Google should be split up. It's wrong to have a company that controls search, one of the major browsers, and one of the two major operating systems for phones.

And then of course it should be made to pay tax where the search results are shown. An ad shown in the EU should be taxed with an indirect EU tax on the ad (30 percent of what is paid for the ad plus an EU VAT). That would once for all make the large advertising companies such as Facebook and Google pay tax in Europe. Every ad should include an ID number with information on the money paid on the ad, who paid it and to whom. Then, consumers and EU tax authorities could easily check if tax had been paid on the ad and who paid what for the ad.

Taxation philosophy aside, I really like the idea of traceable ads in general. There are so many intentionally tangled and obscured intermediate steps between the ad buyer and the ad viewer, a transparent record would benefit the honest parties and the consumer. Prosecuting, or at least deterring, malware served over sketchy ad networks comes immediately to mind.

Is there ever a case where untraceability is a good thing?

And runs the largest advertising network on the planet as well as the major email service provider.

> Google should be split up.

Agreed. One question, though: I somewhat trust Google's security, but is our data still safe after such a split-up?

I also have a fun one:

https://www.google.com/search?q=%22ketmax%22+zip

It's hilarious because it's now gotten to the point where the only non-noise results are the ones where I'm describing how the query in question is broken.

In fact, the top result I'm seeing here is a perfectly-paraphrased oneliner - Google is describing its own brokenness! :D

(I must admit, I'm impressed to see such a perfect summary of the situation presented in a single paragraph like that. Google picked exactly the right portion of the sentence. I take it there's really nothing out there in open-source land that does anything like that, is there? :( )

I think they just picked the sentences that contained the keywords, and concatenated them together (if necessary because the sentences are in two different locations). The sentences themselves can be found using basic word searches once the content is parsed/indexed (obviously, the Google indexing/searching is more complicated than that).

Both of my company's database engine products offer this type of functionality, as do many database engines/servers out there. It's typically referred to as full-text indexing or full-text searching:

https://www.postgresql.org/docs/9.5/static/textsearch.html

and there are lots of ways to control how/why certain words are indexed/search (stop words, synonyms, stemming, etc.). With the right amount of customization, you can get pretty close to ideal search results.

> Google's ranking algorithms are notoriously secretive due to attempts to resist SEO spamming

Why can't they approach it like software security? E.g. Linux's source code is open, and this is considered a good thing regarding security.

The kind of security Linux does is binary and has correct answers: should this user be logged in? Can this user write this file? Should this [more complicated access using SELinux] be allowed?

Search rankings are relative to each other and do not have objectively correct answers. Furthermore, I think the methods people have discovered so far to produce search results users are happy with are all gameable to some degree, so the only option available at this point is a degree of security through obscurity and an ever-evolving arms race between Google and spammers.

But in this case it was lack of original content, which is, pretty much, not a secret.

And Google shopping is full of original content?

Search results were littered with garbage comparison sites when this began though. My comment was more a response to the search ranking point rather than the merits of google shopping, which I'm no great fan of btw.