I had the same experience with Apple when the iTunes team were invalidating ProRes files created from Final Cut Pro.
On the other hand, it's a good thing to not give a free pass for bad behavior in products you produce.
Such as: buying GitHub is unlikely to be Microsoft's attempt to trap OSS; Apple's early adoption of new ports probably isn't motivated by selling an extra $12 dongle for every $2000 Macbook; a journalist criticising Facebook is rather unlikely to be acting on corporate orders to sully the competition, etc.
All the arguments for "but your user might already have such-and-such-version-of-jQuery cached" just have not had a good enough impact on total experience. There are so many versions of whatever you might be importing, and HTTP keep-alive works really, really well.
Of course, I also don't have any tracking setup, say nothing of the 50 dozen systems most sites use. I've never gotten anything out of them that I couldn't have figured out on my own. They always come back to "users bounce when shit is slow". Thanks, tracking scripts, for contributing to my bounce-rate.
So there you go, do you want to create a good user experience, or do you want to sell your users out for a pittance?
Have your first user experience in your site be super optimized. Static HTML. Host all assets on your own. Size images correctly. Minimize, then inline the CSS and JS. There are plenty of design-time scripts to automate it. And if you're making SPAs, you can implement all the dynamic loading you want past the initial page because users clicking to open your app are already engaged. But get that first page on-screen in less than a second on a 2G network.
Lighthouse definitely has some improvements to make for the audit suggestions that offer advice to improve the score. I think people get a little bit frustrated/blinded by the advice when there's nothing that can be done about third parties (I know I do!).
Our product Calibre (mentioned in the bottom of the post) has a n adblocking/third party blocking feature that allows you to compare your pages with or without third party scripts gunking up performance metrics (here's the release post: https://calibreapp.com/blog/release-notes-apr-2019/).
We've seen really good success in separating the two because it sends a clear signal of what a site WOULD score without ads. This is good because it makes it clear to PMs, management and decision makers that tracking tools incur a really visible cost.
In this case it's probably more the performance team wanting to do unbiased measurement. As they should.
How does https://calibreapp.com/ compare to https://www.webpagetest.org/ and https://www.sitespeed.io/ and https://gtmetrix.com/ ?
Why would someone want to pay you for your service, as opposed to using any of the above? Do you support browsers other than Chrome? Do you support a wide variety of simulated hardware? Do you support a wide variety of testing locations? Do you support any page speed test profile other than Lighthouse and Google Page Speed? Do you support running your software in a Docker container of my own choice?
Because I can get all those things via one or more of the competitors in this space, and I'm not seeing any compelling reason to even look at Calibre.
Please feel free to convince me that I'm wrong.
One of the key deciding factors is do you want to run the service yourself e.g. VMs, containers, in multiple regions or do you want to make that someone else's problem and get on with running your business - most of my clients are in the later camp
Both Calibre and SpeedCurve have clean UI's that allow performance to be tracked over time, from multiple locations and integration with other services via web hooks, APIs etc.
GTMetrix is old and outdated IMV, - it still heavily relies on YSlow rules which are well past their sell-by-date
WebPageTest is my favourite (SpeedCurve uses it under the hood) but it's got no ability to track over time, SiteSpee.io is really nice too but is self-hosted.
With the exception of some real devices in Dulles, all of the products rely on emulation for mobile, and Safari is a real gap for them all too
I knew PageSpeed was bunk when that recommendation came up on a web page I was working with. All of the "next gen" formats that PageSpeed recommended weren't supported by any mainline browsers — not even Chrome.
That may have changed by now, but my opinion of PageSpeed hasn't.
What's one supposed to do when PageSpeed only points to Google's own asset delivery services as being "the" big problems which cause a lower score?
- "serve images in next gen formats" - complains about images served from tpc.googlesyndication.com
- "efficiently encode images" - same as above
- "reduce javascript execution time" - cdn.ampproject.org; securepubads.g.doubleclick.net; www.googletagservices.com
.. ?
I can easily get rid of those issues, "simply" by not showing any ads. The site will then be really fast, and... I'll get the ability to display ads again?!
Shame the ad-related revenue will be 0, which isn't unfortunately a good thing for the site.
Why is Google themselves not doing things "right" in order to ensure a speedy execution?
A while back, one of the reasons for a low score was that the GPT.js code (required to show ads) had a short expiration time. Dammit. How am I supposed to fix that, Google?