How bad are Google search results now?

The pages returned do not necessarily have all the words

[beren]

I wanted to find out if ordinary people consider Thomas Edison to be a great American.

I typed this into the Google search field: opinion poll american heroes edison

Number 3 on the results was a page from the National Cancer Institute on colorectal cancer. This is completely irrelevant to my query. I looked at the page and NONE of these words are on it: opinion, poll, heroes, edison.

It is true that I did this search while logged into Google and I was looking at National Cancer Institute pages a couple days ago. In my account settings I have "Personalized results and recommendations" paused. But even if that were not true, Google should show a result that relates to my query.

Twenty years ago Google stood out among the many search engines at the time by having BETTER results. And their results were pages that contained EVERY word you searched for. This is insane.

[not2easy]

Are these mobile or desktop searches? What browser are you using? Have you tried the same query on other search engines such as DDG or Bing for example? Have you tired using fewer words? I don't know that any of these things make a difference but they can affect results. I would also suggest that 'heroes' isn't a word I would apply to inventors or engineers of over a century ago. You may have frightened the AI.

If you did find a poll of American heroes I doubt that Edison is even mentioned since his work preceded opinion polling by several decades. Today's American adults may be only vaguely familiar with the name, based on their confusion between the ancient Greek author of the Odyssey and a lemon colored cartoon dad. You might get a better idea of opinions on Edison searching for the types of books about him that are more or less popular than others (imho).

[EditorialGuy]

I agree with what not2easy said about your Edison example. I'd also point out that search results aren't necessarily about specific keywords or keyphrases. As Matt Cutts said years ago, Google has moved on to "things, not strings."

[lucy24]

I see you tried the same question on reddit before coming here.

How do I know? Because if I enter "opinion" "poll" "american" "heroes" "edison" (each word in quotes) the first hit is a reddit thread.

To avoid contamination, I used a different browser, where there's no history, no old cookies, and I'm not logged in. With this approach, I found a rather interesting article in Discover magazine on people perceived as Heroes or Villains. There I learned to my disgust that Mother Theresa [sic] is ranked #2 in admiration worldwide behind Albert Einstein, while George W. Bush is #4 on the villains side . . . but I digress.

Finally: If I had this question, I would have started by asking it as-is: Do ordinary people et cetera, exactly as worded in OP. (As it turns out, this wording won't answer the question, though G will understand that you want to know about Edison.)

[RedBar]

For me all you have proven is that one should not be logged-in to Google and that one should clear cache and history every time the browser is closed.

BUT by doing this one will not see the garbage answers G serves based upon JP's browsing history!

G assumptions + browsing history = horse manure

[beren]

I know Edison lived before opinion polls. That's besides the point.

The point is that Google returned a National Cancer Institute page that did not contain Edison's name or several other words in my query. This is the sign of a non-functioning search engine. And it is not a matter of AI interpreting things incorrectly. Colorectal cancer has absolutely nothing to do with my search.

[tangor]

Heh ... a Bing search turned up this thread as number one. Number two and three were about Edison and achievements/benefit to mankind. Yeah, you do kind of have to wonder what's going on with g search these days (one reason why I don't use it personally, but do check it for "results" for the g side of the web).

[beren]

"Are these mobile or desktop searches? What browser are you using? Have you tried the same query on other search engines such as DDG or Bing for example? "

Desktop, Brave. I most often use Brave's search engine these days. I know I could have tried DDG or Bing or found another research method, but the point is that Google is not functioning as a search engine.

"If I had this question, I would have started by asking it as-is: Do ordinary people et cetera, exactly as worded in OP."

What? Only someone who is new to the internet or not used search engines would type "Do ordinary people...." People who know what they are doing type in words to see in the pages that come in results.

Google is not an answer machine. It is a search engine. I don't want it to answer what it thinks is my question. If I type 5 keywords into the search field I want at least the first page of SERPs to be pages with all five of those keywords.

[Rndm]

What? Only someone who is new to the internet or not used search engines would type "Do ordinary people...." People who know what they are doing type in words to see in the pages that come in results.

Google is not an answer machine. It is a search engine. I don't want it to answer what it thinks is my question. If I type 5 keywords into the search field I want at least the first page of SERPs to be pages with all five of those keywords.

