Ask HN: Is it just me or did Google search recently get a lot worse?

[alan_wade]

Past few days(weeks?) I find myself more and more frustrated with google search. It seems to frequently omit the most important keywords I'm typing. I constantly have to click "Must include 'Keyword'", because more than half the links on the first page miss it, even if it's the first one out of 3 words in my query.

I'm not the kind of person who thinks about search quality, this is the first time in my life I have ever noticed it being frustrating.

[einr]

It feels like the philosophy has become "return lots of results at any cost".

For instance I just now searched for a specific e-mail address that ends with "@hotmail.co.jp" -- since it didn't find any hits with the full address, it decided to strike out the username part and return thousands of results for just "hotmail.co.jp". This is literally useless and they should know better.

It just gets worse and worse for technical searches too, I find. Searching for stuff like datasheets for chips and vintage obscure computer programs and peripherals rarely returns anything useful without "wrapping" "everything" "in" "quotes".

[icebraining]

They have a Verbatim mode which doesn't do that (as much?), but you have to click on Tools → All Results → Verbatim every time.

Alternatively, if you can edit the URL used by your browser, you can just add "tbs=li:1" to it; for Firefox, you can add it as an alternative Search engine by running this on the Console:

  window.external.AddSearchProvider("https://gist.githubusercontent.com/popthestack/1530165/raw/c99a1c5e6e783aecffa432397a527f557944926d/google-ssl-verbatim.xml")

This doesn't invalidate the overall point, of course, it's just a personal tip.

[kazinator]

The principle is probably more like: return such approximate results that, among them, advertising doesn't look out of place.

[guhcampos]

What really, really bugs me is what they did with the image search at some point.

I always used image search to try and find the original source of an Image. It's super useful to determine the authenticity of a news article, identify the author of a photo for licensing and a million other uses.

Then at some point, search by image started to simply return results based on the image classification tag on the image: so instead of similar images, or other instances of a photo, I get results like "beach" or "bicycle" or "city". This is so frustrating and completely useless. I'm sure anyone is capable of typing "mountain" in the search field to find generic photos os mountains.

So that got me back to tineye.com - I just with they had a larger coverage of the web =[

[EForEndeavour]

I see that particular un-feature as a consequence of undue focus on ML that automatically generates image tags. While it's impressive for a system to be able to auto-tag billions of images at high accuracy, these generic tags are almost always useless as search terms.

[seventhtiger]

I've also noticed this and this is a big loss of utility.

There aren't many tools that find exact/slightly transformed matches of an image. That was very valuable, especially with the power of Google's webcrawling. Imaged get shared, stretched, watermarked, and artifacts build up. This was the single best way to find a source image.

Now the image is converted back into text and it searches for images using the text. We could search for images using text before. This is not new. The unique thing about reverse image search was killed and turned into just a fancy input for the same old image search.

[jkollross]

Have you tried Yandex's reverse image search?

[Timucin]

I've recently moved away from Google to DuckDuckGo, so didn't notice that but I noticed something similar on YouTube.

For some reason, it keeps recommending the videos I already watched, the ones in my save for later and the most annoying one: the ones from the channels I marked as not interested.

My recommendations also getting less and less relevant.

They definitely changed something and broke a perfectly working system.

[NeedMoreTea]

It's frustrating in its simplicity. I don't need 5 more versions of the same track. They're showing far fewer of the same genre and era but a lot more conspiracy theory garbage that I will never, ever intentionally click, yet their algo has never noticed the total lack of clicks on those "Proof of alien life", "brexit is great" or "lizards are running the show really" videos.

"recommended for you", yeah right. How do they arrive here from say 1 from CGP Grey, 2 from ACDC and 1 from Iron Maiden?

Last.fm used to have my music taste well nailed and followed up with a lot of unknown stuff that I ended up loving and buying. YT had more data for longer yet has no clue at all. Now I just search and play direct from DDG.

YT never worked that well for me, but now it's rarely worth even checking the other videos column.

[palerdot]

Youtube search quality is horribly bad in recent weeks. I'm being suggested videos that are minimum 2 years older continously in the home page. There are more recent videos from the same channels, but all I see is 2-3 year old video suggestions as if my phone/computer clock is reset to a date few years back. I'm not even sure if they are running some kind of experiment or is it a bug.

[nihil75]

Second.YouTube keeps showing me the same (promoted?) videos on the right-hand bar, so I keep navigating back to the home page to get half-decent recommendations.

[pjc50]

Searching for stuff you've already seen is reasonable. But reccomendation really needs to have a simple partition into two categories: watched and not watched. Both are valid but they are very different use cases.

Also they need to stop reccomending flat earth, anti-vax and other toxic nonsense without prompting.

[pure-awesome]

Is DuckDuckGo good now? When I first tried it out a couple of years back, I found it didn't give me enough (relevant) results on my searches, so I stayed with Google.

