Monday, December 9, 2013

Defeat

I'm out of runway. My freelancing stint never really took off the way it was supposed to. It's my failure: I simply didn't pursue work hard enough. So it's taken two and a half years to eat through my savings buffer that I was willing to sacrifice for this experiment, and now I'm at the end of the runway and the lift isn't there. Own damn fault.

Oh well, it was like a nice long holiday. I got a little certificate out of the inaugural 6.002x course that is now part of edX. I finally earned my amateur radio licence. Worked a lot on my gEDA fork. Wasted a bunch of time playing the mother of all Freeciv games with an eye to animating the replay of a minimap-like view. Spent a good bit of time with my niece over the 2011/2012 summer. Learned a lot about bitcoin, and refreshed my knowledge about cryptography in general in the process. Figured out how to take credit card payments for merchants through Payfast.

But it was too much fun and not enough work-work. Dull, boring, and frustrating work that yet pays the bills. The straw that breaks this camel's back is when I started noticing that Rocketseed (my ex employer and my anchor client over the duration of this freelancing experiment) started taking their time paying me. When Zanap was still in charge of accounts, I'd get my money within a few days of sending her the invoice. But the new regime was to pay me on the very last day of the month (or even missing that) even when my work for them was done, committed, and pushed by the first few days of the month.

I don't have the energy anymore, in this iteration of the experiment, to try and encourage more prompt payments. Besides, getting my invoices paid quicker only puts off the problem for a short while: my living expenses are slightly higher than what I'm earning. Unless a miracle happens and tomorrow my inbox is overflowing with people asking me to conjure up some C for them, I need to do the realistic thing and take the low-risk option now instead of doubling down on my bet like a gambler.

So I'm now on the job market. Let someone else figure out how to turn value into money, and just give me a regular payslip. What I've been doing hasn't been working, so it's time to try something else. At least for a while, so I can build a new runway and try again later.

Friday, December 6, 2013

Is bitcoin an ecological disaster in the making?

I've been thinking a lot about bitcoin mining lately, even considering building some mining equipment with Nasier. As exciting as it is, I have a new doubt over the enviro-ethical value of the bitcoin protocol as it exists today.

Right now, the total energy put into mining bitcoin is still small. The network hash rate is perhaps 7PH/s. If we take an average mining efficiency of 100MH/J (maybe generous, but I really only want an order-of-magnitude calculation here) then it means there must be 70MW dedicated to mining. It won't stay at just 70MW though.

The current block reward is 25BTC, or about 250000ZAR. Every 10 minutes in the long-term average.  That's the limit up to which mining costs can go before it becomes uneconomical - the point where miners decide to just switch off their machines. Right now capital costs are important, because average mining equipment efficiency is still increasing, which obsoletes current mining equipment in a matter of months. I don't think we can expect consistent increases in mining efficiency for much longer: Avalon's gen2 chips are made with a 55nm process, and I've seen talk of 28nm chips. Once we reach an efficiency wall, the only way to access a greater hash rate will be to consume more power.

Once we reach that point, there will be little incentive to replace older equipment with newer; the incentive for profitable miners will be simply to add more equipment to their operations. Capital costs then shrink as the economic lifetime of mining equipment increases to years instead of months, and electricity prices will become the dominant cost.

Domestic electricity costs no more than R2.50/kWh. 250kZAR per 10 minutes can fund the consumption of 100MWh every 10 minutes - that's 167kWh every second, or roughly half my monthly electricity consumption every second. In standard units, that's 600MW. Notice that this doesn't depend on average mining efficiency, but only on the block reward and on the price of electricity. Cheaper electricity only makes it worse!

Are you okay with that? I'm not sure if I am. Does the standard banking industry, whose death due to bitcoin we sometimes pine for, use that much? I doubt bitcoin would replace the banking industry's carbon footprint; it would rather add to it. Also, things get much, much worse if bitcoin rises to the $100000+ that some think is possible.

Block reward halves every 210000 blocks, or every 4 years. Next halving is in 2016. So we may see only two years of gigawatt-scale mining - if the price of electricity in bitcoin stays constant. What happens to the half of the exahash-scale mining network that would go offline due to being uncompetitive? Does it instantly turn into a 51% attack network, having to make ends meet by conspiring with double-spenders?

In the very long term, the block reward becomes insignificant or disappears completely. So I'm happy that bitcoin mining won't be an ecological travesty in eternity. But in the meantime, this looks like a serious problem to me.