My Own Pirate Radio

If you’re interested in the shifting line that demarcates human and computing capabilities, check out this podcast from Jon Udell. It’s a conversation with Nathan McFarland and Benjamin Hill, both of whom are working on so-called “collective intelligence” projects that farm out tasks to ad hoc networks of people. According to Jon, “Nathan runs CastingWords, a podcast transcription service that uses Amazon’s Mechanical Turk to distribute and coordinate [transcription] work to people. Benjamin’s project, Mycroft, packages up puzzle-like tasks in ways that people can interact with on web pages.” You can find the text transcript of the interview here.

Amazon launched Mechanical Turk late in 2005. It really got me thinking about what types of tasks are undigitizable, the interplay between collective intelligence and outsourcing, and related ethics and economics issues including worker conditions and incentives. It’s interesting to see concrete examples of people working in this area and exploring all these issues in great depth. For example, here is how Benjamin Hill approaches the space:

BH: …We’ve started to think of the whole knowledge workspace as divided on one axis along people’s opinions all the way to things that are absolute. Like people’s opinions would be “tag the image”, absolute would be “OCR check this one line of text”. There’s one correct answer versus “it really matters what people thing” and then along another axis we’ve got everything from easy where it just takes a few seconds of time, all the way up to very difficult where it takes a longer amount of time. If you can imagine both Mycroft and CastingWords occupying different bubbled areas of that knowledge workspace, so actually we’re going after different targets and I think both have a lot of value at that point.

Here Jon Udell gives some examples of the types of work that can be tackled in this way:

JU: …I’ll just read through a couple of the kind of examples you give in your commerce net paper. … “is this image inappropriate for children over seventeen”, “annotate this image with descriptive text”, “what is the text in this captcha image”, “how much do you like the clip from this new pop song”, “which hairstyle makes you trust this politician more”. So actually a whole lot of market research survey kind of stuff would seem to fit really nicely into this model.

Both organizations are still experimenting with how to incent workers. In Mycroft’s case it sounds like the work is sometimes purely voluntary and sometimes in exchange for non-monetary rewards. In CastingWords’ case workers are paid small amounts per task completed:

NM: …I think almost all of our workers are people picking up a job here and there. A lot of them have indicated that they are working at the same time as they are working on our stuff at something else, some other job that doesn’t require anything more than a physical presence for whatever reason. A lot of the other ones are stay at home mothers or something like that. This is just a subsidiary income for most of these workers, and there are a lot of them which is part of the reason I think. Our workflow isn’t, we don’t have so much that we can maintain a steady flow for the huge number of workers we have.

It’s interesting stuff. Ultimately I believe technologies like these will change the nature of work, provide work opportunities to more people around the world, and even open up entirely new lines of employment by making certain kinds of tasks much more malleable, portable, and tractable.

A bunch of things lately have really got me thinking about what human qualities can and cannot be digitized. (This is inbetween more mundane tasks like shopping for winter tires. I’ll bet Greek philosophers didn’t have to shop for winter tires right in the middle of theorizing. Geez.)

Here’s a list of things that cannot be digitized:
* Sensation. Many people have been working on this for decades — vision, voice reco, smell, tactile, taste — with slow progress. I do think this will eventually be cracked for many practical applications. But not all?
* Aesthetic appreciation, e.g. of art, music, sunset, poetry, that sound you hear in sea shells
* Emotion
* Creativity
* Dialog. Remember Eliza? I wonder what the state of the art is now. Not much better I’ll bet, because you need…
* Understanding of context. Surely there is a shorter word for this concept? Lack of contextual comprehension is the reason voice recognition is stuck at 95%.
* Consciousness. Although reading wikipedia’s definition, I think humans are losing ground to machines on some aspects of consciousness. [5 minutes pass….] Wow, it’s worth reading that definition and just spending a few minutes clicking the outbound links from it.

So we still have some high ground on the algorithms. But the list is disconcertingly short! And it isn’t normalized; “consciousness” seems like it subsumes so many other things. What have I missed? Help me make a better list, or point me to someone who already has.