This seems clearly false.

Let's say the page with the best information about dogs took an infinite amount of time to load (never loaded). I don't think you'd want that page to be ranked highly.

What about if that page took a year to load?

What about if that page took an hour?

A minute?

What's your cutoff? Let's say, generously, that you're willing to wait 30 minutes for the page to load. Why such a sharp cutoff? Why do pages that take 30 minutes and 1 second to load get penalized, but pages that take 30 minutes are treated as fine? The user experience is basically just as acceptable (by your standards) / just as bad (by my standards).

Perhaps maybe instead of a sharp cutoff we need some sort of sliding scale, where pages that are only slightly slower than optimal are penalized only slightly, and pages that are much slower than optimal are penalized more.

Gosh, that sounds familiar.

There's a cutoff built into browsers in the form of a KeepAliveTimeout. A longer load time could have a valid reason, that's not for search engines to decide.

Also your example of a year is obtuse. It's like saying you should drive BMW because you will get killed in a Tesla if you crash at a million mph.

What Google's page speed factor does is differentiate between 1.5 and 2.5 seconds loading time. That has nothing to do with the quality of a page.

Your assumption is that the time factor can't be handled by the user. Users try somewhere else when a page doesn't load, and in my opinion that is fine. Granted, it's my opinion. Perhaps most users don't agree and here we are.

If it is truly the best information, it is what I want and I will wait. If I trust that Google can give me the best information then I would be willing to wait. I give up easily now because I can't trust Google to give me quality results. I will not wait for an ad bloated news aggregator.

When a user does a search, clicks on a result, that result does not have what they want so they hit back, and then they click on another result that is a failure of the search engine. "Users can decide the site is worthless after they click" is an opportunity for somebody to come up with a better system.

I guess I simply don't agree. The search engine should give me the best information, period. It does not, and one of the reasons it does not is because it is busy prioritizing things that aren't relevant to content.

That example doesn’t reflect the real world differences in page load time, however. IMO the load time arbitration doesn’t belong in the “search” engine. If your search engine fetches 10 fast loading pages but lacking the depth of information you need it loses utility.

The tricky thing is also the fact that Google does not inform the user that it is prioritising the faster loading pages instead of the pages with the most accurate match. This is a little dishonest too.

It's good if it plays some role when the content quality is similar between results. Some pages are so slow or obnoxious that I'd almost rather not get the information at all, than dreadfully waiting for ads and pop-up-modals to load. Perhaps ranking is the wrong tool, but it would be nice if I could know that a site is bloated before clicking the link. AMP lets me know that.

I agree. And if the speed metric is well-designed, and if making not-slow pages is easy enough (and it really is) then the Nash equilibrium is "Everyone gets the most relevant page and it loads quickly."

I’d argue that pages that are static with few images should be given preference. Simple CSS rules, no JavaScript (or very little of it), not a ton of images, etc. People from all walks of life are capable of publishing this content and it’s equally as accessible.

Pages that should be heavily penalized are those with lots of extraneous JS, bloated CSS files, lots of auto play videos, and huge image files that are supposedly created by “experts” that need a DevOps team to deploy and distribute all the bloat.

What if it's a photographers portfolio? An art collection site?

Sometimes images make sense.

But nobody waits longer than 8 seconds. That's the golden webdev rule for ages. Anything above deserved to be punished.

Art collections are easy to develop with icons and progressive images.

Unless you go there for the artist and you know what you're after.

Progressive images... eh. The js you load for it is 3-4x the images at 720p. Progressive jpeg, that's better.

Once upon a time Matt Cutts used the example of 'a cure for cancer' marked up badly with poor HTML structure but providing the true result versus a well marked up site with the wrong info. Which should rank better?

The same IMO can be applied to speed as a factor. A slow site, (due to depth of content?) badly marked up but providing the correct result should outrank the fast site providing wrong info.

> This only helps the heavy handed SEO optimized sites with deep pockets and time to kill get yet another edge.

This sounds plausible, but doesn't translate to reality. Look at the slowest content sites today: they're large, well-resourced news sites. Look at the small, independent individuals running their own personal site: many are very fast.

In reality, I'd say the biggest causes of website performance issues are (a) institutional bureaucracy - managers requiring their website to do something without regard for the overhead and (b) professional graphic designers defining UI features without technical knowledge of their implemention details. These are both, usually, corporate inefficiencies that have little effect on smaller maintainers.

What also seems to be missing in Google's metric is that of Page Size.

That is, by also having less bytes in the page, then less bandwidth is used. Since less bandwidth is used, less energy is used.

Unfortunately, Google doesn't measure, nor seem to care about energy consumption, as GoogleBot only cares about response times. (Of course, there are energy concerns regarding that too...but it's only every about speed with google and punishes everyone else for not evangelising it)

Using Page Size as metric has important benefits that Page Speed deliberately undermines. It also resurrects the notion that using external cached css and js is a good thing.

Did you know that in order to get 10 urls (the thing that you and I actually want) from google, you need to download 500K? It seems ludicrous to me. Each search result is chock-full of css and js that doesn't change.

Thankfully HN is old-school and uses external files and good-ol' minimalism.

2c

Aren't page size and page speed (and energy consumption) correlated? If you request a 1mb page and a 1kb page from the same server, the 1kb one will load faster in basically all scenarios

If you can test speed close to the requester geographically, it's possibly an even better indicator of energy usage because it also implies the data doesn't have to travel as far.

Also, Google has no idea what external resources your browser has cached locally.

> Also, Google has no idea what external resources your browser has cached locally.

I'm pretty sure they have enough data to figure that out.

I absolutely agree with you. Speed shouldn't play any role in ranking. However in regards to relevancy RankBrain is going to eventually train itself to push the most relevant/respected results. Will have to wait and watch how it spans out.

> The article with the best insight isn't likely to be the one with the perfecly optimized website.

That isn't what search is ranking. It's ranking what users want. And users very most definitely want, on average, among other things, fast.

Did you ever try to look up something over satellite internet? AMP is a tremendous help, not because all other pages are slow but because you know you'll be able to load the page in <3s.

Funnily enough I spent a year or two in rural Australia on satellite Internet and I definitely see your point. I think it's okay for a site to offer an AMP capable site, I just don't think it should factor into the ranking. Perhaps there could be a Prefer AMP pages flag? Or the inverse, ignore Page Speed. I will concede that the majority of users probably get by with page speed as a factor and that my concerns are not those of the general user base.

As an aside, gaming on satellite internet was a real drag, the only game I had that would handle the full seconds of latency was Star Wars Galaxies. Thankfully that was already a favorite.

It is obvious Google is doing it for their own benefit as well. The faster your page loads, the faster they can crawl.