Bacteria make computers look like pocket calculators | Hacker News Reader

Bacteria make computers look like pocket calculators

guardian.co.uk

30 pointsonreact-com21 comments

My computer can solve TSP on three towns far, far faster. This shows calculations can be done with bacteria, but there's no evidence at all that it will be faster, especially on NPC, even with the extreme parallelism the method implies.

It's typical journalistic hype.

indeed, I don't doubt that once we can do parallelism on par with the biological world in neural network models (which are probably already available or being researched) many of these studies will be thought of as ill considered funding streams. Not so sure about chemical processors - I still expect the pay off from this research to be fantastically beneficial for all sorts of domains.

Remember that it's Bactomputer Alpha 0.00001 ;-)

When this discipline was first starting to kick off I remember a little experiment done to tap into and code the DNA of a particular strain of bacterium (probably similar to this E-coli). In doing so the playful scientists encoded the phrase "I am the riddle of life, solve me!" into junk DNA sections of the organism.... it always made me feel a little queasy thinking about it being found by whatever civilization is around a thousand years from now! What would it do to their fundamental picture of life? I'm sure the same could be said about organic computation. It's an open-ended chaotic system, where really all we control is the initial starting conditions. Who's to know that after we've got our "answer" the continued adaptation leads to something vastly beyond our control/comprehension. I guess it's all part of the "ghost in the machine" paranoia surrounding AI, except these things will not be ghosts and the machine is DNA.....

</alarmist diatribe>

To really get your head spinning, what if you used 'strings' on your own dna and you found some interesting ascii segments in there, on 4 billion basepairs the chances of finding something interesting are pretty good ;)

But remember, we see faces in clouds. We seem predisposed to bullshit our selves in similar fashion.

That's what he was saying..

That probably has much more to do with a specialized part of the bran that provides an intensive face processing and retrieval mechanism than it does with some fully generalized predisposition.

Remember that book, The Bible Code? And remember how it worked on any book, such as Moby Dick? Someone should plug the published human genome(s) into it.

see 'a beautiful mind'.

We are pattern recognizers, feed us random noise and we'll see patterns.

DNA is not exactly random noise, but for this purpose it is probably close enough.

Similar "logic" has inspired a number of scientists to attempt to "decode" the background microwave radiation of the universe as it has the capacity to hold the first act of Romeo and Juliet (dunno how many bits of information that comes too) ... but the idea was that IF a creator wanted to leave a message that could be found by any significantly technologically advanced organism, this would be the best place to shove it

But like another poster said - we are pattern recognition machines. Coincidences, faces, conspiracies and their ilk just leap to the forefront of our attention.... I wonder if their couldn't be a little bit of this phenomena at work in this study? Not read enough of the primary material to see if it's something they've consciously tried to factor out of the experiments

I wasn't thinking of Romeo and Juliet but the cosmic bacground noise has it in there, if only I remembered the offset...

Finding "killroy was here" in your own DNA would surely make you wonder though :)

In a thousand years, with no selective pressure to preserve it, the phrase would mutate beyond recognition and probably be deleted altogether.

I always wondered if we could could find something in nature that shows P=NP. It could be bacteria, autistic savants, dna computing, etc. Has anyone set up a test like this?

I'm thinking you simply increase the number and inputs on an NP problem and watch at what rate the time taken by one of these methods increases. (Is this too naive of an approach?)

Here is something that scientists have studied which does this http://www.tjhsst.edu/~rlatimer/techlab06/Students/OuyangPap...

They actually take the approach that you state. They also have to judge how the system solved the problem. If you don't reach an optimal state you could be only approximating P=NP.

That's an interesting paper.

I wonder why we can't make a "soap surface" computer? Just translate your NP-Complete program into a Steiner tree problem, make the appropriate glass plates and pegs model, get the results and translate back?

I wish I knew more computer science. I'd love to try it.

You could do that exactly. The problem is that the soap surface computer doesn't always perform the task to find the best steiner tree etc...

Why is that? Is it approximating then, or just unreliable?

Unreliable it doesn't necessarily give the best steiner tree or a good answer. Sort of hazy on this mainly just googled this information.

That's a shame. I wonder if we could find other natural processes like this? I wonder if electricity seeming to instantly know the path of least resistance could be wired up to solve traveling salesman somehow?

Protein folding?

Reply on news.ycombinator.com