Google returns unwanted search results

Why does Google often return results not containing one of more of my search criteria?

For example I just made a search on: astrid “morning star” venus
The two first results are Wikipedia on Venus, and an astronomy page by a guy called John Pratt. Neither of those two pages have the word “astrid” anywhere on them.

Your question doesn’t make any sense from a boolean logic standpoint. You said that you wanted Google to return pages that contained ONE OR MORE (I assume that is what you really meant) of those search terms and that is what you got.

Both of those pages got highly ranked search results because they are among the most popular sources on the topic “morning star” while “astrid” is a much more rare term so it doesn’t influence the search as much.

Google doesn’t like to do simple keyword searches. It tries to be much smarter than that unlike the older search engines. However, you can make it do some pure keyword searching if you add the plus and minus qualifiers before each term.

For a search of pages that contain BOTH astrid and “morning star”, the correct search terms are:

+astrid +“morning star”

That search works correctly although the results all appear to be gibberish. Maybe those aren’t a good pair of search terms to use together.

No, I meant it to return pages that contained all of the search terms. I thought the + sign was obsolete and all terms were assumed to be prefixed with the + unless specifically removed with the - (minus) or by using the OR keyword.

Like what it says here:

12 Quick Tips To Search Google Like An Expert

Searching +astrid “morning star” now returns this thread as the top result. But I’m guessing that’s not what you were looking for…

At any rate, I’m guessing your results are due to there being very few pages with all three terms. Google’s algorithm will try to show you the best results it can, even if they aren’t exactly what you asked for.

Is there some page that you think should have showed up in your results, but didn’t?

Somebody is lying or at least partially mistaken. Try the two different searches. They have different results and only the one using + operators is limited to pages that contain both of those terms. However, Google doesn’t return any meaningful pages that contain both of them.

Are you sure that is the search you really want?

This is nonsense. The original search was perfectly coherent. In the good old days, before Google turned to shit, and when it still behaved reasonably transparently, it would automatically boolean AND all your search terms (unless told otherwise). No doubt this was the behaviour that Rune expected, not unreasonably, because it still does that, unless it does not feel like it for some reason, and prioritizes sites that it particularly likes for one of the terms, even if they do not contain others of the terms at all.

At one time it was indeed possible to use the + operator in front of a search term in order to force Google to include that them in any sites it came up with, like a true boolean AND. However, the + was quietly decommissioned quite some time ago, and (unless they have secretively changed things yet again), no longer works that way, if it does anything. (It may invoke a search of Google+, but I am not sure.

I just tried +astrid +“morning star”, as recommended by Shagnasty, and it returned no hits at all!

The functionality that the + once provided was supposed to still be available by placing any search term that you definitely wanted included in all results in quotes. Thus Rune’s best bet for getting a truly boolean ANDed search on his three terms would be to use: “astrid” “morning star” "venus"

My impression, however, is that this still does not work reliably. It may give you a better chance of getting pages with all three terms high up in the listing, but I think you will still find age with only two (or even one) of the terms fairly high in the listing, and there may well also be pages that have all three terms, but a low PageRank, lower down.

I just tried “astrid” “morning star” “venus” and got “About 2,180,000 results”, but the first hit is this thread!

Google, these days, sucks. Unfortunately, for most purposes the limited alternatives (limited because most of the competition has been driven out of business by the Google steamroller) suck harder. :frowning:

Again, is there a result that should have come up, that Google failed to find? If the page Rune wants doesn’t exist, it’s not Google’s fault.

Although I suppose you could answer “How would we know if there was?”

Rune, what were you looking for? Maybe we can suggest a better query.

That is a little strange. I get results for that search although no good ones. The result set is different than the original search so the + is still doing something.

ETA, I need to make a correction. It is looking specifically for pages that that contain the literal term “+astrid” and there obviously aren’t many of those. You are correct that the + sign doesn’t do anything useful anymore unless you want to find pages with plus signs in them.

No, it was just an example of something that has been irritating me for some time on Google. There might well be no relevant page anywhere on the net. But it’s irritating having to go through all the Google results only to discover they not even contains the things I’m looking for.

(In any case, it comes from remembering reading many years ago that the female name Astrid was a Nordic pre-Germanic (bronze age) equivalent of the Greco-Roman love goddess identified by the morning star. And was just making a pot-shot for it on Google. But that is another subject.)

I note that if use the search string +astrid +venus the first few hits are from Google+ and after that come some that contain the actual strings “+Astrid” and “+Venus”.

Google has “improved” their searches. So now you have to go to the tools and use the verbatim button so that Google doesn’t decide to “help” you.

I do not know which “should” have come up, but using the string I suggested - “astrid” “morning star” “venus” - does indeed, as I said, return many hits, the first being this thread, followed by several other pages that do indeed seem to contain all three search terms. To that extent, it works. I am quite confident, however, that all those 2,180,000 hits do not actually contain all three terms, so it is still not doing a true boolean AND. I am not going to waste my time going down the list to find how far down the first hit without all three comes, but on past experience it will not be all that far.

But it IS Google’s fault if it gives you results you don’t want. If you sent someone to the store to get bananas, and the store was out, would you want them to come back with a sack of carrots and say “Here you go”? Of course not. You’d want them to let you know that yes, they have no bananas.

If your search yields no results, Google should tell you that. And it used to.

It’s a judgement call though when no results is better than results that don’t quite match. If the user really wants to know “Are there any pages on the Internet that contain the words astrid, Venus, and morningstar?” then yeah, no results would be the more useful response. But if the user wants to know “more information about that other morningstar goddess, kinda like Venus, I think her name was astrid” and then it turns out her real name was Freja or something, then it would actually be better (i.e., more useful to the user) to return that result even if it doesn’t include the name astrid.

I think Google is betting that returning some results that don’t quite match is more often than not more useful than returning no results. (Where “betting” probably means “their research suggests”).

Although I can see how it’s annoying in this case, I’m not so sure it’s the wrong call on the whole. But I agree there should be an option to get just what you asked for. As far as I can tell putting all terms in quotes does exactly this.

Try Google UK; typing astrid “morning star” Venus into it does get results with the word “astrid” in them. My guess is that you’ve run into one of the side effects of the “improvement” to Google US that makes it impossible to turn off safe search. Astrid and Venus are female names; it likely thinks “porn!”.

I don’t know for sure, but I tried some random ones on page 50 of the results and they still seemed to contain all the words.

But presumably Google’s search algorithm thinks these pages are highly unlikely to be what you’re searching for, since they didn’t come to the top when the words weren’t quoted.

I think you will find that that does not behave reliably and transparently either. Even if you tell it to return only pages with “all the words” it may still return some, even fairly high up the list, that do not.

But are you sure that there are not some results on page 3 or 4 or 44 that do not contain them all? Maybe there isn’t, but I would not be confident of it.

I remember masturbating to Delta of Astrid.

2 possibilities:

  1. google also includes a pages header data, and the text of links to that page to determine the keywords for that page. So, one of your search terms may have been used to label a link to that search result. To eliminate this behavior, put “intext:” for a single search term, or “allintext:” for all of the search terms.

  2. Google also includes synonyms and alternate meanings (e.g. psu would also search “power supply unit” “penn state university” in addition to “psu”). If you want to eliminate these results, surround the search term with quotes.