Clear questions and runnable code
get the best and fastest answer
X-prize software challenge?by BrowserUk (Patriarch)
|on Oct 15, 2004 at 14:05 UTC
Update: Please post entries to the challenge subordinate to Re: X-prize Suggestions here please! and leave the main thread for discussion arising from this node itself. Thanks--Buk.
The following was sparked by reading this X-Prize for world's 'Holy Grails' article on the BBC.
Having read it, including the bit about a Skype having recently won a (non-X-prize) award for it's voice-over-IP service, I got to wondering what I would consider a suitable challenge for the new X-prizes. Especially with respect to software (and to keep this on-topic, no matter what it is, there would always be at least some scope for 'it' having a Perl component or two.)
The usual suspects crossed my mind. Artificial Intelligence; Natural language processing; a generic way to optimise algorithms for 'solving' NP complete problems; but each of these seems to miss the mark that:
that X-prize problem proposals must have a good chance of succeeding within a reasonable timescale.
Then I got to thinking about why I felt that the above problems failed to satisfy this criteria? The conclusion I reached is that they are too generic; too open in their specification. The recently won Ansari X-Prize (Nice one Burt), had a very clear judgement criteria. Two flights by the same vehicle, carrying 3-persons (or equivalents), to a height of 100 kms or greater, within a timeframe of 2 weeks. Clear, concise and unambiguous*.
*Unless your a Ferngi lawyer who might consider the definition of "2 weeks" ambiguous:)
More importantly, it didn't carry with it any unnecessary additional criteria. Whilst most of those attempting to pursue the prize settled into emulating the Saturn V model, this was in no way encapsulated into the specification. Unlike say, the world land speed record specification which applies an additional criteria beyond the "average of two runs with 1 hour", but specifying that the vehicle must have 4 wheels--the undoing of at least one very good attempt in the last century.
but if you look at the "judgement criteria", it comes down to this phrase:
...it will be possible to programme computers, [...] to make them play the imitation game so well that an average interrogator will not have more than 70 percent chance of making the right identification after five minutes of questioning...
The problem with this as a judgement criteria (as opposed to an argument in discussion--No. I'm not critisising Alan Turing:), is that it is much too vague to allow for unequivocal verification.
That last critisim may seem esoteric, but if I consitantly and invariably ask "Why?", is that allowed?
How long would a "real human being" tolorate that conversation before replying something akin to "P*** off you moron!"?
So, for an X-prize challenge to be viable, it must be:
So, the point of this ramble. What do you consider would make a good candidate for a Software X-prize?
Ideally, you should supply both the specification of the challenge and the judgement criteria.
It would also be nice, if the denizens of this place permit it, if:
The idea being to refine the original idea in the light of the collective experience to produce a better definition. Each challenge posted would be a separate entity without there (necessarially) being an overall judgement upon which is the overall 'best'. To this end, I will post my favorite idea (if I ever get around to posting this post), as a reply to it.
It may be that this post will itself become subject to critism, discussion or censure, in which case, it might be better to start a spearate thread for the posting of actual challenges? Anyway, I will need time to refine my idea so, I'll post this first and see what the consensus is before posting it.