However, going from 28nm to 14nm (what Intel is using in many of their chips now) is good, but not the same order of magnitude as 110nm to 28nm. Once you see new mining rigs all running on the current process at a high clock rate, then I believe that you will end up with more opportunities for "hobby miners" since they will be able to project an ROI much easier and it will be much more likely that they will be able to have a ROI. People will be much more inclined to purchase a miner if they know it won't be replaced by something 100 times better in 3 months.
This will be a somewhat more stable situation in terms of hash rate increases and miners coming to market. It will be closer to how CPU mining was in 2010 and GPU mining was in, 2011-2012 (even the first third of 2013) where you could buy a few video cards and have a pretty good idea that they would return X BTC/week (decreasing slightly each week due to added GPUs on the network) for a year or two until a new video card came out and that might cause a slight increase in the difficulty increases. And even then you might mine some fewer BTC/week but not 1/1000th of the number in that period like the difference between January 2013 and November 2014.
It is difficult to model and so it is not surprising that they may not be catching all the nuances there. I certainly wouldn't be confident in being able to predict it all going forward. Particularly with so many variables - regulations, halving in ~18 months, improvements (e.g. sidechains etc), new uses.
...
Yup. Additionally, large miners are likely to, eventually, vertically integrate other services. There are some potential negatives (as well as positives) to that, but it also probably means that there are a number of workable business models for a number of large fairly-efficient miners, as opposed to an endgame where strictly the least-cost producer gobbles all hashpower.
Regardless of all that, the same arguments that are thrown around about mining could be thrown around about data-centers in general (OMG! there will only be a single least-cost/massive-scale producer that will run everything!), or pretty much *any* industry for that matter; this is really just a (mostly) extrapolated-short-sight criticism of capitalism in general, not specifically bitcoin mining. Few industries actually coalesce to pure monopoly in practice.
Anyway, the natural incentives of bitcoin mining make it less likely than other domains.
Yup. Additionally, large miners are likely to, eventually, vertically integrate other services. There are some potential negatives (as well as positives) to that, but it also probably means that there are a number of workable business models for a number of large fairly-efficient miners, as opposed to an endgame where strictly the least-cost producer gobbles all hashpower.
Regardless of all that, the same arguments that are thrown around about mining could be thrown around about data-centers in general (OMG! there will only be a single least-cost/massive-scale producer that will run everything!), or pretty much *any* industry for that matter; this is really just a (mostly) extrapolated-short-sight criticism of capitalism in general, not specifically bitcoin mining. Few industries actually coalesce to pure monopoly in practice.
Anyway, the natural incentives of bitcoin mining make it less likely than other domains.
