Hacker News Reader

Comment by lykr0n | Hacker News Reader

I wish there was a way to directly help Bing/DuckDuckGo/Yandex improve their search results. I've tried both, and it's just not the same.

Google I can bang in cryptic queries like > centos 7 tuned no daemon

and get the 3rd link about how to run tuned in a no daemon mode. Bing/DuckDuckGo have the article at around 7th or 8th place, but prefaced by a lot of "while technically not wrong, not what I'm looking for" links. It's even worse for more niche errors or code snippets.

We cannot, as a healthy internet, let Google control so much of the web.

View on HN

I felt the same about Google alternatives up until about 3 months ago. Google's results have been declining in quality for a decade, with much more rapid decline over the past year or three.

Google's results are uglier and blatantly revenue based. They have now lapsed behind DuckDuckGo in usefulness for me. I fall back to Google a few times per week, with inconsistent results when I need a "second opinion."

I'd suggest giving DDG another try.

I plan to remove Google from my life this year, at least as a central dependency. Search is already behind me. Mail, calendars, docs, and drive will be taken care of throughout the year. And my Android phone will be replaced with an iPhone.

Er, who do you think Apple pays for search on iPhones?

Pretty sure google pays them for search

Google pays them to be default search engine. Users do have option to switch.

My bad, thanks for the catch.

Done. Let's see how it works out.

Google always says x many results and when you click page 5 6 or some it is usually empty... not as many results as displayed on first page

Do you (or anyone else) have any recommendations for an email service? I've cut pretty much everything Google out of my life, minus Gmail, mostly due to not wanting to go through all the trouble of transferring everything over to a service that I end up not liking. I've heard ProtonMail is good, but other than that, I'm not sure.

I've used Fastmail for many years now, and I have nothing but good things to say about the service. In particular, it's insane how well notifications work in Fastmail, especially compared to Gmail (which I use at work). (Honestly, you'd think Fastmail was the giant multi-billion-dollar super-advanced tech company, if you look at the quality of their email experience vs. Gmail.)

However, some folks are a little spooked by the privacy implications of it being ran out of Australia, so be sure to research that if you're interested in Fastmail.

I'm leaning toward Tutanota, but I can't claim to have experience with them yet.

Proton has appeared somewhat bumpy to me -- I can't say for sure why, but they give me some spidey tingles.

Migrating / transferring is indeed a problem. I would suggest using Google Takeout, their data export tool, and permanently archiving your data with a third party service and / or physical backups. See https://takeout.google.com/. You probably won't be able to import into your new provider.

Thanks for this, I didn't realize you can actually export all your data from Google.

If you own a domain through Gandi (possibly others) you get email included for free. This is what I use. I don’t know why I don’t see this recommended more often?

I would do this through bluehost but it just straight up doesn't work. No matter the tutorials followed or time spent with support, I can't receive emails at my domain.

Migadu is great because they bill based on emails sent rather than billing by how many accounts or domains you have.

ProtonMail isn’t supported by iOS’s default mail app, which for me personally meant I hardly used it. It’s just too much hassle for everyday email.

I ended up trying fastmail and found the transition much easier.

I used to work at Bing. If you really want Bing to improve, the best thing you can do is just use it: clicks on search results, plus backs and dwell times, are vital training data.

Ideally you could use Bing as your default engine, then fall back to Google whenever there's a search that doesn't yield good results. If you have the time, you can also use the Feedback link on the bottom-right of the page to report bad search results; people do actually triage and read those.

>Ideally you could use Bing as your default engine, then fall back to Google whenever there's a search that doesn't yield good results.

This is what I do with DDG. Unless I have a pretty specific search, or want Google's really nice live sports scores widgets, DDG is usually pretty good.

DDG doesn't track what you click on, so unfortunately you are not contributing much to improving its and Bing's search quality by using it instead of Bing.

> clicks on search results, plus backs and dwell times, are vital training data

Interesting. Can you explain more on what is tracked after a user clicks a result and how this data is used?

