Forget Quantum computer - This can come earlier

[JackH]

An even faster approach to computing is going to be done through synthetic biology. A rapidly growing field, taken "public" by "Dr Craig Venter" that created the first artificial lifeform only a few years ago. (Google Synthia).

Fast forward today, we are looking at logical gates being assembled by the use of synthetic biology. For great reference look here: http://www.kurzweilai.net/scientists-create-computing-building-blocks-from-bacteria-and-dna or here: http://www3.imperial.ac.uk/portal/page/portallive/AF95133B97CC7745E040C69B47390348

Question is how soon we can expect a synthetic biological ASIC device and how much such can outperform silicon based products. Looking at the synthetic biology and how it develops this could hit us anytime out of the blue.

I havent been able to really figuere out what team in what country is close to getting a wet-ware computer up and running. If you know more about this please do feel free to post. This could be the next mayor development in computing, and what better way to test speed is there than mining Bitcoins.

For cool synthetic biology stuff you will enjoy this: http://www.youtube.com/watch?v=AKxmqMH4w_A

[Jutarul]

It's unlikely that molecular computation will ever be used as a means to compute the block-chain, because the "problem" changes every 10 minutes, requiring to reprogram the genetic material. Input and Output to and from a molecular computer is non-trivial.

It's more likely that molecular computation is a means to brute-force the private keys of bitcoin addresses. Especially those addresses which are considered dead and have huge balances.

[tytus]

People publish nonsense even in high profile journals.

[Bugpowder]

An even faster approach to computing is going to be done through synthetic biology. A rapidly growing field, taken "public" by "Dr Craig Venter" that created the first artificial lifeform only a few years ago. (Google Synthia).

Fast forward today, we are looking at logical gates being assembled by the use of synthetic biology. For great reference look here: http://www.kurzweilai.net/scientists-create-computing-building-blocks-from-bacteria-and-dna or here: http://www3.imperial.ac.uk/portal/page/portallive/AF95133B97CC7745E040C69B47390348

Question is how soon we can expect a synthetic biological ASIC device and how much such can outperform silicon based products. Looking at the synthetic biology and how it develops this could hit us anytime out of the blue.

I havent been able to really figuere out what team in what country is close to getting a wet-ware computer up and running. If you know more about this please do feel free to post. This could be the next mayor development in computing, and what better way to test speed is there than mining Bitcoins.

For cool synthetic biology stuff you will enjoy this: http://www.youtube.com/watch?v=AKxmqMH4w_A

No team is "close". No team is even "far". It's not going to happen. "wet-ware" computers will always tremendously suck except for very specific pattern recognition functions using DNA oligonucleotides.  Erik Winfree at CalTech is the closest to having a generalized computational device, but he is a long way away from anything practical.  Silicon computers will always be at least 100,000,000 times faster. I am an expert in a related field.
  

[salty]

People publish nonsense even in high profile journals.

And a reasonable proportion of that nonsense will become tomorrow's technology.

[molecular]

Going off on a tangent:

If you think about it, evolutionary algorithms are often a good solution for search and optimization problems. Mining is a search problem.

However the nature of sha256 makes it impossible to construct a good fitness function.. you just don't know wether you're close to a solution or not.

So maybe what you could do is utilize massive parallelism by implementing some bacteria that is able to somehow evaluate two rounds of sha256 and somehow mark itself when difficulty is met. Put in it's dna the necessary input data (the block hash, right?) and a random nonce. Then make it reproduce under mutation of the nonce and select any marked individual and read its nonce from its dna. Voila, mesdames et messieurs: block found biologically

[tytus]

Yes, a small portion ... this is basic research ... but to publish something in Nature (derivatives) You must claim that You discovered America. Only a small percentage of nonsense get's through the referees but this is still much bigger than the number of discovered Americas :-)

[tytus]

@molecular ... reading of DNA is definitely slower than 10 min :-) You will have a solution that was good a day ago :-)

[Bugpowder]

Biological computation has broad appeal to the layperson, engineer, comp sci person, hence it's common publication in Science, Nature. In practice, synthetic biology's promise is not there, but rather in metabolite engineering, biological production of difficult to synthesize molecules, biofuels, etc.

Also, at Nature about 60% of submissions get through referees, but < 10% of submissions get sent out for review.

