Hacker News Reader

Comment by rovr138 | Hacker News Reader

Wondered how well it would rank on those insight scores,

https://developers.google.com/speed/pagespeed/insights/?url=...

Looks like 80kb and they still find things.

It's pretty darn hard to hit their page delivery targets unless you're serving from something close to their testing site. https doesn't help, because it adds round trips (www.spacejam.com doesn't support tls 1.3, and I wouldn't want pagespeed to be using a resume handshake for initial load anyway).

An ultra small, static site can sometimes get 100 though: https://developers.google.com/speed/pagespeed/insights/?url=...

> It's pretty darn hard to hit their page delivery targets unless you're serving from something close to their testing site.

Are you sure? Have you tried using a CDN?

> An ultra small, static site can sometimes get 100 though

Here's an example site I run that gets a close to perfect score that has a fairly complex landing page:

https://developers.google.com/speed/pagespeed/insights/?url=...

(it should load close to instant once it connects in your browser: https://www.checkbot.io/)

The main tips I can give for a high page speed that most websites don't do are avoid large header images, make sure text is visible before custom fonts load, use minimal CSS (and/or inline the CSS for the header into the top of the HTML), don't use blocking JavaScript and especially avoid huge JavaScript triggered cookie popups (the blocking JavaScript + big delay for the Largest Contentful Paint will kill your score).

Google can't even meet their own metrics. The Gmail landing page gets a PageSpeed score of 24.

https://developers.google.com/speed/pagespeed/insights/?url=...

98 on first try, 97 on refresh. 99 on 2nd refresh. Are they just making stuff up?

If they were making stuff up, you'd think at least their own Google.com would score higher than a 76.

That’s an extremely small variation in score. Why would that make you think they’re making stuff up? Networks and servers don’t always respond with identical timings.

Yes, 98, 97, 99 are small.

But when I punched in Google.com, I got 84, and then 78, then 90. That's a pretty wide range.

There's a bit about the variability here: https://github.com/GoogleChrome/lighthouse/blob/master/docs/...

Reminds me of that quote from Ryan from the Office:

I'm such a perfectionist that I'd kinda rather not do it at all, than do a crappy version.

Seems that Google's software shares that mentality.

Reply on news.ycombinator.com