Thriving Hive #2

I don’t think my last post was forceful enough.

It is completely absurd that for 130 million, so much computing power has been put on the task.

For comparison, BP alone has announced plans to spend $100 million on a super computer searching for oil – with a sliver of the computational ability.

This is a hectic example of how a well designed incentive system can trump the combined power of states, in a sphere where you’d expect them to be totally dominant.

Thanks to the hype around bitcoin – and the kicks and coin people get out of participating in the enterprise – the infrastructure developed almost overnight to create most powerful computer the world has ever seen, for a fraction of what it would cost the most powerful governments in the world. Indeed, for a fraction of what they spend on far inferior systems.

And it feels like it appeared out of thin air overnight.

Last year supercomputer sales were $5.6 billion. If these (admittedly not quite apple) comparisons to the bitcoin processing hive are to be believed, the every-ten-minute-bitcoin-releases completely trumps them.

And the bitcoins were conjured out of the aether at no cost by the system itself. Disorientating.

Perhaps politicians, economists, entrepreneurs – actually maybe all of us – should give some thought to how such an outcome has been so freely engineered with such a phenomenal result.

Thriving Hive

While hipcoin fever has finally dropped to sub-hysterical levels, there have already been some interesting lines of though to come out of the mania. One is the realisation that all money should probably be designed with a similar level of security features embedded.

Another being the stunning acknowledgement of the sheer power of the the bitcoin mining hive.

As a quick recap, you can harness the spare processing power of your computer to contribute to the brute force ‘mining’ of bitcoins.

Dizzying

It was recently noted that the combined processing power of Bitcoin miners outperformed the top 500 supercomputers in the world combined, with 1000 petaFLOPS. By comparison, IBM’s Sequioa –  the fastest commercial supercomputer – operates at 16 petaFLOPS.(1)

What a shame that such a force is applied to pointless, arbitrarily difficult calculations.

The existence of hardware optimised from the bottom up makes the comparison not quite apples and oranges, but save to say this is as much computing power that has ever been thrown at a single problem.

A thought.

Now bitcoins are offered at a rate of 25 every 10 minutes. Using $100 price (let’s keep the calculation simple and our assumptions conservative) the combined value is around $131 million.

But remember supercomputing itself is a saleable good.

Pull out your envelope

So let’s put a rental price of $1000 per hour, which (while hard to pin down) seems to be the going rate. If the bitcoin mining hive is operating at 100x this, that would put an hourly value of $100,000 for the network, or $876 million / year. This may sound a little silly, an obviously would crash the market somewhat, but the supercomputer industry measures its revenues in billion, so it’s feasible that a network large enough to effectively replace it would be worth the same order of magnitude.

You can see where I’m going with this: even if the assumptions above were weakened drastically, this still implies a substantially higher monetary gain for orientating your spunky little butterfly lab product away from bitcoin hashing and towards finding a cure for cancer, simulating the human brain or probing the depths of the Big Bang. The added benefit being that it’s done for the good of us all, rather than helping facilitate the trade of crack, smack and weaponry.(2)

Precedent

There is something of a precedent for this, as there are projects that allow you to contribute your computer for free. I know this first hand, as about a year ago I donated my mostly idle computer hardware to the solution of a protein folding problem.

Unfortunately the system was poorly designed and took up massive amounts of bandwidth, smashing through my limit overnight and condemning my household to a week of dial-up speed.(3) This was actually a design flaw, and there’s no need to require enormous amounts of data transfers to harness a remote processor.

So, in a soundbite, it should be possible to replicate the money-for-processing trade at the heart of the Bitcoin endeavour.

How could you do it?

Without giving too much away, you would need a sales function, a nerdy portal for users of computers to access your network (I’m unqualified to comment, but it appears this is the real value-add from the true players in the supercomputing business), and a system of financial rewards to pay back any investment a punter makes in hardware. You could even pay in bitcoins.

Hey, whoever set this up could probably even do it for a profit.

(1) As a side not this is about 10% of the value of Bitcoins, so those hoping to profit from their machines are relying on fairly substantial growth (reasonable – numbers on the drug industry are hard to find, but I’d bet that its very nature makes growth).

(2) Those protesting at the fuzzy flawed logic, a better way would be to compare the billions in revenue for the supercomputer industry to the paltry 9 figure sum offered by mining pools.

(3) Sorry Pete