[salty]

Yes, a small portion ... this is basic research ... but to publish something in Nature (derivatives) You must claim that You discovered America. Only a small percentage of nonsense get's through the referees but this is still much bigger than the number of discovered Americas :-)

Yes, that's a problem with science and the media, one is concerned with finding the best theory, and the other is concerned with finding the most readers through hype and exaggeration

[Bugpowder]

Yes, a small portion ... this is basic research ... but to publish something in Nature (derivatives) You must claim that You discovered America. Only a small percentage of nonsense get's through the referees but this is still much bigger than the number of discovered Americas :-)

Yes, that's a problem with science and the media, one is concerned with finding the best theory, and the other is concerned with finding the most readers through hype and exaggeration

Which is which?

[molecular]

@molecular ... reading of DNA is definitely slower than 10 min :-) You will have a solution that was good a day ago :-)

That is probably not the only problem with the approach .

I know this is completely unfeasable, but for the fun of it: I only need to read a very small portion of the dna. Find primer (did I use that word correctly?), read data. Still 10 minutes?

[tytus]

reading of 1 nucleotide takes 1 hour on the fastest machines. You will need to read more than one nucleotide :-)

[JackH]

My reason for posting is that I see very fast development in the world of synthetic biology and a great number of companies shooting up around this field.

Its not possible to say what will happen in terms of computing power. I for one will keep an eye out on whomever comes out with a wet-ware computer. First model may be "crap" but at the same time, this industry has 25+ years to catch up with silicon.

[labestiol]

I'm not an expert in synthetic biology, but I work in system biology and I've been to some conferences with talks about synthetic biology.
I concur in saying that the first application will probably be oil productions, and that we are extremely far from biological computation.

And people don't have the biggest esteem for Craig Venter. He seems better at making big claims than deliver

[Bugpowder]

@molecular ... reading of DNA is definitely slower than 10 min :-) You will have a solution that was good a day ago :-)

That is probably not the only problem with the approach .

I know this is completely unfeasable, but for the fun of it: I only need to read a very small portion of the dna. Find primer (did I use that word right?), read data. Still 10 minutes?

The advantage of DNA is not speed, but parallelization.  See, Helicos's or Illuminia's sequencing systems. They do say 5Gbp / day in short fragments, but do that by sequencing millions at once, not by high speed.

[Bugpowder]

Academics have issues with Venter's attitude, not his results.  He knows how to spot talented people and ideas and get the funding to make them happen. Celera's shotgun sequencing really pushed the Human Genome Project forward and basically beat the academic team (though assembly would not have been possible without the high quality scaffold already generated by the project).  The sequencing of the sargasso sea was also a great and  fascinating idea and yielded red-Channelrhodopsins from Volvox and probably much more I don't know about.

I tried to get the guy who invented shotgun sequencing to buy some bitcoins. He was all about amassing gold and shorting the Euro though. His loss. :-P

[labestiol]

Ah ah, not the worst strategy either imo. And I'm pretty sure he'll come to bitcoin at some point.
Anyway, thank for the complementary opinion on Venter

[molecular]

reading of 1 nucleotide takes 1 hour on the fastest machines. You will need to read more than one nucleotide :-)

damnit, ok then make the bacterium that finds a solution not mark itself, but instead produce some huge molecule we can actually see using a microscope and encode the nonce into that somehow.

[Bugpowder]

reading of 1 nucleotide takes 1 hour on the fastest machines. You will need to read more than one nucleotide :-)

damnit, ok then make the bacterium that finds a solution not mark itself, but instead produce some huge molecule we can actually see using a microscope and encode the nonce into that somehow.

I think the fastest approaches are quite a bit faster than 1 nucleotide an hour.

http://www.helicosbio.com/Portals/0/Documents/Helicos%20tSMS%20Technology%20Primer.pdf

DNA oligos are pretty good at picking out single sequences from a big mess of others.  I'm not that familiar with the computation we are trying to do here, if someone could explain the search space we would need to look in, we could have a discussion on the feasibility of using DNA to find the needle in the haystack.

For example, if you are searching for a known sequence that can be encoded in 25 base pairs (50 bits) and either is, or is not present in a soup of random sequences bound to a glass surface, IN THEORY you could fluorescently tag a complementary oligo, wash it over the surface, wash unbound material off, and check for the presence of a fluorescent dot somewhere on the surface.  IN THEORY.  In practice, noise, dust, etc would make that approach difficult, but there may be work arounds.