why is the growth in the number of bitcoins not convex?

and I check the trend for 'All Time', it looks like the rate of growth in the number of bitcoins INCREASED around January 2010. I thought that the rate of growth (enforced by a computer algorithm) would halve every 4 years and that it would never increase... Am I missing something?

The rate of bitcoin production is set so that on average, regardless of how powerful the network is, a block should be discovered every ~10 minutes. If the network becomes more powerful due to new entrants to mining or better technology, the network will adjust (after a short lag time) by increasing the difficulty - and the same process would work in reverse if people left mining or the hash power dropped for some other reason.

The fact that the # of bitcoins per block halves sequentially doesn't really apply as much as the difficulty targeting which would still happen if bitcoins per block never halved again.

Because of this, we see a linear growth in number of bitcoins mined while network hashing power increases geometrically. All this means is that your 5GH/ps Jalapeno is already obsolete and by the time you get your 600GH/ps Monarch delivered in April or May 2014 it will already be nearly obsolete.

:)

• but doesn't this contradict the previous answer, then? if the rate of growth of bitcoin production is the same regardless of the network power, than the rate of production between jan 2009 and jan 2010 was the one dictated by the algorithm. Why is it, then, that the speed of production INCREASED in jan 2010, when the algorithm should only allow for DECREASES in the speed every 4 years? – user114618 Dec 10 '13 at 18:10
• when there are very few miners (or not much computational power) in the network, the changes to difficulty which be mucj more volatile of course. If say there are 3 miners each contributing 1/3 of the network and one drops out there will be a large impact. Now if there are 1 million miners and 1 drops out nothing will happen. For a graph showing that the growth of bitcoins in existence has been fairly linear since inception please refer to this: blockchain.info/charts/… – user10263 Dec 10 '13 at 22:56
• this is the same link as the one I posted initially and it does show an acceleration in jan 2010 – user114618 Dec 11 '13 at 9:54

It is decreasing by half every 4 years, just that graph is just not showing it well for this short timeframe.

The dip up until 2010 was due to there only being one or two miners in the entire network, so the maximum number of Bitcoin wasn't produced in that period. The speed of the entire network couldn't keep up with the minimum of 6 blocks an hour which has lowered the total available Bitcoin in that period, though the number overall is just fine. You can see the sharp decrease in late 2012 when the block reward halving first occurred, which is the start of the downward trend in block reward production.

Currently the network hashrate is increasing so quickly that it's producing more Bitcoin than expected, we'll probably reach the next halving in a little less than the intended 4 year cycle. That's not a negative effect in itself though, it just means the network is increasingly secure month on month. There's a similar hump between June and July of 2011, when GPU mining was publicly introduced.

• so you're saying that between jan 2009 and jan 2010 the actual growth rate was actually BELOW the algorithm-imposed limit? And the reason why it picked up is that Bitcoin caught on and people actually started mining it, making the algorithm-imposed growth limit binding? Would this understanding of mine be correct? – user114618 Dec 10 '13 at 10:27
• That's all correct. The hash rate was below a level that could sustain difficulty 1, the minimum for Bitcoin mining. – Anonymous Dec 10 '13 at 10:40