(Disclaimer: as I said, I don't work at Bing anymore. Things may have changed in the intervening period and I could be mistaken or plain wrong about things)

Obviously ranking search results is immensely complicated with lots of accumulated insider wisdom, but the short answer is that "static" page features (PageRank, how well the query matches text on the page, etc.) are good for pruning down zillions of pages to generate a candidate list of 10 or so results to show the user, but not great at ordering those results, and serving the results in the best order is really important for user satisfaction. For generating the final top-10 ordering, "dynamic" features like how many people clicked that link, how long did they spend on the page, etc. are the most useful.

Regarding what is tracked, obviously Bing can only track what you're doing on Bing itself, so clicks on blue links and backs are tracked, but they can't see what you do on the destination page. This data is used to train models/NNs/etc. but the raw click data isn't kept for very long: it's too large and it has to be removed for GDPR compliance.

> For generating the final top-10 ordering, "dynamic" features like how many people clicked that link, how long did they spend on the page, etc. are the most useful.

Do Google do this as well?

I've never worked there and can't say for sure, but I have to assume that the broad strokes of their search engine are the same as Bing's. Bear in mind my response above was a very very high-level description of a really complex system with tens of thousands of person-years of accumulated knowledge and experience inside it.

Thanks for posting, it's nice to hear from someone on the inside!

Downvoters: why on earth are you downvoting Analemma's post? It's constructive, on point and overall the kind of post which makes HN comments valuable.

I still don’t understand the hive mind downvote patterns.

What I'm about to say will probably get downvoted as well, but I've noticed HN getting more polarized in terms of votes in the last maybe 2 years.

Because of the way that downvoting works on HN, the few times I use it, my thought is "nobody should see this comment". Because enough downvotes delegitemize and hide a comment, it's effectively telling other people that it's not worth being read.

I don't know if that's actually going through people's minds when they downvote, but that's what I think which is exactly why I rarely use it. But too many people now are simply using it on posts they don't like, even if a post deserves to be there. For some reason, it's an unpopular opinion to believe that an idea or opinion you think is wrong deserves to be seen and rebutted rather than silenced.

Downvotes for dislike is supported by HN/pg. Their choice of fading out comments, when combined with downvotes-for-opposition is really terrible for a high quality discussion site. Just goes to show the community around a site makes all the difference.

> I used to work at Bing. If you really want Bing to improve, the best thing you can do is just use it: clicks on search results, plus backs and dwell times, are vital training data.

A million flies are attracted to shit. I don't believe this sort of training data will ever become useful if it's in the same pool with the rest of the world. See also: voting with your wallet against the tyranny of the majority of uninformed consumers who buy whatever is most marketed. Those pennies don't matter.

In fact, I believe this sort of training and optimization for the mainstream plays a role in allowing bad results to proliferate.

This is before we even consider the fact that clicking on many results can indicate that they're bad (I click another result because the previous wasn't good), or because they're good (I'm browsing choices). Dwelling long can be bad (crappy & slow site, it takes me long to find the information I want or turn away) or it can be good (I found good stuff and I'm spending a while on it). Whatever conclusion your training system draws might be completely wrong. And probably prone to being gamed.

The above heuristic arguments seem inherently weaker than the direct experience of someone who used to be on the team that improves the results. They are well-funded and should be able to back out the effects you mention.

How are they ever going to know whether they improved results for me or not? You might as well train AI to play a game without ever checking their score. Oh, it's spending 20% more time in each room now and firing fewer bullets than before. Surely it is a better AI now.

Sorry, appeal to authority is no argument. Direct experience is valuable if there's an argument or some real scenario we can dissect, otherwise it's nothing more than a baseless claim. Without concrete examples, it's not even an anecdote. And there are plenty of anecdotes about search results for in-depth content becoming harder and harder to find.

The person I was responding to posted that they're using this training data to order the top 10 results or so. That's already an indication that it's not very helpful for me. I don't get frustrated if the top few results are in suboptimal order, I get frustrated when I get pages and pages of garbage and irrelevant results and can't seem to get anything useful out of it.

See the replies elsewhere. The metrics go beyond simple, naive "click counting."

Specifically they also measure metrics like "regret" - if you went back to the search results and chose something else this indicates it wasn't a useful search result for the query and this counts against it.

In theory this is quite a nice defense against clickbaity, content free results.

