:: economics, computer

Do not eat the free lunch: it has probably been poisoned.

On 2015–03–12, Google announced the closure of Google Code, the latest in a succession of services they have switched off over the last few years. This is a perfectly reasonable thing for them to do: they are a commercial organisation and need to focus on the things that make them money — selling advertising and acquiring as much personal data as possible from users of their services to help them do that — and hosting source code repositories is probably not a very efficient way of scraping such data off people.

So there is no reason to complain about this, however annoying it is: it was a service that was being offered for free, after all. But of course, a number of people will be significantly inconvenienced when things like this go away because they have come to rely on them, either personally or as part of their business: this turns out not to have been the smartest idea. The interesting question is whether they will learn from the experience and what they’ll do to stop it happening again.

Too cheap to meter

The cost of many things related to computers and networking has fallen dramatically over time, and continues to fall. We’ve also found out that more things are related to computers and networking than we realised: music, still and moving images, books and so on. In particular the marginal cost — the cost of making an additional copy of something — has often become extremely low because the cost of storing and moving data around has become very low indeed.

It’s quite tempting to think that ‘very small’ is the same as ‘zero’1, but this is a fatal mistake: if it costs nothing to do something then it costs nothing to do an arbitrary amount of it, while it it merely costs a very small amount then you can make the cost arbitrarily large by doing enough of it. If something with a non-zero cost, however small, is given away for no cost then the giver is in a dangerous situation: nothing is too cheap to meter unless it is free2, and nothing is completely free. So if an organisation is giving away a service ‘for free’ there is reason to be suspicious: either things are what they seem, in which case they are going to run out of money at some point and disappear, or things are not what they seem.

If things are what they seem there’s a fairly obvious problem: you probably don’t want to build anything substantial around a service which is inevitably going to evaporate when the organisation providing it falls off a cliff.

Things are more interesting when they are not what they seem: how is the organisation making money if they’re providing something for free?

The first hit is free

One approach is the one traditionally used by people who sell recreational drugs: you get a free taste of the service, but the taste will be limited in ways which make it annoying to use and probably will prevent you from doing some things altogether. Eventually, all being well, you become both dependent on whatever it is they are pushing and frustrated with the limitations of the free version and decide to pay for the unrestricted version.

There is nothing very wrong with this approach: you’re getting something for free, after all: just not what you really wanted. And you have the option of paying for that if you choose to: that’s what the supplier wants you to do, after all. This is not, however, a very good long-term solution: the supplier could always simply stop offering the limited version or, worse, stop offering any version at all.

The place where there is no darkness

Another approach is one you might associate with a person wearing suspiciously well-cut clothes that you once met late at night at a crossroads somewhere in the deep south. Now you can play the guitar pretty well, but can you remember just what it was that you you bargained for your new talent and when the debt will become due?

This is not the sort of bargain you want to make3. But it is exactly this sort of bargain on which a lot of large companies have built their businesses: they provide you with some service, and in return you provide them with your soul, which they then package with a lot of other souls and sell on to you know not whom. They’re not, in fact, interested in providing the service: they’re in the soul collection and resale business.

A lot of people quite clearly think this is all just fine. They’re quite happy to trade their souls for an endless set of distractions: perhaps the point of the distractions is so they don’t realise just what it is they’ve lost and what exactly it was they gained in return if anything; or perhaps they have souls which are not very valuable and the bargain is a perfectly reasonable one, after all.

There is worse. When you met someone late at night to make this sort of bargain, you made very sure that you got a bit of paper with signatures on it detailing just exactly what the deal was4. That’s not how the deals that are made so willingly now work: you get something momentarily useful or amusing, and in return you irrevocably give away something of yourself, and that’s as far as it goes. If, later, it becomes convenient for the entity you did the deal with to stop providing whatever entertainment it was, then one day it simply goes away and you have bargained your soul for air and darkness, and precious little of that.

Better living through chemistry

The answer is quite conventional. If there is something you want and on which you might come to rely, then you sign a contract for it: a document which obliges you to pay for it, and in return obliges the provider to actually provide the service.

Contracts really do three things.

  • They make it clear what exactly is being bought and sold, and avoid the ‘too cheap to meter’ fallacy I talked about above: the contract should detail what you get and what the limits on it are — how much bandwidth or storage you can use for instance — and what you are paying for it, which should generally not be ‘your immortal soul’.
  • They ensure that the interests of the consumer and the provider are the same, or at least similar: the consumer wants a service or a product that works well, and the provider gets paid if they provide that.
  • They specify what happens if the contract is terminated: what the responsibilities of each party are and what they are not. For instance the organisation providing your cloud storage might be obliged to give you a way to recover your data.

The second point is particularly important: for a contract to be of any use at all both parties have to get something out of it: you can sign a contract with someone to provide you some service for free, but if they decide to stop doing that what are you going to do — perhaps you could ask them for your money back?

But, well, this is a very conventional and rather boring answer: surely we all live in a future where all this awful tedium is no longer needed. Wasn’t the internet meant to do away with all that? What happened to the gift economy? Are there no flying cars, after all? Sadly, no, the internet didn’t change all that: it simply enabled a collection of large corporations with toxic business models to fool a really large number of people. There are no flying cars.

On 2015–07–16 SourceForge fell over: perhaps it will recover, this time. Once upon a time it was the bright future of source code hosting: who knows what will be lost when it finally goes away?

  1. It is particularly tempting to people who want to make the argument that ‘no harm is done to the artists if I just download this song, because it costs nothing for them to deliver an extra copy: they have already been paid’. I am not sure if this argument is ever made in good faith, but it’s very easy to see that it doesn’t work by reductio ad absurdam: what would happen if everyone made it? However I don’t want to get sidetracked by that here. 

  2. ‘Metering’ may be simply restriction of supply — for instance a limit to the amount of data you can transfer, which may not seem like metering although it is. In the limiting case the limit may be the physical capacity of the system: you can only transfer so much data per month over link with a given bandwidth. I suspect that the original ‘too cheap to meter’ claim was made based on this assumption for domestic electricity usage (if it was ever really made at all).  

  3. Well, perhaps it is a bargain worth making, but probably not in exchange for anything related very closely to computers. 

  4. Perhaps in the hope of later renegotiation, although it generally seemed to turn out that the counterparty had rather better negotiation skills than you and, obviously, expensive lawyers with dead eyes.