Judging because of extremely rare 2/10 second rounds? People are upset when the luck is bad, but people are also upset when the luck is good... :-) I don't remember how such "good luck" happen last time, it is really rare.
no, no, I'm not upset as I completely understand the logic behind it and as I said in the long run I earn exactly what I would expect. Having said that, I do realise, that as difficulty increases even gh/s miner will mine less & then even less than less.
btw if the round lasted really short and there would be only one share submitted in the whole pool in that round and if that share was mine, then I would earn 49.xxx BTC for that round
and would send a bottle of very good wine to slush.
my point is: 1gh/s rig will mine about 838 diff1 shares/h = 0.23... shares/s ~= 1 in 4 chance rig will submit a share in that 1 s. if round lasts 2 seconds then we would have approx. 50% chance that rig will submit a share in that time. since network latency is about 20-50 ms depending on your location (20ms in internet terms is very good indeed) and can jump to 80ms, then your chances of submitting a share are much lower because of round trip delay (which is 2x latency) and processing time ( cgminer, or, network card, router with nat, adsl etc, which I cannot measure but would estimate to be another 20ms ) then my 2 second round allows for 1.8s of hash finding or even less. this is where this is painful.
i understand your argument that I could be lucky and submit 3 shares in those 2 seconds by pure luck, but the latency bias still remains.
on the other hand Jeff running his asic miner on diff1 (he should not do it, but there is no way to stop him) with 68 gh/s, is submitting 15 shares / s so he has and extra edge in this situation. above the obvious gh/s advantage.
dialling back share difficulty to 1/2 or 1/4 is probably also not such a good idea. delaying payoff till more than one round is completed who's combined duration is bigger than some threshold is also probably not such a good idea....