> Specifically they also measure metrics like "regret"

I explicitly addressed this:

> clicking on many results can indicate that they're bad (I click another result because the previous wasn't good), or because they're good (I'm browsing choices).

You don't know the reason why I clicked back. Maybe I got what I needed at a glance, and I'm browsing to the next thing to see if there's more of what I need.

You don't know that the Average Joe (and a million other Average Joes) wasted 15 minutes on clickbaity content free crap despite not getting anything useful out of it. You know, those same SEO spam sites are trying to find ways to make the user stay as long as possible; they are in the game. And I'm sure there are plenty of gullible users who will stay. Who knows, maybe SEO spam is working and that's precisely why we see SEO spam sites in results?

Again, I think training like this is pretending the problem is much simpler than it is. And I think that there is a real danger that training like this just optimizes for 1) average users 2) sites that managed to game the system with their SEO. Neither is optimizing for quality results.

That theory has so many holes and pitfalls.

I just think you're assuming these search engines are using way fewer factors than they actually use. It's probably exceptionally easy to tie your search behavior to someone else who's much more like you than "the average", unless you're using a search engine like DDG, Qwant, etc.

Maybe the best bet is for a search engine to try to include a simple, non-intrusive method of getting user input to determine the usefulness of a search result. Like, if it determines you came back as a "regret", maybe a ThumbsUp/ThumbsDown icon next to the last link you clicked to get you to say whether that link was useful in any way, or maybe a "block" option as well to say that you never want sites like that again.

> I just think you're assuming these search engines are using way fewer factors than they actually use. It's probably exceptionally easy to tie your search behavior to someone else who's much more like you than "the average", unless you're using a search engine like DDG, Qwant, etc.

I should, then, expect dramatically different results when I use someone else's PC or a public computer. I have not witnessed dramatic differences.

It's just you pretending search engines are simpler then they are. Noone's just ranking most clicked sites first - that's too naive and encourages clickbait. Clicks are an ingredient in the overall ranking system, which is generally also trained on relevancy labels by human annotators/raters/assessors. These human labels are far more reliable and can penalize all the bad clickbaity stuff which clicks can't, but unfortunately also very expensive to obtain and thus low volume.

We may not know exactly why you clicked back, or why Joe wasted 15 minutes on a site, but from all clicks in aggregate and from the human-labelled data, we can tell how that correlates with the clicked site's relevancy and quality and utilize your clicks accordingly to improve the ranking.

Finding such correlations is still by far and large what all search engines do, rather than trying to truly understand the query. It's only now starting to change with recent advances in deep learning and NLP.

A lot of this does, in fact, get solved through machine learning. If one search query is conducted a hundred times by different people with otherwise similar profiles, the machine learning can constantly modify the results until the hundredth person sees the best results first. For example, if the first 25 people all click on the 2nd, 4th, 5th, and 8th result and then stop, it can modify the order of the results for the next 25 people, and randomize the order, so that some see 8-4-2-5 first, and others see 2-5-8-4, etc. The it re-assesses the results and gets smarter.

The gaming part is where there's a lot of importance, since it forces the machine to learn which users are acting nefariously.

Shuffling the top 10 results is not fixing the big problem in search for me. I don't care too much whether the company I search for is in first or fourth place. But I'm really irritated when I get pages upon pages of useless results and nothing remotely relevant.

Clicks are noisy and have their problems (especially clickbait), but as far as ranking goes they still blow out of water everything else you can use. Analemma_ is right on the money. We need more Lemon Pledge^W^W clicks!

Speaking as someone who worked for another big web search engine in the past (not G, not B)

And counting good/bad clicks for a given (query, url) pair is just the tip of the iceberg. There's a lot of other interesting stuff you can do with them _if_ you have the data. Deep learning/NLP with clicks as training signal is probably the most exciting area to me, to name one. Unfortunately almost all of the data currently gets nabbed by google, other search engines are just getting by with the scraps. And it's very hard to bootstrap a competitor from scratch - for example Cliqz had to resort to some shady deals with Mozilla just to get any data to start with.

Yandex does that. At the bottom of search results page there are links to other search engines. So if you aren't satisfied with results, you're one click away from Google. I believe they track these clicks and know which results are bad.

> We cannot, as a healthy internet, let Google control so much of the web.

It's getting worse too.