I assume it must have improved in that time; maybe I should try it again.

[SquareWheel]

It seems likely that people would search for videos they've previously watched. Must consider that these products are made for "normies", not power users.

[zython]

I had a similar experience with youtube lately: I was looking for a song that I knew the tune and a related artist and for the love of god could not find it with the current youtube recommendations. Most of the videos I'd get recommended when searching related songs, were not even music-related. So I went to archive.org and within a couple of results found what I was looking for.

Youtube's recommendation has really changed for the worse.

[saiya-jin]

its even worse with youtube - if I leave the TV, it will continue playing videos youtube considers somehow related to original one. But it often forgets what it already played in a row, so sometimes plays the same video multiple times. Like 1 2 3 4 2 5 1 6 5 3 etc.

[buboard]

most frustrating when you re watching a series of lectures and it recommends the previous one , or breaks the series

[keiferski]

The "must include" functionality makes zero sense to me and I'm dumbfounded as to why it was ever added. If I didn't want my results to include a particular word, I wouldn't have included it in the search. Boggles my mind how Google could mess up something so simple.

[samwhiteUK]

I'm dumbfounded as to how you don't see the usefulness of this feature. Sometimes there exists more than one word for a situation, and you might miss a forum thread, for example, that is relevant to you, but you're saying to Google "only use these exact words" and not accounting for differences in dialect etc

[DarkWiiPlayer]

Because people are dumb and type in entire questions or sentences, so google tries to be "smart" and remove words that hold no real meaning.

[tinus_hn]

One of the original defining features of Google was that all results included all terms. Now even when you search for +”term” you get results without “term”.

It seems just stupid to return these completely useless results but I guess it makes them money somehow.

[Yetanfou]

Search engines - Google amongst them - are moving from using the search term as (possibly stemmed) literal words to search for to attempting to interpret the 'meaning' of the term. This interpretation can be somewhat... lacking, with the observed result of nonsensical answers.

[natch]

They have definitely messed up in many ways, but having a must-include operator was not one of the ways.

It allows (allowed?) you to tell the search that the other words can be used for ranking and are nice-to-have words, while the must-include words are treated differently and must be present and match exactly.

If all search terms were automatically must-include with no override, search results would be garbage because as a user you aren’t going to predict the perfect inflected forms of words used in documents every time.

Being dumbfounded should not lead you to think other people are wrong. It should lead you to wonder what you are missing.

[Jedi72]

It seems only a matter of time until Google is just an interactive yellow pages. Why would you spend millions trying to index peoples personal blogs etc. when companies will pay thousands for the spot in the results and give you the link to put up? All they need to do is maintain their position as the default internet portal.

[quickthrower2]

People advertise on Google because of the massive audience, and that audience exists because of quality organic search results. They are not going to mess with that.

[charlesism]

It's trash now. I noticed this a few months back:

    The final straw for me is that it now "helpfully" drops 
    search terms on my behalf. I was already using DDG for a third 
    of my searches. I'm going all-in now. For people who don't spend 
    much time online, maybe what Google is doing works for them.

https://news.ycombinator.com/item?id=18266966

[brazzledazzle]

I think Google is focusing on the friendlier natural “human” queries. I noticed this some time ago when I’d ask it in a “lazy” way because I’m tired or distracted. E.g. “What day is Mother’s Day?”, or “What time is it in London?” I’ve noticed it seems to be more and more optimized for this sort of query as time goes on, to the point that I use the quote modifiers in many (most?) of my searches and the other modifiers more frequently than I used to.

I suspect that the people who’ve noticed this and feel frustrated have, like me, been searching on Google since back when it was on its way to unseating Altavista/Yahoo/etc (or maybe somewhere between). But I don’t hold it against Google because they’re optimizing for the widest possible audience and if it makes it easier for most people I’m ok with being a little inconvenienced. As long as adding quotes or other modifiers kicks in the old/advanced search I’m ok with that price for something I not only don’t pay for, but don’t click on the ads for either.

[pure-awesome]

I remember the days when I had to include "AND" and "OR" between words in the queries.

[saluki]

It feels like maybe Google had a changing of the guard and a new batch of engineers are making decisions and missed out on a transfer of knowledge from the previous engineers with both search and SEO.

I'm seeing my search results show less and less of what I'm looking for, it's like google search is forgetting how it used to work.

The same with SEO, I'm seeing sites ranking again on the first page that have useless content using all the tricks/ghosts of SEO past. It seems all the legacy SEO hacks/tricks filters that had been in place at Google have been removed recently.

I was a huge fan of Google, but I'm definitely seeing a decline in quality of the SERP.

[quickthrower2]

Don’t forget that the web is changing. There is more junk produced every day and Google needs to combat search engine spam and blackhat practices. This means changes and not seeing search results you are used to.

