Not Invented Here

- by Paul Murphy, -

While watching CNN last week I suffered one of those brain spasms that leave you holding an idea you can neither rationally assess nor forget about. I still can't assess the idea --and I'm about to ask you for help on that-- but my attempts to do something with it did give me a deeper insight into the forces that made the Linux development process so successful.

The idea that started this has absolutely nothing to do with open source, Unix or Management. I was watching a CNN report on casualties in Iraq when it occured to me that scratch and lose lottery tickets could be used to address a critical intelligence issue.

Right now American intelligence officers in Iraq and Afghanistan face a problem akin to spam email - they get thousands of petty criminals and vindictive fools asking for money and favors in exchange for mostly useless or erroneous information while the people they want to hear from, solid citizens with real knowledge, extended families, strong social ties, and something of substance at stack in the outcome, are mostly silent.

There is a key reason for this: a petty criminal who comes into a few dollars selling information doesn't expect to explain the income to anyone and sees his action as placing only himself at risk of retribution. An honest citizen, in contrast, cannot easily explain a windfall earned by selling out his former oppressors and must, therefore, either hide it from everyone or expose his family to the risk of violent retribution. As a result the people an intelligence officer most wants to hear from act like bankers being asked for loans: they'll only contribute if the final outcome is so clearly assured that their information is valueless.

My idea for fixing this is simple: give away, with as much publicity as possible, a large number of small denomination scratch and win lottery tickets and then use winning tickets of known denomination to reward informants.

In Iraq, for example, $25 million - the amount offered for information on Saddam's whereabouts - would let military personnel give away 250,000 tickets per week for six months with odds of about 1:25, 1:100, and 1:1000 of winning a week's, month's, or year's average Iraqi wages while retaining one fifth of the winning tickets for use as rewards.

Informants could then be given tickets of known denomination, scratch them in public, be widely seen talking to American officers as they pick up the cash, and have their entire families help them rejoice in their new wealth with the risk of retribution decreasing in direct proportion to the extent to which a black market in tickets evolves.

I thought this was a great idea, cheap, easy to implement, and probably infinitely more effective than offering multi-million dollar rewards to people whose limits are measured in hundred dollar bills. Unfortunately, I'm a Canadian with no access to the US chain of command or political process and thus no obvious way to get it in front of people who could either tell me what's wrong with it or take action.

I tried a few possibilities and got a quick lesson in "not invented here" syndrome at work. Organizations like the DIA, for example, have groups dedicated to deflecting well meaning fools who want to profer ideas for consideration, but they don't have a mechanism for accepting outside ideas. When you think about it, you realize that organizational imperatives mean that there's really no other way for them to behave and that the same forces must affect almost all larger organizations.

This column, for example, has a largely technical audience. Most of you may not be any more qualified than I am to tell whether creating a market in low denomination scratch and lose tickets makes sense as a way of encouraging middle class informants to come forward (and if you are, or know someone who is, please contact me) but its a certainty that many of you have ideas which could help companies like Sun or Red Hat win their particular wars - and no effective way to get them evaluated or acted on.

With this in mind I went back and reread Eric Raymond's account, in the Cathedral and the Bazaar of the early days of Linux. What I discovered is that you can sensibly look at the process as he describes it and claim that what made the open source process so insanely great was the almost complete absence of "not invented here" syndrome among its earliest and strongest backers -including both Linus Torvalds and Eric Raymond.

That's a new thought, at least to me; and suggests that the bursts of creativity coming out of conflicts such as those recently seen in the BSD development world are the best possible harbingers of future success because almost all of the sides to the schism actively recruit new members, and new ideas, using no stronger a criterion than that the ideas brought in fit their particular tent.