Guys, doesn't it look for you like we have some issues with invalid shares?

That's how pool looks right now:
KanoPool: 22,098.71THs
Shares: 703,336,606,279 (51.55%)
Invalids:
25.4M (21.5%) · 104.5G (12.9%)
Last Block
Pool: 40h 56m (495143)
Network: 2m 27s (495434)
Odds are it is old hardware. For example, Canaan has removed a majority of their hardware, but they probably leave old S7 or stuff running which would have a lot of invalids. Same goes for Cobra etcetera. That is my guess.
I just sent a message to nicehash thread. Would it have anything to do with the insanely high diff they have on our machines. S9 65536 diff. Lowest Vardiff will drop. Same goes with s3 etc. I have about a 4% loss just in rejects due to high diff also my local work completed showing on s3 since they remove it on s9 is almost double that of accepted shares. Could this be a problem. A form of inadvertent block withholding. Nicehash being slow to push shares is causing stales etc. just curious