[Yetanfou]

A good solution to all this $search_engine misery is to use something like Searx [1] as an in-between. This gets rid of most of the profiling - especially when using a shared instance run by someone you trust (to avoid being profiled by the Searx instance, self-host and have family members and friends use it) - and gives results from a whole host of search engines. Configure all your devices to use that instance and you no longer have to cope with any single $search_engine's 'new and improved' algorithms, instead being treated to a search salad with ingredients of your own choosing.

[1] https://github.com/asciimoo/searx

[whateveracct]

Sounds like someone in Google had a successful feature launch :) good for them on delivering something to production!

[ur-whale]

The quality has indeed declined drastically for pointed technical queries versus "where's the nearest hot dog stand" type queries.

Two possible explanations:

1. The latter is much more of a moneymaker.

2. I don't think there is anyone left at Google that actually understands how the search engine works these days.

[enz]

I have the same feeling, I have to click "must include keyword..." to get better results. I thought it was because I use google.co.jp since I moved to Japan a few weeks ago, so I was guessing Google tried too hard to correlate the results with my new geographical area. It seems I'm not the only one.

[s9w]

Not sure if it was in the last weeks.. I would say in the last months. But google searches without verbatim mode are increasingly useless.

[synesso]

Now you mention it, yes. I've been seeing this for a few weeks. It seems often the most salient word in my search phrase is ignored.

[EGKW]

Totally! Actually not 'worse' as such, but search results are more centered around earnings, rather than around providing information. You can't blame them for that, but by shifting focus they can expect a shift in public too.

[apatters]

I commented on a different post a few days ago with my thoughts about why this is happening, in my opinion it is a problem that has evolved over several years:

https://news.ycombinator.com/item?id=19111306

A Search Quality engineer at Google responded to me but I didn't find their arguments very convincing (nor did a number of repliers).

I have switched over completely to Searx on most of my devices, and by telling it to weight results from Bing, DDG and Google I feel I get better results than from any of those search engines alone.

[hallihax]

I agree. Google has consistently gotten worse imo - I believe the general move towards third-party 'trust' verifiers, and Google's obsession with 'relevance' (as in, relevance to me) has destroyed the entire point of a search engine. I rarely use Google these days except to quickly re-find something I discovered somewhere else. I never really use it for finding information these days - just getting a link to something I already know exists. If I had a better way to search my bookmarks, I'd probably use that instead.

[Jefff8]

Similar. I can't quantify, but I seem to be getting less useful results, certainly on the first page, and often beyond. I'm not sure it's not just perception though because I've not kept any log and it's also possible there are genuinely no matches to the searches I've been doing.

[twic]

Also similar. What do we mean by "recent"? I feel like it got a lot worse a year or two ago, and has been consistently bad since then.

[r_singh]

This question has been on my mind too and I'm glad someone thought of asking on HN.

Google has been returning really obvious results that one expects to get based on search history, other internet usage, etc (much like Netflix's matching). So much so that in order to find really unexpected gems of info I need to really spend a great amount of time searching a lot of different websites (hacker news, reddit, linkedin, corporate affairs data, marketplaces, truecaller, etc etc).

I feel like we need a new search engine now that there are so many different types of search results (blogs / media, aggregators / marketplaces, directories, social profiles, company websites, etc). One that lets us take more control over the type of results one can expect and more importantly one that doesn't show you what you expect to see (after being heavily influenced by your previous activities).

Also, Google may have millions of results for every keyword, but the quality of those links deteriorates quite fast, post 3rd page results are just crap.

I genuinely believe that Google worked and innovated when other search cos were wasting users' time by not rendering results quickly, but today we need something that better suits the complexity of the web and doesn't heavily rely on usage histories, etc.

[thiago_fm]

Just search using "". For me it didn't change, as the way I search by using "", + etc since I started googling didn't change and it is even obvious in some ways that maybe I could just ddg all the way.

[x38iq84n]

I have had the same feeling for months and not limited to Google, sadly DDG suffers from the same. Search result quality has been getting worse and my search queries contain more and more quotes just to get back to a previous level. What I miss is a tick box for "power users" to disable spelling suggestions, search terms omission and similar "features".

[pi-victor]

I use DDG but when i wanted better results i used google. It gets harder to differentiate, they both miss the point.

Granted search engines are no psychics, you also need to better phrase your searches sometimes.

I remember Google used to return way more accurate results a few years ago. I guess privacy regulations, increasing demand for higher ad revenue have made it a lot worse than it used to be.

[brudgers]

It's just worse in different ways (and better in different ways as well). As Google's index grows it shards across more machines. The odds that latency will affect the result of a particular query increase simply because more machines and network connections are involved. So there's caching.

