Signal to noise

A friend of mine forwarded me this FT article yesterday, as part of our discussion on the importance of Open Source. Not only is the logic flawed, but the FUD-based swipe at open source is especially cheap.

FT.com site : DROWNING IN NUMBERS: The rising cost of ‘free’.
24 January 2007
The rising cost of ‘free’

The 21st Century appears increasingly benevolent, particularly regarding technology, products and services. Consumers and businesses are being offered an ever widening range of goods, for a ticket price of zero. Apparently.

Open Source has disrupted the software sector, to a large extent because of its perception as a free product. But once you factor in installation, maintenance and customisation, Open Source may no longer appear cheap.

The impact of free services can, overall, be destructive. The minimal marginal cost of sending e-mails has enabled spam to flourish. Spam now represents more than 95 per cent of all e-mail sent, equivalent to more than 60bn messages each day.

Free e-mail also allows viruses, worms, spyware and other malicious code to be carried gratis to millions of unwitting recipients. Free though the service may be, it is estimated to have cost consumers $7.8bn to repair or replace infected PCs during 2005 and 2006.
By contrast, paid-for SMS services, on which consumers spent approaching $100bn in 2006 alone, were relatively spam-free; most likely because the economics of spam simply do not work within a commercial setting.

Instant messaging (IM) has suffered a similar fate. The average IM user can expect to be interrupted with junk messages about five times a day, and those interruptions are forecast to rise to 27 times a day by 2008.

The genuinely free lunch may still be some time away.

*Statistics and opinion by Nick Pflaeger, technology partner, and Paul Lee, research director, at Deloitte.

Here’s the argument in a nutshell:

1. Open Source is often regarded as zero-cost
2. Other zero-cost services exact an unwanted cost, which is often malicious
3. Paid-for SMS services have little of this malicious unwanted cost
4. Ergo, “The impact of free services can, overall, be destructive.”

Analysis
First, open source does not represent a zero-cost solution (and those who consult on the implementation of open source solutions are negligent if they imply that). The definition of open source is code that is open to everyone to view, amend and enhance. Not for everyone, to be sure, but it ensures that if the code is not fit for purpose, it is possible to modify it to do so. The ‘free code’ of open source that he refers to is confused with the zero acquisition cost. (The ‘free’ in ‘free speech’ is not the same as ‘free’ in ‘free beer’).

In industrialised countries labour costs are high. The investment of time in learning a new software package is significant (which is why well-designed human interfaces, can save time and by extension, money). The time it takes to adopt a new package is a very real cost, irrespective of whether that package is free. The free services he refers to provide a benefit to the users, for which they trade the currency of their attention (a subset of our available time).

Second, the malware they refer to – which is egregiously ubiquitous on Windows machines, and completely absent on my Mac – has for example, nothing to do with the cost of that operating system, and everything to do with the size of the user base and the security weaknesses in the system. That contradicts his original premise: the impact of paid-for services can also be destructive.

Third, paid-for SMS services are malware free, because they are such a simple technology (160 characters, piggybacking on a voice transmission infrastructure) and because they work within a walled garden of operator constraints, not “because the economics of spam simply do not work within a commercial setting”.

Finally, there is an egregious non-sequitur between second and third paragraphs. One moment, the authors are talking about Open Source, the next about spam, viruses, worms and spyware. And by implication, linking the two. Naughty.

The point that the authors miss is that increased connectivity (or ‘connexity’, to use Geoff Mulgan’s phrase) always means an increase in background noise. But where the noise level becomes too high, we change our behaviour to reduce it to a level which is comfortable (switch mail providers, use spam filters, pass laws to put a financial burden on spammers). We are accustomed to wall-to-wall advertising (something that would shock, nay offend, a person a hundred years ago) but we have adapted to filter it out. And now, we can do so using software (my NY Times and Economist.com sites are beautifully rendered sans ads).

Still, we have always been willing to put up with noise in order to get want we want. 150 years ago, Ralph Waldo Emerson wrote: “We are too civil to books. For a few golden sentences we will turn over and actually read a volume of four or five hundred pages”.

I wonder what he would have said about today’s TV?

Order! Order!

Earlier this week, I was at a dinner party where I heard this extraordinary story. One of the guests had been invited to attend the recent Arsenal-PSV Eindhoven match. He’s not an avid football supporter, but faced with the excitement of the game, he cheered when Eindhoven scored their 83rd minute goal. Apparently he was the only one in the box to do so. Immediately, he and the other members of his party, which included the Arsenal season-ticket holder who had invited him, were asked to leave the box and the game. I think it didn’t help that he is Dutch.

One of the other dinner guests, a keen football fan, expressed approval at this decision (“Most rightful judge! The court awards it and the law doth give it”) explaining that as an Eindhoven supporter (which he is not) he shouldn’t have been in that part of the stadium (tough call when you are invited there as a guest).

freedom of expression (but just not around here)

When questioned why this should be, she gave examples of supporters getting beaten up because they were in the wrong part of the stadium, explained the proscription of sartorial choices in specified parts of the ground “you can’t wear that shirt/scarf in here” because they might also get beaten up, but suggested that there were ‘neutral’ parts of the ground available for types like him, who wished to express an independent opinion.

kick him out because such behaviour has incited violence in the past

When this debate was broadened to the general condition of whether it was acceptable to kick him out because violence (or the fear of it) could be used to enforce normative behaviour – we were relieved to hear that she thought that “violence was unacceptable” though we were treated to colourful examples of how in any case, you shouldn’t tempt fate where there is a risk of physical violence: “you don’t go flashing a Rolex in downtown Johannesburg”.

The discussion in its general form then centered around the transgressing of micro-social norms (ie. is it socially unacceptable to cheer for the opposing team when you are not surrounded by the opposing team’s supporters?). She proposed that it was trangression of these norms that made it correct to segregate supporters and remove those who don’t adhere to these norms.

segregate supporters to maintain in-group cohesion

Analysis

In order to characterise reasonable choices, one uses the concept of preference relations. Without getting into broader issues about the debate (like the tacit acceptance of violence), I was fascinated by her ordering of the following choices:

1. protection of minority expression
2. fear of the threat of violence
3. maintenance of in-group cohesion

2 and 3 can co-exist, of course, but it’s the ordering of these preferences by her that is important, as it seems to violate contraction consistency.

Amartya Sen‘s contraction condition states that if you have a set of choices from which you express a preferred choice, and subsequently have a reduced set that also includes your original choice, then your preference will stay the same.

But here we have:

• 123 -> 2 (private box condition)
• 23 -> 3 (general condition)

This could be explained by Herbert Simon‘s ‘satisficing criterion‘, where the threshold utility for adequacy of an alternative can be less than the maximum attainable from the set (eg. faced with the purchase from a range of 99 different personal computers, you are satisfied with not going into a rigorous analysis to find the perfect choice). But this applies to large sets, not the simple case we have here.

Of course, the broader issue here was the value of artificial in-group creation, which she was wholly in favour of, at least in the case of (British) football. The situation she describes is an excellent example of this, both in its use of physical separation (supporters in different parts of the grounds) and the threat of unfair distribution of resources (only one winner from the game) and these and their consequences are well-documented in the literature (eg. Sherif’s Robber’s Cave 1961, Blake and Mouton 1979).

Needless to say, as the evening wore on and the trees were darkly swayed, Godwin’s Law was also obeyed.