XF 2.2 Pagespeed Insights is making no sense

[avalanch]

How does it figure that I have a TTFB of 12 seconds on desktop and 1.3 on slow 4G mobile? Makes absolutely no sense at all when at Slow 4G throttling it's fully loaded in 2 seconds. And by god they really did gimp the F out of the 4g throttling...

Network throttling: 150 ms TCP RTT, 1,638.4 kb/s throughput (Simulated)
Browser location: North America

So for desktop, it is "fully loaded" in 1 second flat (on their test) but yet supposedly has a TTFB of 12 seconds? How is that possible? Pingdom, cloudflare testing, webpagetest & gtmetrix all disagree with googles "12 second TTFB". I don't understand this at all, do you?? Like how TF is desktop testing going to be worse than a mobile result.... HOW?? Someone, please try to explain this weird methodology that google has.. I'm all ears.

PageSpeed Insights

pagespeed.web.dev

 

[Jeremy P]

The data up top comes from CrUX rather than that specific Lighthouse run (which had a TTFB of 330ms). It also notes that there is insufficient data for that specific URL so it's aggregated over the origin (domain). It's possible you just have some really unlucky samples in the dataset. I'd imagine it would improve over time.

 

[avalanch]

I think what did it is I had the notices set to pop in after 10 seconds but I've since disabled that, did some changes to the notices (moved some of it into threads for example), leaving me with 1 notice left with 0 delay until it shows (fixed notice, so that delay option isn't even a option for it anymore so it loads instantly).

And even if that was the trigger for it, I don't see how that could affect the TTFB, google search console shows 0 speed issues at all, no matter what time span I ask it to show..