Pocock: Android betrays tethering data

You're taking that value way too seriously. I don't care wether it's the national average, or what the actual value is. I just picked a value for the sake of the illustration. What interests me is the concept of a "reasonable price per Gb for mobile consumers". I only reasserted the value because you were questioning it and saying that it must be higher.

But since you ask: I know because that's what I currently pay from one of the major carriers, and because I very recently had a look at most of the carriers' offers in my country.

> Does it still fit “under 1€/Gb” cap?

You don't need to look hard to see that paying 200% more for using 10% more isn't a linear €/Gb price. That "reasonable if within quota, shocking if over quota" price is what I'm complaining about.

> If not then you've just discovered one way companies use to bridge the gap between what they could profitably offer and what customers demand.

Firstly I haven't discovered anything, I just go biten by something that many people (myself included) have long complained about. Google for "bill shock" for countless annecdotes.

Secondly, the argument that a different carrier behaviour wouldn't be financially viable is a strawman. The price of overage is so high that users do their best to avoid it. It's an exceptional occurence and can't contribute much to the carrier's bottom line. Or if you view it from the "economical incentive for users to not use too much data" angle, then a linear per-GB rate would be a much more efficient incentive.

> As long as we have enough devices on our networks to saturate them this will not be possible.

I fail to see the logic in that. Since we'll always be able to saturate the network (if only because the network eventually gets sized by demand), we need ways to push back against overusage. You may not like the price-per-gb model, but it's the most direct incentive you can think of. It's not just possible, it's obvious. All other schemes currently in use (expect the "actually-unlimited" one) are more complicated.

> If you know how much data will be transferred [...] you can plan for the bandwidth requirements

Yes and no. Sizing a carrier or ISP's network hardware is not like starting an AWS instance. Each base station costs an awful amount of money and time to setup. You can plan ahead (months, not days or hours) but you'll also need to reach a compromise between costly overprovisioning and costly annoyed customers.

> So you are proposing a “solution” which will need more hardware to implement

Absolutely not, no hardware change. Metering is already being done. All you need to change is the routine that looks at the data usage and creates the invoice.

And from a network usage point of view, you're likely to see decreased usage. Price your service accordingly so that you still get the same overall amount.

> and will raise the perceived price (because now both people who use less traffic than “prepaid amount” and people who are using more than “prepaid amount” pay more, often much more, than what they say in ads)

It's hard to gauge what would be perceived, but people who use way less than their prepaid amount would pay less (the cutoff point depends on the new per-gb price), and people who use even just a little bit over the prepaid amount would pay way less (because current overage prices are so high).

From a psychological point of view, most people by now have grown highly suspicious of tempting ads that forget to mention the fine-print limitations. It's an uncertainty that adds to the perceived price of an offer (another way of saying that people are willing to pay more to avoid surprises - which you cited as an advantage for prepaid/unlimited plans). Since a linear price naturally solves the carrier's "abusive users" problem, the carrier will be less likely to add silly limitations like "no tethering", and that'll lower the perceived price.

> yet somehow expect that prices will go down. Don't you see any problems with that idea?

While I described a few problems with your reasoning, I've not explained why I think a linear price would lower prices. The reason prices would go down (appart from techonolgical/infrastructure progress) is that simpler prices are easyer to compare. Easy comparision drives competition up and prices down.

> Once more: “big users” are not a problem at all

Really ? How come "unlimited" isn't the only plan available worldwide then ? Heavy users cost more than light users.

> Bursty users are problematic, and flat rates encourage them

As you say, the usage pattern (bursty/unpredictible vs regular) matters more than the actual amount, but I wasn't talking about the usage pattern, and none of the discussed pricing schemes (unlimited, pseudo-unlimited, prepaid, linear...) influence the usage pattern much :

> if I know that I don't have enough monthly traffic to watch football matches from airport then I'll go and find WiFi access point, if I pay cheap plat rate then I'll just use my 3G or 4G connection

What you're describing here is not a problem of burstyness, but a problem of total usage. You'd get the same behaviour (deciding to use wifi instead of 3G) if you had been using data at a steady rate during the month.

Burst usage means that you are using a lot of data in a short time. But a single bursty user isn't likely to matter to a carrier : what's important is the aggregate usage, with the network suffering only during "prime-time".

For this particular problem, the per-gb pricing scheme can help smoothe the usage graph by setting a higher price during prime-time (like we have for electricity). You could actually do the same with prepaid plans (substituting Gbs by arbitrary units), but it's a bit more abstract and therefore complicated/unapealing to the customer.

I just bought a €10 prepaid card with 10 GB of data to use, and that includes €5 I can use for phone calls / SMS.

This gives me €0.5/GB price. Provider: MEO.

Tethering allowed.

The thing about this price strategy that will perhaps not strike you as coincidental if you know who BT Openreach's biggest customer is (hint: It's BT retail, part of the same group) is that it punishes ISPs for being small.

Random variation caused by the behaviour of a small number of individuals gets drowned out when you're a big ISP, so in fact BT retail can and do offer "unlimited" service (e.g. on "Infinity 2" their most expensive and best performing normal home offering) even though by law they're paying the same fees you've described. A handful of customers may use the service very heavily sometimes, but compared to millions of households on BT's services this handful doesn't matter.

Still, I'm not sure your description is quite right, or rather, I don't think it tells the whole story. The ISPs are not so far as I can tell obliged to take more bits than they want. Smaller ISPs often have only a gigabit feed from BT. No matter how much BT tries, no more than a gigabit will fit through that pipe and they only get to charge for a gigabit. If 500 users on 80Mbit FTTC who buy service from TinyISP are trying to download the entire Breaking Bad simultaneously, too bad, they're rate limited by congestion at TinyISP's 1Gbit fibre in some DC in an industrial estate. Of course those customers may not be happy, regardless of whether they were promised "unlimited" or not, but Openreach doesn't get to charge for 40Gbit transit that didn't and couldn't take place over that fibre.

Eventually this all shakes out. But it shakes out for the biggest players first. I'd guess that the average (mean) household with Infinity 2 only ends up using a hundred gigabytes per month if that. BT retail makes a profit on this product even though in theory every user could run it full tilt 24/7, chewing say 20TB per month and bankrupt them.