To put that image at a size that would still be too big for the importance and accuracy of the content:
There is no magic 100 is "Looking good" but 101 is "Too far"
One problem with that image is that it is telling p2poolers to centralise p2pool.
Right ...
The idea of p2pool is to run it yourself so you can indeed make the claim that p2pool is decentralised.
The image is telling people to centralise p2pool and mine on other nodes.
Now if you were silly enough to mine on someone else's random who knows what set up p2pool node, then their node configuration is just as likely to make even a 500ms ping time almost totally irrelevant.
Most of them are, to be blunt, poorly configured and slow to handle work changes.
--
So now, when discussing pools with high performance hardware and software (and pools that don't have that ...)
The performance of solo.ckpool.org (and kano.is closely behind it
) means that the fast pool work change time will make even a sizeable ping time worth your while mining.
The ping time
vaguely infers your stale share rate, so of course a higher ping time will mean higher stales.
But also remember that a block time average is expected to be around 600s (or less at the moment) so even a 600ms work reply time would put your stale shares caused by that at around ... 0.1% ...
The problem isn't so much your ping time as it is the mining hardware being able to switch work quickly.
So yes, ping time is quite relevant if it takes you 2000ms to get a packet from the Antarctic via penguin satellite to Tasmania then weave it's way through the crappy internet of Australia and make it to a pool somewhere else on the planet, but in general the pool's work change time will usually be way more important to determining if you will find a block ... which is all that matters here on solo ...