User count is evenly distributed with the number of errors.

PlzNavigate is unrelated.

There's a good chance this is caused by issue 734978.

(blink-worker triage)

Error rate is still gradually increasing:
https://uma.googleplex.com/timeline_v2?sid=177660ed5db44a49c6c2f22bfcf9f56e

AFAICS, ERROR_START_WORKER_FAILED and ERROR_TIMEOUT are also a bit increasing.

1. The rate of SERVICE_WORKER_ERROR_START_WORKER_FAILED is increased by BrowserSideNavigation.
BrowserSideNavigation enabled: https://uma.googleplex.com/timeline_v2?sid=51a2a26d51304dfee5ff48a68b186659
BrowserSideNavigation disabled: https://uma.googleplex.com/timeline_v2?sid=5277d80e252965a37faccc8693889c8e
I think this is because Chrome tends to create a new process for SW when BrowserSideNavigation (PlzNavigate) is enabled.

2. The rate of SERVICE_WORKER_ERROR_FAILED is increasing even in the same version
https://uma.googleplex.com/timeline_v2?sid=f0147e66c03dd621ba313d73f1ad7c13
I don't understand why yet...

It looks like something strange happened externally between June 10-12. The number of page loads doubled temporarily:
https://uma.googleplex.com/timeline_v2?sid=c70e9b1c87026a4cbcd3885317348618

This just happens to line up with the Beta 60 release.

I found a strange spike of ServiceWorker.MainFramePageLoad.RedirectChainLength during June 10-14.
https://uma.googleplex.com/timeline_v2?sid=d978bde5561eb865b17d5b1d04cb2065

It looks like that a site introduced a redirection related bug with Service Worker during that period.

Handwavy conclusion is that both 59 and 60 dropped at the same time, so it was probably an external site change, not a Chrome regression:
https://uma.googleplex.com/timeline_v2?sid=f706168347531179444af4261dd641b4