I just tried this. It has a lot of overlap with YSlow, but there are some interesting extra rules in it, most notably a rule that finds inefficient CSS selectors. That's kind of cool.
These tools (like YSlow) are oh so interesting when your clients install them in their browser and examine the site you built for them. Now, I'll get to hear complaints that we used too many inefficient CSS selectors (and that we're incompetent), ignoring the fact that a .0003 millisecond makes no difference to the client's website. It's not Google's fault, but I just had to sigh when I saw some of the rules (I'll expect a call by tomorrow haha). :-)
I could actually see it working in the other direction. Things like YSlow highlight good technique in a (seemingly) objective way.
As for your issue, I would hope that if a client is savvy enough to check with these tools, they could also understand when you tell them what is important and what is not.
I find that my smartest clients have the capability of being the most annoying but usually just demand more explanation out of me. And while sometimes they take that to the extreme (where it's smarter to limit your interaction) its usually for the best because in teaching I find I learn a lot more.
The best thing I liked about PageSpeed as against YSlow is that it provides very specific action items. Very novice website owners can also follow the instructions.
A good addition would be the Printable Version of the report the way YSlow provides.
On Page Speed Activity tab, it seemed to be all messed up. It did the activity profiling for all pages open currently in the browser and not only for the specific page open in the current tab. Bug ? or did I miss some settings.
Was I the only person who was a little surprised that they didn't mention YSlow by name? That strikes me as somewhat rude, though I know Yahoo and Google aren't exactly best buddies.
Edit: Looks like some of the commenters there agree with me on that point.
One annoying thing with this thing is that it creates a bunch of dirs in your home dir and dumps stuff in them when performing the analysis. Shouldn't they put things like that in /tmp, if they need to store them at all?
Seems to have some overlap as mentioned in other comments - buy why the need to compare the two? Why not just add both to your toolbox and gain the features of both?
It looks really useful. Much more than Yslow, and Yslow is great too.
I posted an example http://dodeja.posterous.com/762803
It even gives images that are optimized and JS that is minified. Developers <3 Google
I also liked the specific action items. Like which exact css selector is inefficient. Also liked it called out the exact images that could do with compression, what % can be gained and even a sample compressed image!
Slow for Google's crawlers does not necessarily mean slow for you or vice versa. It depends on location, whether logged in an not logged in users see the same site, the time of day, the functionality of the site etc.
They could adjust for the factors picked up by this plugin, but their importance is going to cary from site to site.
Also, it could easily be gamed.
Hopefully slow sites should annoy people and get fewer links.
I think they do already have a bias towards faster websites, but as with everything else with google it's only speculation as things constantly change with these guys.
I'm not sure Google needs to cook that book. People like faster websites, so they'll get more link love, so they'll probably rise up of their own accord.
If I'm searching for something to link to - often a news story, if I'm writing for the datablog we run - then I'm going to open five stories in tabs and probably pick whichever one finishes loading fastest.
Yes, I agree - I really value a fast-loading website, and am very annoyed if the site I clicked on takes forever to load.
Your theory that it could happen naturally is interesting, but I don't see it working in practise - ie, I don't notice any correllation between pagerank and site response time. I am sure Google et al have the resources to measure it in a reasonably non-game-able fashion; I'd certainly welcome such a development.
