https://www.technologyreview.com/2012/04/12/186779/computer-scientists-build-computer-using-swarms-of-crabs/

…Now Yukio-Pegio Gunji from Kobe University in Japan and a couple of pals have built what is essentially billiard ball computer using soldier crabs. “We demonstrate that swarms of soldier crabs can implement logical gates when placed in a geometrically constrained environment,” they say.

These creatures seem to be uniquely suited for this form of information processing . They live under the sand in tidal lagoons and emerge at low tide in swarms of hundreds of thousands.

What’s interesting about the crabs is that they appear to demonstrate two distinct forms of behaviour. When in the middle of a swarm, they simply follow whoever is nearby. But when they find themselves on on the edge of a swarm, they change…

ChrispenEvan said:

https://www.technologyreview.com/2012/04/12/186779/computer-scientists-build-computer-using-swarms-of-crabs/

…Now Yukio-Pegio Gunji from Kobe University in Japan and a couple of pals have built what is essentially billiard ball computer using soldier crabs. “We demonstrate that swarms of soldier crabs can implement logical gates when placed in a geometrically constrained environment,” they say.

These creatures seem to be uniquely suited for this form of information processing . They live under the sand in tidal lagoons and emerge at low tide in swarms of hundreds of thousands.

What’s interesting about the crabs is that they appear to demonstrate two distinct forms of behaviour. When in the middle of a swarm, they simply follow whoever is nearby. But when they find themselves on on the edge of a swarm, they change…

you forgot the ridiculous leading bit though

This is motivated by two lines of thought. The first is theoretical–ordinary computers are hugely energy inefficient–some eight orders of magnitude worse than is theoretically possible. The second is practical–Nature has evolved many much more efficient forms of computation for specific tasks such as pattern recognition.

motivated by efficiency right

https://arxiv.org/abs/1204.1749

looks more like Texas sharpshooting to get the outcome they want to be honest

SCIENCE said:

ChrispenEvan said:

https://www.technologyreview.com/2012/04/12/186779/computer-scientists-build-computer-using-swarms-of-crabs/

…Now Yukio-Pegio Gunji from Kobe University in Japan and a couple of pals have built what is essentially billiard ball computer using soldier crabs. “We demonstrate that swarms of soldier crabs can implement logical gates when placed in a geometrically constrained environment,” they say.

These creatures seem to be uniquely suited for this form of information processing . They live under the sand in tidal lagoons and emerge at low tide in swarms of hundreds of thousands.

What’s interesting about the crabs is that they appear to demonstrate two distinct forms of behaviour. When in the middle of a swarm, they simply follow whoever is nearby. But when they find themselves on on the edge of a swarm, they change…

you forgot the ridiculous leading bit though

This is motivated by two lines of thought. The first is theoretical–ordinary computers are hugely energy inefficient–some eight orders of magnitude worse than is theoretically possible. The second is practical–Nature has evolved many much more efficient forms of computation for specific tasks such as pattern recognition.

motivated by efficiency right

https://arxiv.org/abs/1204.1749

looks more like Texas sharpshooting to get the outcome they want to be honest

QI, but yes I doubt efficient.

OTOH, it has often struck me that if you want to analyse the stability of a pile of sand at the granular level using computers it would take weeks of computing time, but the pile of sand does it in microseconds, so maybe they have a point.

The Rev Dodgson said:

SCIENCE said:

ChrispenEvan said:

https://www.technologyreview.com/2012/04/12/186779/computer-scientists-build-computer-using-swarms-of-crabs/

…Now Yukio-Pegio Gunji from Kobe University in Japan and a couple of pals have built what is essentially billiard ball computer using soldier crabs. “We demonstrate that swarms of soldier crabs can implement logical gates when placed in a geometrically constrained environment,” they say.

These creatures seem to be uniquely suited for this form of information processing . They live under the sand in tidal lagoons and emerge at low tide in swarms of hundreds of thousands.

What’s interesting about the crabs is that they appear to demonstrate two distinct forms of behaviour. When in the middle of a swarm, they simply follow whoever is nearby. But when they find themselves on on the edge of a swarm, they change…

you forgot the ridiculous leading bit though

This is motivated by two lines of thought. The first is theoretical–ordinary computers are hugely energy inefficient–some eight orders of magnitude worse than is theoretically possible. The second is practical–Nature has evolved many much more efficient forms of computation for specific tasks such as pattern recognition.

motivated by efficiency right

https://arxiv.org/abs/1204.1749

looks more like Texas sharpshooting to get the outcome they want to be honest

QI, but yes I doubt efficient.

OTOH, it has often struck me that if you want to analyse the stability of a pile of sand at the granular level using computers it would take weeks of computing time, but the pile of sand does it in microseconds, so maybe they have a point.

ah but what if you’re using a pile of smelted sand* to model the behaviour of a pile of natural sand that is acting as a set of logical gates à la a modern-vNA-computer

*: in modern-vNA-computer form

The Rev Dodgson said:

OTOH, it has often struck me that if you want to analyse the stability of a pile of sand at the granular level using computers it would take weeks of computing time, but the pile of sand does it in microseconds, so maybe they have a point.

Maybe not as slow as that – When you have time take a minute or three to look at the videos in this channel. He finds some very impressive stuff.

Two Minute Papers

Spiny Norman said:

The Rev Dodgson said:

OTOH, it has often struck me that if you want to analyse the stability of a pile of sand at the granular level using computers it would take weeks of computing time, but the pile of sand does it in microseconds, so maybe they have a point.

Maybe not as slow as that – When you have time take a minute or three to look at the videos in this channel. He finds some very impressive stuff.

Two Minute Papers

Thanks SN. Looks like my strange idea of fun :)

Spiny Norman said:

The Rev Dodgson said:

OTOH, it has often struck me that if you want to analyse the stability of a pile of sand at the granular level using computers it would take weeks of computing time, but the pile of sand does it in microseconds, so maybe they have a point.

Maybe not as slow as that – When you have time take a minute or three to look at the videos in this channel. He finds some very impressive stuff.

Two Minute Papers

VI, but note that these are very simple problems, and took millions of iterations for the AIs to find any clever new solutions.

From the article:

“To be fair, the results were mixed. While Gunji and co found they could build a decent OR gate using soldier crabs, their AND-gate was much less reliable.”

This is just another way of saying it doesn’t work, IMO.

Timing is key to AND gate operation, the two groups of crabs would need to consistently and reliably meet at the intersection of their respective input tunnels.

For an OR gate to work in this way, all you need to do is join the two input tunnels into one output tunnel. It can’t not work, if you know what I mean.

No processing or logical operations are going to occur without reliable AND gates. Technically, I’m not even sure you can build a working OR gate without using an AND gate, so I’m not even sure I’d call their OR gate a success.

> This is motivated by two lines of thought. The first is theoretical–ordinary computers are hugely energy inefficient–some eight orders of magnitude worse than is theoretically possible. The second is practical–Nature has evolved many much more efficient forms of computation for specific tasks such as pattern recognition.

Eight orders! Prove it.

Often it’s not nature that is more efficient, it’s that bad algorithms are horrendous.

As an example of a horrendously bad algorithithm, there’s the travelling salesman algorithm. Say 1,000 points
Solving it requires 10^307 steps.
But that’s for an exact solution.

Nature doesn’t do exact solutions. I could sit down and in a few minutes come up with an approximate solution.

A more rigorous approximate solution is the “ant-colony” algorithm. https://en.wikipedia.org/wiki/Ant_colony_optimization_algorithms
As each ant visits each nore it leaves a pheromone trail. The solution is the best pheromone trail. But there are other approximate solutions not inspired by animals.