Staying Employed in the Turkish Automaton's World

by Kurt Cagle

Related link:

Welcome, one and all, to the Metaphorical Web. My name is Kurt Cagle, author of books on XML, occasional blogger, computer consultant, and pundit wannabe. I want to thank O'Reilly for giving me the privilege of writing a blog under their banner, and hope to be able to take advantage of it to expound pell-mell on XML, the industry, programming, programming ethics, and the world of the technically bizarre.

I've noticed an interesting trend lately, an extension of a practice that's been going on for a while but seems to be gathering steam. Recently, A9, the "new technologies" off-shoot of Amazon, launched a service called the Amazon Mechanical Turk. The concept here refers to a rather intriguing scam performed by the Hungarian nobleman Wolfgang von Kempelen in the late 1760s. Herr Von Kempelen had a most ingenius automotaton that he created, in which a mechanical man astride a complex block would challenge (and as often as not beat) all comers in a game of chess. He intrigued much of Europe with this particular robot, including such luminaries as Benjamin Franklin and Erasmus Darwin, and it was only much later that it came out that the secret of the automaton consisted of a midget with a prodigious talent as a chess grand-master, who was carefully hidden within the box amidst mirrors and gears.

Amazon's service uses the idea of the Turkish Automaton to handle a rather intractible problem with many tasks today. In essence, the idea is to take the notion of web services and turn it on its head - take those tasks which can't be readily done by computer - anything from identifying photographic content to writing copy to evaluating essays - and turn them into web services that people can bid upon to complete. Once the process is completed and approved, the results are then dropped back into whatever computation or application is currently being done and life goes on.

Similar ideas have been tried before, with mixed results - posting jobs to be done in a marketplace, answer services such as Google Answers, and related efforts have occasionally run afoul of simple market dynamics, the difficulty of setting up "atomic" tasks and a need to establish a reasonable means for determining reliability of service both serving as barriers to adoption to these types of offerings. However, I've also seen, and used, other types of "bulletin board" services, including in my most recent move (with both good and bad results, including some damaged furniture).

However, I suspect that while the kinks may still be a few years getting out, the future does seem to point toward the notion that business is transforming into the individual entrepeneur model that was extolled (perhaps somewhat ridiculously) by much of the Wired digiterati. I'm not necessarily sure that this is altogether a good thing, but before going into that, its worth looking at where this process is leading.

If you take a look at eBay, what you see is the rise of a whole new class of mid-range entrepeneurs amidst a sea of weekend merchants. The weekend merchants may occasionally put a few items up on eBay, but the mid-range entrepeneurs are now acting as middlemen to aggregate multiple other sellers into a package, and then selling into that package. They are, in essence, acting as the electronic equivalent of department stores, producing little themselves but acting as agents for others who have neither the time nor the desire to be involved with eBay full time. For these people, eBay IS their business.

I suspect that this will likely end up being the path taken within the Amazon model as well. Having run my own business, I can say that a significant proportion of the time and energy in that business goes into the process of securing work and accounting for that work when its done - operations which are necessary, but for which in general you do not get paid. Initially, the "contractors" on the other end of the equation will be independents, but while the overhead is reduced somewhat in this kind of model, it is replaced by establishing the business process and handling the monitoring of contracts. Those that succeed will not be the ones doing it themselves, but rather small to medium businesses that essentially specialize in being generalists.

A system like this is not, however, necessarily all that beneficial to the seller of the services (nor to the buyer, but for different reasons). The seller ends up with the headaches associated with running any small business, but without many of the legal or governmental benefits that acrue to businesses. Indeed, because current tax law requires that you maintain taxes on every 1099 type job (i.e., contract based) that you take, doing the taxes for a business that's heavily built upon this model can become nightmarish. You become responsible for your own health care and benefits packages on top of that, and depending upon how you're declared you may get hit with self-employment taxes as well.

Additionally, while the model currently proposed for Amazon requires that the money for the job be placed in escrow, the possibility that the work will not be found satisfactory is high, and the risk of litigation in this regard should be seen as significant.

For the buyer of these services as well, there's a certain degree of risk, accounting overhead and unexpected costs that enter into this equation. I've worked on a contract basis for years, and have never been in one where a client can specify exactly and precisely what he or she needs out of the gate. Instead, most such projects are iterative in nature, with the client refining their vision with each successive cycle of work and review.

Ultimately, this raises a question that has disturbed me more than once about such technology-based solutions ... in the attempt to create ever increasingly automated systems are we reaching a point where we're sacrificing many of the things that contribute to quality - solid design, effective communication between parties, succesive refinement of concepts and the time to think through problems? Moreover, are we destroying the social aspect of business in attempting to make business more machine-like?

One of the concepts I've always felt to be somewhat ludicrous was the use of UDDI as a means of automating the sales process - you post a web service with your request to the appropriate UDDI server, which will in turn send you back the vendor which best meets your criterion, based upon the SOAP bundle you send. From a technological standpoint, this sounds like a can't lose proposition. From a sales standpoint, however, this is an unmitigated disaster, because sales are ultimately social transactions, not automated ones. I do not find it surprising that UDDI has not even remotely taken off except in very narrow environments.

The principle challenge of this decade of the twenty first century is to move the web from being a largely static medium for displaying content into a framework for social networking. I suspect that true eBusiness will likely only take off once we recognize that the most powerful aspect of computers is not in their processing powers but rather in their ability to act as communication devices that best facilitate human to human transactions, that take advantage of the social networks, and not just the physical ones, that bind us all together.

Should be interesting to see how well we do with that...

So what do you think about the Turkish Automaton? Flash-in-the-pan idea, or a harbinger of things to come?