The majority of people do not use search engines as you do and in fact, they do the exact thing that you suggest that no one does. They also expect Google to answer their question. What Google is and what you hope/expect it to be are very different things.

[not2easy]

Today's users are more likely to be using a mobile device and may be searching verbally - without typing anything. Most on this forum may be accustomed to typing in searches, but worldwide, most users are on mobile and have been using voice search for awhile now.

[RedBar]

27% of the global online population is using voice search on mobile.

I'm not too sure whether this is 2018 or 2021, nonetheless it's a big number!

[thinkwithgoogle.com...]

[NickMNS]

@RedBar
That's 2018

27% of the global online population is using voice search on mobile

Be careful with the wording of the statement. The statements is vague, it is not saying that 27% of all searches are voice searches. It says that 27% of people are using it. I guess I would be included in that because every now and then I ask "siri" something. But 99% of my searches are not voice.

[randle]

How bad are Google search results now?

The answer to the question is a bit of a paradox. I remember once I met someone at an event and I started the conversation off with, "boy its a nice day". His response was, "relative to what?". When looked at from a pure ranking of quality information perspective, Google search results are pretty bad, but as my new acquaintance so eloquently put it; relative to what?

People that search for things online have been conditioned to view the entire concept of search through a lens Google has created and perfected. The less you know about search, the better the results look and those that know the least offer the best monetization opportunity for Google. They see what Google wants them to see, they go where Google leads them and because they have no clue how the sausage gets made there happy with the results.

[Niresh12495]

Google results are total Bad.
I remember it being great between 2012-2016
This is my personal opinion.

[robzilla]

Yours is an awkward query and Google certainly struggles with those sometimes. You can tell when lots of Google Books results show up or they start scrapping keywords early in the result set. I agree with not2easy that you've probably "frightened the AI" (LOL) with your query. But, in your defense, and I'm sure they'd agree with this themselves, ideally Google ought to have good answers for lesser queries, too.

I'm not sure where the cancer result would come from. The combination of edison, american, opinion and poll would sooner point towards US elections (parts of which may be described as cancerous, but let's not give the AI too much credit). I suppose some unhelpful associations are made behind the scenes.

Maybe it's collateral damage from their improved understanding of natural language. Give the "phrase" opinion poll american heroes edison to humans and they may not really understand what you're looking for either (though it'll be clear to them it's probably not cancer-related). Ask them, "do people consider Thomas Edison to be a great American" and everyone's on the same page, including Google (as evidenced by the result set). We've been taught to search using keywords by old search systems, but that's not the natural way, of course.

Saying Google results are bad because "the pages returned do not necessarily have all the words" seems to me a little too old-school. We've passed that station at least a decade ago.

I see the algo struggling with some queries, and sometimes I get frustrated by that, but on average I think the quality is still higher than the alternatives. Certainly not "total bad", I think that's an absurd thing to say.

[tangor]

I just think if I put four terms in a query I should get returns that contain all four keywords ... or at least a button what will "Do you want to see ONLY pages that contain w,x,y,z?"

There should be alternatives.

[robzilla]

Query intent trumps keyword matching. You don't want to put the onus on searchers for finding the right keywords to use.

If you insist, just put quotation marks around the keywords.

[tangor]

That's possible, doesn't always work without the "+" operator ...

Then again, DDG, Bing and others do it the first time so ...

YMMV

[robzilla]

I'd forgotten, but Google removed the plus operator about a decade ago.

Then again, DDG, Bing and others do it the first time so ...

If you mean to say those search engines only show you results containing all your keywords, that's simply not true.

Easily verifiable using beren's query.

[engine]

How bad are Google search results now?

That's a very subjective title.
I wouldn't turn it round, either, to "How good are Google search results now?"

The whole way that people search now has changed a great deal, and days of ten blue links are way behind us, and that'#s nothing to do with SERPs crowded with ads.

If you take out the ads, just look at all the other material in the SERPs.

Today's major search engines are much more sophisticated than they were at the start, and it's about learning to use them differently.

BTW, I often find that the SERPs can be a mess during updates of various kinds, and i'll usually indicate the problem when I see it. It often works its way through in time.

