This... it would recalculate in 2 weeks utter max.
Even if the hash rate dropped 95% and transactions took hours in 2 weeks utter max it would be back to 10 minutes..
I suppose you could shut your stuff down kill the hash rate, let the difficulty drop then turn it back on and mine a crazy amount of blocks. That is a different topic though.
You are wrong. If the hash rate halved, the blocks would be discovered around every 20 minutes. And the retargeting would happen after 2016 blocks or 28 days.
50400 coins would be mined in 28 days instead of 14, so there would be 25200 less coins mined in 28 days.
Apart the need to wait around 20 minutes to get a block, the network would work without problems as the mean size of a block would rise from 400KB today to 800KB in this case (supposing the same number of transactions happen). This would be dangerously near to the limit of the block size, but increasing it would be pretty easy and fast, as the current limit to 1 MB is artificial.
If the hash rate would fall 95% it would be a problem for a long time (like 40 weeks), but the same 50400 coins would be mined in 280 days instead of 14. Inflation would halt near completely and the price would be pushed up for lack of new bitcoins hitting the market. In the same 40 weeks, normally, there would be little more than 1M bitcoins mined, but instead just 50400 would be. Inflation would drop from 9.49%/year today (weekly projection) to 0.4745%/year. The main problem would be a block would be mined every 200 minutes and there would not be room enough for all transactions. Surely the commission to get a transaction in the block would raise a lot (no microtransactions or free transactions baby). But 40 weeks are a lot of time and I do not think Gavin Andresen would need more than 4 weeks to get out with a fix increasing the the block size directly to 32 MB (the current maximum with all limits removed) and this would be enough to make room to all transactions. And with 32 MB blocks the only nuisance would be the 3 hours and 20 minutes (very variable) needed to find a block. Bitcoin confirmations would not be slower than Credit Card confirmations even in this case (even if the block is found after one and half day).
BTW, a 95% reduction of the hash rate would cause the remaining miners to get 20x revenues than before and they would be a lot more profitable.
In my opinion, if this happened, bitcoin would be able to adapt and, as result, a lot of people would invest in it given its resilience.