The whole trial and error thing is known to computer science as genetic programming. There are some important gotchas.
The first is that it is very easy to hit a local maximum – somewhere you think is the best fit whereas there is a better one somewhere else (see Wikipedia). The usual solution is to keep mixing in some real screwballs as they’ll help probe other areas.
The second is that you have to have some way of measuring fitness in order to pick winners (and losers). Evolution does that by looking at fecundity. If you were using trial and error for tax policies then you’d need a way of measuring what was most promising, and that could be fairly hard.
The trial and error principles can also be applied beyond science. Lookup “evolutionary art”.
Question for Roger… Did you leave this comment in the hopes of teaching other readers or with the hopes of educating the author? If it was to educate the author you might try to lose some of that GIGANTIC EGO your pushing around as he has published several books on the subject with many references, I highly doubt you can say the same. If not, thanks for the extra info.
None, whoever you’ll be.
I wonder why you had the urge to critic Roger
I don’t what he was trying to do, I know that he brought a great piece of information.
Although it’s not new to me, I tend to forget about it sometimes and I’m sure I’m not the only one.
Thanks for that one Roger!
Hi Tim, this talk was very inspiring for me and I felt especially touched by the emotional end. I will keep up my trial & error lifestyle, from now on without regrets. Thank you!
Very interesting talk, and I agree with Tim on his insights. One question did come to mind which I’ve not been able to figure out –
By saying that certain people tend to have the God Complex, is he not showing a God Complex as well? Or by saying that trial and error works, is that not a God Complex?
11 Comments
Gareth says:
Really fascinating talk. Inspiring, too. I’m looking forward to reading Adapt.
18th of July, 2011Roger says:
The whole trial and error thing is known to computer science as genetic programming. There are some important gotchas.
The first is that it is very easy to hit a local maximum – somewhere you think is the best fit whereas there is a better one somewhere else (see Wikipedia). The usual solution is to keep mixing in some real screwballs as they’ll help probe other areas.
The second is that you have to have some way of measuring fitness in order to pick winners (and losers). Evolution does that by looking at fecundity. If you were using trial and error for tax policies then you’d need a way of measuring what was most promising, and that could be fairly hard.
The trial and error principles can also be applied beyond science. Lookup “evolutionary art”.
18th of July, 2011none says:
Question for Roger… Did you leave this comment in the hopes of teaching other readers or with the hopes of educating the author? If it was to educate the author you might try to lose some of that GIGANTIC EGO your pushing around as he has published several books on the subject with many references, I highly doubt you can say the same. If not, thanks for the extra info.
18th of July, 2011Yoni says:
None, whoever you’ll be.
18th of July, 2011I wonder why you had the urge to critic Roger
I don’t what he was trying to do, I know that he brought a great piece of information.
Although it’s not new to me, I tend to forget about it sometimes and I’m sure I’m not the only one.
Thanks for that one Roger!
Wolf Chiappella says:
Hi Tim, this talk was very inspiring for me and I felt especially touched by the emotional end. I will keep up my trial & error lifestyle, from now on without regrets. Thank you!
19th of July, 2011rich says:
Really great talk tim!
19th of July, 2011Abhijit says:
Thanks for the talk, Tim. An interesting thought indeed. Enjoyed it. Was reminded of my Bayesian Inference classes in Statistics course.
20th of July, 2011Joe Perez says:
Fantastic talk, Tim. I’ve transcribed a bit on my blog and begin to explore how these insights can help to inform our relationship to language.
20th of July, 2011Don Smith says:
Good job on the talk. Nicely done, I enjoyed it. Many new thoughts for me. Looking forward to exploring other things you have to say. Cheers.
22nd of July, 2011Dirk says:
Brilliant talk. Thank you, Tim!
24th of July, 2011Subir says:
Very interesting talk, and I agree with Tim on his insights. One question did come to mind which I’ve not been able to figure out –
By saying that certain people tend to have the God Complex, is he not showing a God Complex as well? Or by saying that trial and error works, is that not a God Complex?
Never say never?
Maybe I’m overthinking it..
26th of July, 2011