you off versus paying a cooperate lawyer their minimum fee.

It seems like they're trying to get rid of the unlimited plans altogether, so I'd bet they'd let you out of your contract if you were one of the people who would be throttled. The people who use that much data aren't nearly as profitable, and they'd probably love to see most of that group switch networks or accept data limits.

The problem for a potential competitor to AT&T that could offer a 'REALLY unlimited' contract is that the people who would switch would be all the really high bandwidth consumers. All the 'juicy' decently paying, (in reality) low bandwidth people don't feel oppressed by AT&T. So the competitor would just be setting themselves up for providing a better service, for the worst ($/Mb) customers.

Let's hope that everyone or at least a majority start using up more and more bandwidth so that the threshold for the top 5% is a very high amount ~30gb. At that point, we can bring the future of high usage to the present so that AT&T and these other telco's can create policies that are future-proof rather than vague, temporary measures. We should make it clear that these high-bandwidth activities are here and the telco's are the bottleneck blocking these new services from going mainstream.

I had every intention of getting an iPhone 5 when it launched and renewing my contract with AT&T at the same time. Now I'm going to run the contract out and then go elsewhere.

IF you want to look at it this way, if they average users speed was 3mb/s , then over the course of 1 month they could download 7593.75GB, which isnt truely unlimited. And if they decide to leave AT&T, who would they go to? T-mobile throttles, Verizon does not offer unlimited, and Sprint has some deffinatley has some issues with users being able to connect at all.

they do not tell us how many GB is too many

Why is this comment not at the top. They need to define what "unlimited" really means.

Reply on news.ycombinator.com