I don't think home miners have any real impact on the hashrate anymore, there are just so few of them relatively speaking. 4.3ph (which switched to some shiny new coin) is roughly 1.29% of the total network, where as the complaint up above is about much wilder swings in hashrate. The swing OP is pointing to is an extremely small sample size which shows nothing. Finding 6 blocks in 16 minutes is not a rare occurrence, albeit it isn't common either, but in and of itself no conclusion can be drawn. The graph I posted was to show small sample sizes, as in the bright green line, will produce wild swings but don't have enough information to indicate anything. The price of bitcoin *might* be manipulated, but the difficulty is dictated by the free market.
People who can afford to mine profitably will mine, that's all there is to it. A small portion may go mine something else, and *maybe* a low single digit percent of mining power is either switched off or under clocked, but the 5% up or down you see every two weeks is large data centres being filled up, emptied, rotating their stock and occasionally burning to the ground, not miners switching pools or turning on and off. The big guys mine 24/7 and there is no shutting off. There is simply too much money to be made and too much to be lost by being idled or gambling on alt coins. People can't afford to pay rent on a warehouse that's fill with turned off miners. And if they own the warehouse, they're kicking themselves wishing they'd done something more profitable with the money.
The graphs at
bitcoin.sipa.be look really flat, especially by bitcoin standards. I'd say this suggests economics (ie. how much you make vs your costs) is dictating most mining. With the shitty price (and given the investment a lot of people made it is an awful price) very few miners are adding power and some miners simply can't pay the bills at a certain point, let alone justify expansion. There are a lot of miners paying 12 cents/kw or more to mine, to make only 2 or 3 cents from the same kwh after paying that electricity. Where 95% of the network is conducted on an industrial scale, very large bills are due monthly. There are a lot of facilities who can't break even below the current $70/TH/month, which is $105 made per KW of use at 0.66W/GH. (Although, obviously most hosts are above water since we're still at 96% of the hashrate we were at two weeks ago.) For older gear, which still makes up a lot of the network, 1W/GH yields only $70/KW used. That's below a lot of hosting prices. We're scraping those break even points, and that puts pressure on difficulty. More efficient machines means a non stop rise in difficulty over time, but that rise is becoming extremely slow.
I think the bigger story is the burn out rate versus replacement rate, both for miners and hosting. Right now, we're facing a significant drop off in old machines starting to die and several hosts shutting down. Half the network is at least 7 month old equipment which is going to start seeing failures in machines. It is a far different market than late 2013 and early 2014. At this price, low single digit growth per month is all you can expect, including the current pace of efficiency gains. However, this means the market is in a position where it will be much slower to react to any significant price rise. (Just like oil, natural gas and minerals, low prices mean tighter supply.) If the price spikes to $500, there will be an acute shortage of miners and hosting for months. Even at $500, doubling the hashrate would take 6 months or more.