Coding HighDEAs

Anyone on /prog/ ever smoke marijuana and then start coming up with really great ideas for your code, or just computer science ideas in general?

For example, I got an idea that perhaps a way to implement an A.I. is with a chain of randomly-generated finite state machines, linked together so that one's output becomes another's input, etc. The structure of this chain could be determined by some sort of "DNA" with "genes". Then you could use a genenic algorithm to find better and better genes until the resulting chain of finite state machines exhibits macro behavior in the form of A.I.

they were actually all terrible ideas, OP.
you were just high so it seemed good at the time.

>>2
Isn't that true of most ideas, at least ideas of the significant magnitude, regardless of whether formed under the influence of a psychoactive substance or not? I've worked on an idea, flushing out the details, until I eventually find some reason why the idea is bad/impossible/too hard/etc., and then I give up.

This idea is horrible.

NO EXCEPTIONS

>>1
computationally infeasible

No random number is ever truly random.
Enjoy your wasted clock cycles in pursuit of perfection in an area that doesn't need it.

>>6
Unless you want to play the deterministic universe card, then how about using some outside source of noise i.e. a thermocouple, hooking it up to an A-to-D converter as your source of randomness?

>>7
I believe time would be the most efficient way(in both effort and time) to get a "random" number.

>>7
Almost all noise isn't random (especially heat).

Even randomness in quantum physics is hard to use, because it is often weighted to one outcome.

>>5

I started thinking that even when I was high. I suppose the general idea is to use artificial selection to produce an artificial intelligence, just like natural selection produced natural intelligence. Why try to manually make something all-at-once when we already know of an incremental process that has successfully made what we want?

>>6

Genetic algorithms aren't supposed to find perfection, just good approximations of perfection. I suppose it's hard to define a "perfect A.I.", but it's easy to define the characteristics we'd like it to have, and we could select from a set of A.I. candidates those which most follow these characteristics.

And true randomness shouldn't be required, as current GAs function properly without it.

>>1
Why has no one pointed out that OP's idea is merely another FSM, complexity wise? Seriously, go check out some evolutionary programming if you want something that's actually, I don't know, Turing complete, maybe?

Also realize that we are a long way from achieving anything of any importance with GA's, evolutionary programming, neural networks, other such tomfoolery. They all fail in their own way.

>>11
Because >>1's "idea" isn't anything more than words strung together.  It doesn't do anything.

>>1
Anal Touring thought of that, right before he ate the magic apple.

YHBTBMAWHBTBY

>>12
I was thinking that there would be two FSMs, one whose input is received from an outside source (keyboard), and another FSM at the end of the chain whose output is sent to another outside source (display).

>>11
Good point. Maybe it'd be better to use a neural network design, and have the "genes" encode for the structure and behavior of this network at the neuron level. Still, I agree that this is a very complex idea, both conceptually and computationally, and it won't work in all likeliness. I was just using it as an illustration of a coding/compsci "HighDEA".

>>13
YHBTBMAWHBTBY
You Have Been Trolled By My Anus. We Have Been Trolled By You.
?

>>15
oMG EPIC!!!

>>13,15
Same person.

>>1 There were two great contributions to modern society that came out of Berkeley in the 60's, one was UNIX (BSD specifically) and the other was LSD.

This is not a coincidence.

UNIX shell scripting and Perl actually make sense when you are stoned. It may be the only state of conciousness where they do make perfect sense.

>>18
Taking acid is not being stoned.

>>18,19
i THOUGHT i'D NEVER GET TO POST THIS IN A RELEVANT POST BUT FINALLY i WAS GIVEN THE CHANCE!

'uSE OF lsd-25 IN COMPUTER PROGRAMMING'
http://www.maps.org/news-letters/v18n1/v18n1-MAPS_24.pdf

>>20
http://www.wired.com/science/discoveries/news/2006/01/70015

>>21
eYE DON'T READ WIRED BCUZ ITS PURE BULLSHIT, BUT i'VE READ MY PROBLEM CIHLD (PARTIALLY) AND MYSELRF i'VE GOT SOME AST.

wHY DO YOU TYPE like this ALL THE TIME?

>>23
Because it's poring to type like this all the time.

>>24
hIHIHEHEHEH

This post brought to you by the Gay Nigger Association of America

Don't change these.

Name:		Email:
Entire Thread Thread List