Hacker News Reader

Comment by d-sc | Hacker News Reader

Tdlr: Googling “Powered by” seems to cause google to take excessive time to return results.

2.64 seconds for me (i guess it's cached now?).

Honestly, the mentality in this twitter thread is a joke. Google handles insane amounts of data beyond what 99.99% of engineers have ever dealt with. The fact that their search works at all is a miracle. But I guess people like to ignore that and much rather complain on Twitter that it takes a few seconds to query a database of trillions of records.

Google is famous for raising the bar on speed and yeah, their performance is usually impeccable. That’s why the existence of a consistently slow Google search is newsworthy.

It very clearly has to be a bug. I also got ~2.5s. It makes no sense for it to take that long for everyone, even after running it multiple times in a row. The system most definitely has many types of caching, so if it takes that long, there's probably something going very wrong.

There are comments on the thread that suggest that it is an intentional delay to prevent Google itself from being weaponized. Essentially, "powered by" is special because of automatic inclusions like "powered by wordpress" or "powered by nginx."

How can it be weaponized? Can you not search any other keyword used by those platforms? "by wordpress" is pretty fast.

I'm just pointing out the hypothesis and the train of thought. I don't know the real answer here. It's not my hypothesis.

If, for example, you wanted to harvest domain names because you had a hot 0day on wordpress, this would be a convenient way to do it.

Yes, there are probably other signatures, too.

For me:

About 6,060,000,000 results (6.82 seconds)

Well, thankfully you were here to defend one of the largest corporations on the planet from these Twitter meanies who dare discuss a slow responding page.

I don't see any factual objective discussion in the twitter thread. Btw, I'm not a big fan of Google, which I guess is what you were implying. In fact, I think Google is the most dangerous tech company currently in existence and I consider it a severe threat to our society. But that doesn't mean I cannot objectively judge the technical quality of their products and put things into perspective.

> I don't see any factual objective discussion in the twitter thread.

What do you mean? People are seeing slow queries for "powered by". Unless you think they are lying, I don't see how that isn't factual?

It's not what you say, but how you say it. This is just mindless disproportional bashing.

"Come on Google I don't have all day!"

"This is actually one of those crazy "facts from the future" to go back in time and tell people 10 years ago: Google has the dominant browser and mobile operating system, but their search sometimes takes 7 seconds."

"@Google have you tried implementing these recommendations? https://developers.google.com/speed/"

Doesn't seem like bashing to me. It is simply interesting.

Just seems like perfectly normal snarky banter making fun of a faceless corporation.

Still, you do you. I'm sure Google appreciates the peanut gallery coming to their defense for even the most trivial of issues...

I, for one, support the effort to raise the signal-to-noise ratio.

It’s really weird. If you add “powered by” to a query it’s really slow, but if you add a minor typo (eg “powere by”) it finds the same results, but faster.

According to https://twitter.com/syndk8/status/1157930276208750593

> Powered by is a footprint that black hats/hackers and spammers use to find targets.

So maybe Google isn't being slow, it's just sleeping intentionally to avoid abuse?

It looks like it is the exact substring that is slow. "asdfpowered byasdf" is slow, too. I bet it's happening at some layer that doesn't even know how to parse the query.

Since a lot of pages contain these exact words (WordPress, github.io, etc.), maybe it's actually the sorting of the result that takes time?

"amex platinum powered by mastercard" as well takes > 5 seconds

"a powered by b" takes >5 seconds

"x powered by header" takes >5 seconds

"x-powered by header" takes >5 seconds

"x-powered-by header" takes ~0.5/0.3 seconds

what.

“powered by” a popular term used in a lot of footers for a lot of different web forums/cmses, I wonder if that has something to do with it?

I had the theory that maybe they are trying to filter it out to frustrate searches for vulnerable CMS

Yea, I was thinking along the same lines. You’ll often see publicly disclosed vulns include google dorks to find vulnerable sites, ex: a google dork to find forums running a vulnerable version of phpbb or something

They're not the same query. They don't return the same results. I seem to recall a dash links words together somehow. So that probably narrows down the search space more.

> x-powered-by header

Is that a thing?

Yeah, it generally returns some variant of information about the technology that is powering a web application such as ASP.Net or express.

The last one may be from cache.

b-cache-busting-powered-by b

~0.4 seconds.

Google doesn‘t use stopwords because if they did I wouldn't be able to search accurately for information regarding the group "The The". But because we can find the occasional slow-running query they might be using "stop phrases" but weren't aware "powered by" should be one.

There is probably a small special index for things like "the the", and then the regular big index filters out such common words.

About 7.780.000.000 results (4,14 seconds)

I've found that if you go past page 2x on most of these results, Google just stops presenting pages and will tell you that it's omitted the results.

The number of results may represent number of hits in a database, but it's not actually accessible to the end user. After you get into the last few pages, it'll also just cap the result count to the number of pages they've decided to show you.

Reply on news.ycombinator.com