It's long been an open secret that many major telecommunications companies, including Internet service providers (ISPs) and cellular data providers, impose specific limits on the volume and type of bandwidth consumed by their customers.
"Open" in the sense that these companies almost universally reserve the right to impose such limits, and occasionally make public statements defending their right to do so. "Secret" because the companies rarely reveal their specific limits, and because it seems like these limits are constantly being rediscovered by people who ought to know better.
Two such shocking cases are before us in the news this week.
Next, Verizon Wireless settled a dispute with the Attorney General of New York state stemming from its use of the word "unlimited" to describe its wireless Internet plans, which were very definitely limited by secret company policies to a total of five gigabytes per month.
In both cases, the primary issue is fairness...and the contradictory definitions of fairness held by customers and their service providers.
For ordinary customers, fairness means being able to use all the bandwidth they're paying for, continuously, forever, with whatever software they like. Providers apparently want to define fairness as satisfying the maximum number of customers while still making a profit.
I have to take the customer's side on the definition, but you can't argue with the business goal. Without profit, there's no business. Some ISPs don't care about profit, of course--chiefly the ones in foreign countries where the Internet has been subsidized by government for political reasons. But here in the US, Internet services have to make a profit, and that means bandwidth supply can never meet bandwidth demand.
On one hand, ISPs could build out their networks to provide the maximum rated bandwidth to all subscribers at once...but that would be extraordinarily expensive, and wasteful because that kind of load would never be seen in the real world. On the other hand, they could promise the lower throughput they truly can guarantee, then deliver more when possible...but that would look terrible as a marketing campaign. Verizon's 5GB limit, for example, works out to about 2 kilobits per second. Can you imagine advertising "2-kilobit high-speed Internet!"? Nope. Supply and demand isn't just a good idea--it's the law.
So ISPs use what's called "statistical multiplexing"--sharing bandwidth among multiple users on the theory that they won't all use it at once. A cable-modem service such as Comcast's might share some 30 megabits/s of bandwidth among dozens to hundreds of subscribers in a single neighborhood.
Although a mere half-dozen users could overload this shared channel, in practice, the natural load-sharing behavior of Internet protocols provides a reasonable experience for all the users online at any given moment.
Except when it doesn't. An active BitTorrent user might have dozens or hundreds of transfers in progress--multiple files, each with multiple partial downloads in progress. Each transfer competes for a share of the channel's total bandwidth, so a handful of BitTorrent users on one cable-modem service could consume the vast majority of the available bandwidth.
This gives us a third definition of fairness, the BitTorrent definition: fairness is being able to take everything you can get without regard to other customers on the same service.
So is it more unfair for Comcast to cut back on BitTorrent traffic, or for BitTorrent users to exploit the Internet's load-sharing behavior? I have to take Comcast's side on this one.
I was writing a reply to a badly-written article and subsequent ignorant comments about the Comcast situation over on Engadget earlier today when I realized this situation creates a market opportunity for some software company. So I figured I'd write this instead of posting the reply there.
To summarize: although I suspect most BitTorrent traffic consists of pirated software, music, TV shows, and movies, there's also some important, legitimate content on BitTorrent--Linux distributions, collections of classic e-books in the public domain, Linux distributions...wait, I mentioned that already. Actually, there probably isn't that much legitimate BitTorrent activity. But however much there is, it deserves to pass unmolested on Comcast and other Internet services.
It's unfortunate that these files are being subjected to Comcast's traffic shaping, but that's what happens when people put legitimate content into a distribution channel designed and optimized to facilitate piracy.
So clearly we need a separate public-access peer-to-peer system. How would it differ from BitTorrent? Well, the content would have to be legitimate, and probably so. That means a central authority and a master list of authorized content.
Of course, this is basically the Joost business plan. Joost is a peer-to-peer file-sharing service for video. It seems to me that either Joost ought to expand the concept to include other kinds of content, or someone else ought to get into the business. It'll have to be ad-supported, I think, because if anyone has to pay for the service they'll probably just keep using BitTorrent instead.
Such a service should also get support from ISPs since it would actually reduce their overall bandwidth load for legitimate file sharing. Peer-to-peer distribution is very efficient--that's why Joost uses it--so the providers could leave this new service unfiltered while BitTorrent gets quite legitimately squeezed out of operation.
No doubt such a plan would be highly unpopular with active BitTorrent users, but if that's the price of keeping ISPs profitable--and honest in their marketing policies-- I think it's a reasonable price to pay.