Back in the day, you could Google something for a manufacturer and include the name of the manufacturer like, "Pioneer 10" Subwoofer" and automatically the first result would be a like to Pioneer's subwoofer page or their main ecommerce site.

You type that into Google today? You will get 15 results for AMAZON pages with Pioneer speakers. No, I want to buy it from Pioneer, not Amazon. Oh yeah? The actual link to the actual company, who actually makes those speakers? They're on Page 2.

When you have the actual manufacturer being buried in the results, we have a major problem.

IMO Google is doing the best they can to serve you relevant results, just that everyone and their uncle is spamming internet to get up in the ranking game.

Think of modern google as a question answering machine not a serch engine. And do "Show me Pioneer 10 subwoofer manufacturer specs" instead of just jamming keywords and hoping for the best.

> IMO Google is doing the best they can to serve you relevant results

Maybe, but if so then they're singularly bad at it. Google consistently gives me worse results than DDG and Bing, anyway.

I really think it's because Google is trying to figure out what I "really mean" by my search queries, and it's really terrible at guessing.

I'm genuinely asking: can you share an example where the Google results are worse in your opinion than DDG or Bing? Ideally with a screenshot in case there is some personalization going on. I just want to see a very clear example of it so I can try it out for myself.

What you really mean in a google search is you want to buy something. You only think you don't want to buy. It doesn't matter that you've lost the leaflet for your Pioneer 10 Subwoofer and want to remember which wire goes where in your new house. Or just want to remember some detail of something you had years ago. Google is there to assure you that you really want to buy a shiny new set.

Not buying stuff you already own again serves neither the economy nor Google's advertising profit. Go on, buy another set. Infinite growth depends on it.

That really is their attitude to every product and service.

because you are being swamped by the order of magnitude more people that want to find a price or buy it.

I work at Google (not Search though sorry), opinions are my own.

> When you have the actual manufacturer being buried in the results, we have a major problem.

How do you know though? What if the majority users actually do what you describe in hopes of buying something and thus the shopping results are more relevant for them?

I imagine either there is a bug or this is the case, because I'm sure the links people actually click feed back into the algorithms and the results are modified accordingly.

This being said, it seems like the fact that your results aren't personalized enough to your liking is a shortcoming, assuming you were signed in.

It's a self-fulfilling prophecy. Third-party shopping results are shown on page one and the manufacturer's store is on page two, and people rarely go beyond page one so naturally they will click the Amazon results. This in turn feeds the algorithm and reinforces its original assumption.

> Bing/DuckDuckGo have the article at around 7th or 8th place,

You should take a closer look at DDG, it has the answer from serverfault.com in an instant answers box for your query and it highlights the required setting in a perfectly chosen excerpt from the correct answer:

  As of CentOS 7.2 tuned now has a no-daemon mode which  
  can be turned on by setting daemon = 0 in 
  /etc/tuned/tuned-main.conf. This is mentioned in the 
  RedHat Performance Tuning Guide.

Weird, I do not get that when running the query on DDG.

Do you have the URL?

I can see the excerpt too. Try this:

https://duckduckgo.com/?q=centos+7+tuned+no+daemon&t=iphone&...

> I wish there was a way to directly help Bing/DuckDuckGo/Yandex improve their search results.

I am amazed that no search engine gives me easy way to blacklist domains from my results. That would make the usefulness of any of them to increase by orders of magnitude. (and if they are careful, they might be even able to use the blacklisted data to adjust their general results , not only my personal results as well)

And I can't help to add qwant.com to the list of alternative search engines. No affiliation whatsoever, but I am pretty happy to use that as my daily search engine here in Europe.

Google used to have that, but of course they stopped it. There are browser addons that have he same effect though.

Reference for the curious: https://github.com/codyogden/killedbygoogle/issues/674

> Google used to have that, but of course they stopped it. There are browser addons that have he same effect though.

Yep. Ironically enough, that was killed soon after I found about that. And anyways, it was implemented far from easy to use. It should be just behind the green arrow that gives the options to see cached version etc. And obviously with cross-platform effect on all my devices as long as I am logged in.

I would pay a subscription fee to my search engine to implement this feature. Currently use duckduckgo with a add-on to filter blacklisted sites but would love for this to be integrated.