Because no one wants to wait forever (delay is indistinguishable from failure), Google returns the best results it can obtain within some time threshold. Unfortunately, the utility of a search term is often its rarity relative to other search terms and the rarity of a search term makes it less likely to be cached nearby. If "nobody ever searches by that term" it is more likely not to be indexed or on a far away partition. Think of adding a Korean character to a string of English search terms : In Korea : In the US.

[xorand]

Google left for another dimension and they needed all their computing power for that, so they resorted to this trick. The results we see are from a NN previously trained on the real google, which is now hosted by a computer in the janitor office.

[fergie]

It definitely did. As a complete outsider (albeit one who has done a bit of contract work at Google a few years ago), it seems that they are going all in on ML-driven personalisation, rather than an "objective best match" approach.

[untangle]

Google search has gotten worse and DuckDuckGo has simultaneously gotten better! I have (again) been using DDG for the past few weeks and I type "!g" once a day at most. Very exciting if this seeming trend holds.

[jim-jim-jim]

Glad I'm not alone. It's so bad I'm actually _using_ DuckDuckGo these days, instead of instinctively typing in g! before all my queries like I usually do.

[anc84]

Absolutely it got even worse in the past few weeks after an earlier, perceived drop a couple of months ago.

[shash7]

I've noticed a lot of crappy youtube videos being inserted in the search results. These videos are far from relevant from the search query.

Unfortunately, duckduckgo and the like aren't still as good as google else I would've shifted to those search engines a long time ago.

[segmondy]

I had an issue this morning where I wanted to download some youtube videos for offline viewing and they were already deleted. I used google to try to find other copies and it found nothing. I used bing and was able to find all copies.

[qnsi]

Does anyone have suggestions how to find good tutorials/blogposts? I am interested in startup world, but when I search for things like "how to find startups ideas" I am tackled with lots and lots of content marketing, or poor medium posts.

(I know of firstround search and I think this is the best you can get. I was thinking there should be a website that aggregates articles linked on twitter by people from VC/founders world, but I don't think there is one sadly)

[BidCoins]

Interestingly enough, I was JUST today thinking to myself, has google auto-suggest been consistently degrading or is it just me?! Now my thinking is there is some over-engineering going on (for lack of better articulation) as type auto-suggest as well as speech-to-text are just plain "off base" + your observation of search results, I'm sure they're all on a common, core "function" ; )

[nullandvoid]

I have encountered this with YouTube which prefers to offer me garbage results back rather than searching all my keywords. E.g I was searching on a react topic and all I got back was 'x reacts to this'. It looks like it weighed the word 'react' much heavier than the others even though the other words were essential to getting a useful result..

[luord]

Indeed, that's one obnoxious "feature" that has become omnipresent (as in in every search) lately.

[belltaco]

It seems to be weighting recent web pages a lot more since a year or more. It's annoying to search something and get 10 pages of recent pages rather than something from a year ago mixed in somewhere.

[buboard]

I am noticing an effect of "my searches don't get better as i add and change keywords". Becoming more specific has become difficult. thankfully there are alternatives

[growlist]

For me it's been getting consistently worse for years, not weeks or months. Is there an 'exact search terms only' search engine out there?

[dennisgorelik]

Several days ago Google Search slowed down significantly for some of the queries I was making.

Several search queries took 2-4 seconds to return results.

[e9]

pretty much garbage at this point, I honestly don't know what to do anymore. I tried Duck Duck Go and Start Pages but they are not as good. I am testing Yandex, so far it's really good and performs like old Google but I don't know what are the implications of me using Russian search engine :/

[qbaqbaqba]

Mobile searches are even worse. First page is occupied with links to search within huge portals. Nothing relevant.

[dfgert]

Are there any alternatives? I am ok with indie, slow, paid, not-super-accurate options.

[operatorequals]

What's the business model of it anyway? Is it providing you good search results?

[SquareWheel]

Search's business model is showing adverts at the top of the page.

[kowdermeister]

No, good as always :) But seriously, yesterday I wanted to deploy my beta to a fresh server install and I did like 10 google searches of various server and tool issues. I got it all resolved and the beta is running but I didn't stop at how bad these results are.

What are you searching for when you see a decrease in quality?

[trumped]

Since Google+ is pretty much gone, now is the time to bring back the + operator.

[paulcole]

Can you give a single example?

[tannhaeuser]

Google search superiority is a myth IMHO. If you're a SW engineer, you're better of searching StackOverflow and using an old-school curated list of links for on-topic sites. Basic thinking tells you that Google Search are incentivized to show you pages with the most ads, rather than relevancy. There aren't that many good resources for any given topic anyway; creating a simple search index over those isn't rocket science. Certainly, "searched 130.000.000 sites in 0.13 ms" style info isn't helpful, unless you find Google giving you random crap sites entertaining.

[isostatic]

If google works for you, great.

> 99% users of this website are software engineers

Citation Needed