h a l f b a k e r y
My hatstand runneth over
idea:
add, search, annotate, link, view, overview, recent, by name, best, random
meta:
news, help, about, links, report a problem
account:
Browse anonymously,
or get an account
and write.
Login
Create account.
|
|
|
Utilize Evolutionary Computing to evolve new ideas from ideas on the HB.
Computer would select 2 random ideas then combine them to evolve a new idea. The ideas would be scored and then the better ideas would be mated to evolve a new set of ideas. Eventually really great ideas, really bad ideas or
just really strange ideas could be evolved.
Darwin's Halfbakery
Darwin_27s_20HalfBakery
[bungston, May 05 2006]
New ideas via concatenation
New_20ideas_20via_20concatenation
This one is actually close enough to make this redundant. [bungston, May 05 2006]
Like this?
Flaming_20Rooster_20Tail_20Skis
Shameless, perhaps, but an example of using existing halfbaked technology when creating newer halfbaked technology. [Shz, May 05 2006]
[link]
|
| |
Shirley, we need to find a way to stop bad ideas from multiplying... |
|
| |
This made me think about what would consitute the genome of an idea in order to even attempt crossover. Not for very long, mind, but just long enough to realise that it would be extremely difficult. |
|
| |
At the very least you'd need to look for 'species' of idea, such as 'object with particular feature', 'means of achieving x' and so on, and limit crossover to within species. That might provide some interesting ideas. However I doubt that a full taxonomy could be determined for all ideas, and in the end, as Bungston notes in considerably fewer words, I suspect that what you'd end up with is very similar to the Concatenation idea. |
|
| |
[jhomrighaus], nice idea, but until you formulate and explain some method of codifying an idea in such a way that it can be combined with another in order to create something new, this idea isn't really doing anything for me. |
|
| |
Once that part is figured out:
What would be the mechanism for selection?
What would be the mechanism for survival?
What would be the (or would there be a) mechanism for mutation?
|
|
| |
//think about what would consitute the genome of an idea in order to even attempt crossover// |
|
| |
//formulate and explain some method of codifying an idea in such a way that it can be combined with another in order to create something new// |
|
| |
are essentially saying the same thing which reflects yet another problem to overcome. |
|
| |
zen_tom, jhomrighaus did say that selection would be based on 'scoring' (presumably voting), I'm guessing that survival would be as for any other idea - if it's not mfd-ed it survives. |
|
| |
Mutation could happen if the species suggestion above was followed: say we had a flying car and a jumping house. We could crossover to produce a jumping car or a flying house. If it were possible to boil all ideas down into such simple terms then presumably other objects and aspects could be added to the potential lists that don't actually appear in exisitng ideas; e.g. add 'cloud ' to the object list and 'pontificating' to the aspect list despite the fact that neither currently appear. Mutations could then occur to provide pontificating car, jumping cloud and so on. |
|
| |
Of course object-with-aspect is just one species of idea, and there's no way of ensuring that a full, consistent taxomony could be discovered. |
|
| |
I do suspect that a partial taxonomy could be done along these lines, but probably with a huge amount of effort, a fair dose of dispute about the taxomony adopted and probably very few ideas worth reading. |
|
| |
A species like object-with-aspect would likely be too simplistic as it fails to address factors such as 'why do this?'; it's just intended as a hypothetical example. |
|
| |
In a way 'bakers already perform the crossover and selection aspects of this idea when we see something that already exists, wonder what would happen if we added a particular twist, post it and see if it survives and prospers or crashes and burns. |
|
| |
I'm new to this forum, but even though most ideas I've posted are original (thought up then checked for repetition), a couple came from thinking of a greatly modified version of something else I've seen here (i.e. more fun or interesting). The intuitive leap of idea generation via the human brain can't be beat. |
|
| |
That said, this reminds me of something I've been thinking about for a fun and creative community process...(see upcoming post ;) |
|
| |
I was thinking less along the lines of just mixing up the words as in the to links and more along the lines of mixing the content of the ideas themselves as was described in the annotations. I have not gone too much further in my description as I was curious to see what kind of input was offered on the core concept. [boysparks] hit the nail on the head in his description I think the difference here is that this would have the potential to evolve completely new ideas in areas and ways that noone had ever really considered. |
|
| |
A computer scientist by the name of John Koza actually has applied the technique of Genetic Programming to developing new inventions. He has successfully received at least one patent (US 6847851) using the technique. |
|
| |
I read the same article and that is one of the inspirations for this idea. |
|
| |
// then combine them to evolve a new idea.// |
|
| |
your talking about AI. this idea was well worded, but you dont mention how to make it work. shouldn't that be a requirement? you're basically saying "lets make some AI that can solve our problems" and thats all your saying. a better idea would be to put 1000 monkeys in a room with 1000 typewriters for a year. |
|
| |
You have any idea what kind of mess that would make of the typewriters? Not to mention having to clean up all the dessicated monkey corpses. I think that would be a much worse idea than just saying "lets make some AI that can solve our problems". |
|
| |
Just trying to keep things in perspective... |
|
| |
Actually this is quite simple, Take two ideas and mix them together(keep some parts of each and drop some parts) Then evaluate the new idea. if its good it moves to next round. if its bad it gets droped. then take all the "good" ideas and start the process over. This is a very simple process. I admit to lack of the intimate details of the current evolutionary computing technology that is sited above but it is not imaginary, it exisits and is being used, it has even generated a product that has won a Patent. |
|
| |
The problem with that approach is the problem of evaluating goodness, which a few hits of the random button will show is not a simple problem. |
|
| |
I have thought about this approach using essays or poems. The check against goodness would be the AI built into Word or something similar: these AIs can check grammar and style. One could even allow the AI to correct the concatenation as it saw fit. The progeny would then go back for another round of concatenation. |
|
| |
Ultimately, the survival fitness of these creations would be determined by posting them to one of those services that sells college/high school essays, seeing which one was used most often. |
|
| |
Actually I was intending for the good/bad judgement to be made by the bakers of the Half Bakery. Only the best will survive. Allow them to "live" for a limited time then move to the next round. More buns means survivial. This will not however be a super fast process. |
|
| |
in what way would the computer recieve the original ideas? are they strings of text or are they mathematical interperetations?? |
|
| |
We are way overthinking this. You could do this with a scissors and paper if you wanted. The idea is the "concept" of evolving the ideas, not the process of evolving ideas. you could do it a million different ways, the thing that I was proposing was to do it then use the resulting ideas to generate new ideas. The existing approaches to this did not continue the process forward. |
|
| |