In my opinion, the results on DDG et al are a matter of users being trained by Google knowing too much about them. If you search "Django" on Google and get relevant results, it's because Google knows you. On DDG you need to search "Django framework".

To me it is not so much result quality as integration. Currently, with Google no matter what I am searching for, I just type it into my location bar and hit enter. Whether I am looking for a new bar to check out on a Friday night or what does a specific compile error mean in Haskell. So in the first case, I will get a map with bars around me and in the second case I will get a link to stackoverflow. With DDG these become functionally separate. To do the Haskell error search I do the same thing, but to do a local bar search I have to open a separate tab, go to Google and do a search there. Same goes for looking for things like theater plays (Google will give me reviews, showtimes and a link to buy tickets all right there at the top of my search page), address or place name searches (map, directions, open hours and website link all come up right away) etc. The only thing I can do with DDG is the old fashioned "find me links relevant to these keywords" searches.

Feels like a perfect opportunity for you to try out search engine keywords. On Chrome, it's under Search Engine > Manage Search Engines. This allows you to type (keyword) -tab- (query). For example, my work computer (on which I can only use Chrome or Edge) is force-defaulted to Google, but I have Bing set to keyword "b" and DDG keyworded to "d", so I just type:

b(tab) Whatever I want to search for

and it goes to Bing with that query. It's a great example of what you're looking for. On my personal devices, I try to use Bing for as much as I can to see whether its personalized results will ever marry up to Google (it's gotten extremely close, lately), but I often prefer searching Google Maps, so I have gMaps added as a search engine with keyword "m", so I can search for anything with "m (tab) place" and immediately see gMaps results.

It's very easy to add new engines to, just take the query URL from any site, pop in %s in place of the query, choose a keyword (same on FF).

I have about 50 keywords.

> Google I can bang in cryptic queries like > centos 7 tuned no daemon

Not only can you bang in whatever query you want to DDG, you can bang it in via Google, using bangs (https://duckduckgo.com/bang). I think that "!g <query>" is the second stop for many DDG'ers, when DDG itself disappoints. (My understanding is that it still offers some anonymisation over searching directly through Google, but I'm not sure.)

I have noticed that when trying to find the most recent solution Google shows me an answer that is 10 years old.

ex. Adobe updates come faster than I can update, so I run into issues, so finding the most recent answer is what I am looking for and not a similar issue from Adobe pre CC.

I find that a lot too. Trying to search for anything Rails related always shows Rails 3/4 results before Rails 5/6. If Google really knew that much about me, they should notice I always end my searching by clicking on a Rails 5/6 result (because those are the versions I use) and every time I land on a Rails 3 or 4 result I almost always click back and try again. Same with Python when I get 2.x results, with all their data they should know I've used 3.x for years.

If companies want to track me and build profiles on me, I accept that I can't stop that without refusing to use their products and every site that's integrated with it (Google Analytics, Adwords, etc). But if they're going to do it, at least use that information to help me. What's the point otherwise?

You click back.

That means you see a second helping of ads.

You still use Google.

From their perspective this is probably working perfectly.

or at least filter out results that are irrelevant. They are lacking the ability to remove dated results when you are looking for the most recent.

I know you can manually filter results, but why show 10 year old answers to software that gets updated so frequent! I am sure there are other areas where this exists, dated results should be pushed back a few pages.

Do they factor in the number of clicks on each result? If so then it would presumably improve over time.

I know Google also factors in how long people spend on the resulting webpage.

Yes, they do factor in number of clicks. They also consider "regret" -- returning to the search list and picking a different result.

I can pretty much only use Google in verbatim mode now

My latest annoyance, nevermind them showing pages that don't have "quoted" keywords. They rewrite the search query in the textbox and remove the quotes. Just when you thought it couldn't get more annoying.

Oh yeah that kills me I often have one word that is the important one and instead they show me the low-value responses that do not contain that one differentiating word that is the reason I'm doing a search in the first place

Does verbatim mode actually work for you? It doesn't seem to for me.

yeah there is no verbatim mode any longer

Do you have a region set? I got an answer on the right from stackoverflow (don't know if it's the correct answer.)

Reply on news.ycombinator.com