[londrum]

sometimes the words that google chooses to ignore are really important words, try searching for these
-- london pub without music
-- london pub with no music
and most of the results returned are for places with live music -- the complete opposite of what was asked for
google should be able to tell by now that the word "without" is what the searcher wants

[engine]

I agree, sometimes it's odd, however, there's a clue if you put the keywords in quotes. "No results found for "london pub with no music"."

[NickMNS]

@londrum
I have experienced his too. Google is really bad with negation. I find nearly impossible to get good result with terms "no" or "without" or "not including". I'm speculating that this is due to the fact that words such as "with", "no" and "not" can be considered stop words and then are eliminated from the query at some stage of the processing. What I find interesting is that if you reword your query in such a way to preserve the intent but eliminate those words you end up with good results. Example would be quiet london pub instead of london pub with no music.

[squarebracket]

It's true that some searches are difficult, but that problem affects all search engines. That's why Google search has special operators, like the 'negation' operator (-). Try this search:

london pub -music

Also see:
[makeuseof.com...]

[robzilla]

The results from TripAdvisor suggest "without" isn't completely ignored, it's right there and highlighted. But it does seem to be the first word to be thrown out. Google, Bing and DDG all give mixed results, mostly unhelpful. Bing arguably has the best result at the top, Google has a higher number of relevant results, DDG has absolutely nothing. Clearly, like beren's, this is a tricky query for a modern search engine.

I can sort of deduce from the query that you're probably looking for a quiet pub, not necessarily one completely devoid of (background) music, but I suspect that's still a bridge too far for these algorithms. A pub without music is, after all, not necessarily quiet.

It's possible that more of an exact match approach to this query might have yielded more results, like the ones you get for "london" "pub" "no music" (mostly UGC, which may be a hint; there's a trade-off somewhere). Many queries will benefit from the AI's interpretation, but not all of them just yet.

Try this search:
london pub -music

If only :-)

[londrum]

@nickMNS
yup. and interestingly, if you try adding the original 'without music' and 'with no music' to your 'quiet london pub' query in quotes, then it does return relevant results, and the pages include the search terms 'without music' and 'with no music'
-- quiet london pub "without music"
-- quiet london pub "with no music"

so the results were there to be had, but google just chose to show the opposite
if google gave greater weight to the words on the page they might have shown them

[NickMNS]

@robzilla yes "quiet" as a concept is not the same as no music but neither are your suggestions:

"no music" indicates that you want Google to return results that explicitly contain the words "no music" as is.
-music indicates that you want results that do not contain the word "music". There are certainly many pubs that play music but do not make mention of it.

This thread illustrates the challenge which remain for search with things such as negations. After all at some basic level the interpretation of such statements is ambiguous. The ambiguity is easily resolved in natural human to human conversation by context, but for Google or any other search engine it remains a struggle to capture and then understand such context.

The truly amazing things (regardless of you opinion of Google as a company) is how far search has come over the years, from matching words 1 to 1, to where it is today.

[robzilla]

@robzilla yes "quiet" as a concept is not the same as no music but neither are your suggestions:

"no music" indicates that you want Google to return results that explicitly contain the words "no music" as is.
-music indicates that you want results that do not contain the word "music". There are certainly many pubs that play music but do not make mention of it.

The latter was clearly not my suggestion, the former was merely an illustration of more exact match type of search results, where you're essentially forcing the search engine to turn off its query interpretation and focus on the words on the page, as was suggested as a solution earlier (i.e. for Google to give more weight to the words on the page). It shows you the results that could be considered relevant (lots of pub suggestions to be extracted there) but that you're not getting with the original query because just words on a page aren't enough to rank. Other signals are taking precedence, but clearly not leading to query satisfaction here.

The reason the TripAdvisor pages rank first is probably that, for those, Google is at least confident enough that they're fully about what is queried. They're just about the only pages on the Web that have all the keywords in their title. From there on, it all falls apart as keywords appear to be variously ignored, so that you get some results thrown into the mix that you would also get for queries like london pub music and london pub without. That's what it looks like to me anyway. I don't think that's limited to words of negation, but different types of words (i.e. parts of speech) probably do get different priorities.

Missed opportunity here for the owner of Quietpubs.com ;-)