You must be talking about altcoin mining since there is virtually no choice when it comes to mining software with modern asic hardware.

Looks like the number of nodes is back to the baseline of ~800. Both the second showstopper bug that was exploited and their embarrassingly bad attempt to cover it up by alleging core nodes were affected, and trying to fake the number of core nodes in existence look like less than exist have not perturbed them.
https://www.reddit.com/r/Bitcoin/comments/5zhmwn/andrew_stones_bu_dev_fake_screenshot_is_a_poor/
The fact that the number of BU nodes is unchanged in light of the above speaks volumes for the type of people running them...

This is one of the most fucked up public stunt screw ups yet.

Some of the nodes have come back online, presumably patched, but it's still under 400 on coin dance. I wonder how much time is needed to know how many have patched their clients and restarted them or have abandoned BU after this. It will probably be many days before we know since not all nodes will be attended to or monitored daily.

Looks like about 200 nodes have stayed online which are probably mostly patched clients, though it could be very persistent users repeatedly restarting their node (and least less likely sybil attacks.)

Well the pool hasn't solved any blocks of late so just for some interest here's poolbench which is showing block speed changes to a client in USA (slow to load):
https://poolbench.antminer.link/
Note solo pool being the fastest of the full verifying node pools, and that de is actually faster on change than main solo since the server upgrade. All the ones above it mine empty unverified blocks.

Now all we need is a few blocks sprinkled around to make the most of that rapid block propagation/change...

That doesn't make sense for the former reason; the latter (integration with other codebases forked off earlier) does. There are many improvements in 0.13 and 0.14 that are totally unrelated to segwit and if segwit never gets activated the segwit code goes untouched, so forking off 0.12 is just leaving us with ancient code. The problem is everyone's rushing to get on top of each other with new proposals and choosing the shortest cuts possible. Oh well stick around long enough and *someone* will probably create a BIP100 for current core...

For core version 0.12... why? Presumably to predate all segwit code and be compatible with BU? Since it's a hard fork proposal independent of segwit there is nothing that would prevent this from being made for core 0.14. BIP100 was the most popular choice amongst mining pools when it was first proposed. Don't be surprised if it makes a surprising comeback now. If this was coded up for core instead of BU without any other changes it would be very popular very quickly...

32MB was the limit of the message size between nodes. There was in fact no limit at all to block size, but the message size wouldn't cope with anything larger than 32MB. Satoshi was the one who imposed a 1MB limit later on to minimise DoS risk. I suggest you do some reading in the more general bitcoin/ and development and technical/ discussion sections of the forum to understand why a massive increase all alone like that would be a potentially unmitigated disaster for the network at this time for multiple reasons. Even an increase to 2MB alone is massively different to 1MB; it doesn't just carry with it 'double the work, storage and bandwidth'. Some things do not scale linearly at all to size unfortunately due to the original protocol design of bitcoin transaction verification. The original design was not perfect (nothing ever is) and only with the benefit of hindsight can we see the flawed components in the design and the code which was a monolithic code dump from Satoshi. However it's here to stay and we need to work with it. Assuming we need to scale, everything must be factored into the scaling approach.

By the way, talk of BIP100 is back too, which was by far the most popular pool scaling choice at one stage, though currently they're trying to tie it in with BU:
https://bitcointalk.org/index.php?topic=1822824.0

There are also substantial performance improvements in core 0.13 and 0.14 that haven't made their way into the BU code so miners would lose all those benefits by abandoning the core client.

Bah, he wasn't ambitious enough; he should have been aiming for 1 bit.

The number of sigspammers who posted absolutely nothing useful rehashing exactly the same pseudopost in this thread gives me a headache.

Oh look, only the 7th thread created about this 

Thanks I wasn't aware of that. Probably something worth offering in conjunction with BIP102 then.

Right, there is no *public* code that I'm aware of, and I do hack on bitcoind for my own purposes, especially the mining components so I'm quite familiar with the code. As for "up until the point in time that it is not", well that's the direction *someone* should take with their code if they wish to not pursue other fixes for sigop scaling issues as a matter of priority then - if they wish to address the main reason core is against an instant block size increase. Also note that header first mining, which most Chinese pools do (AKA SPV/spy mining), and as proposed for BU, will have no idea what is in a block and can never choose the one with less sigops.

You keep talking about miners mining this more quickly validating block... there is no code currently that can try to validate two different blocks concurrently and pick the one that validates faster. The first one that comes in will be under validation while any other blocks come in wait before they can be validated so unless someone has a rewrite that does what you claim, the problem still exists. First block that hits will always win.

Transactions yes, but in prune mode the 'network' flag will be disabled meaning you won't be advertising that other nodes can download older block information from you to sync up.

From memory the closest we've come to that to date was that single transaction 1MB block from f2pool that took nodes up to 25 seconds to validate. It is possible to make it much worse, but newer versions of bitcoind (and probably faster node CPUs) would have brought that down. Rusty at the time estimated it could still take up to 11 seconds with 1MB:

https://rusty.ozlabs.org/?p=522

So yeah it has been used... possibly unwittingly at the time.

Core would not because they're all convinced we must have segwit before increasing the block size to prevent a quadratic scaling sigop DDoS happening... though segwit doesn't change the sigops included in regular transactions, it only makes segwit transactions scale linearly which is why the blocksize increase proposal is still not on the hard roadmap for core as is. If block generation is biased against heavy sigop transactions in the core code (this does not need a consensus change, soft fork or hard fork) then pool operators would have to consciously try and include heavy sigop transactions intentionally in order to create a DDoS type block - would they do that? Always assume that if a malicious vector exists then someone will try and exploit it, though it would be very costly for a pool to risk slow block generation/orphanage by doing so.