[alopecoid]

Google can and should customize search results based on location, and it's not just about "local articles" as the article suggests. If enough people from the same general location click on certain results more frequently, then those results should rank higher for others who search from that same general location.

We use Google because the results are useful, not because they are "unbiased". Ranking implies some sort of "bias" and is what makes search results generally useful. We don't want a search engine that does nothing clever and just spits back unranked results. Otherwise, we would be inundated with results containing credit card scams, porn, Bitcoin scams, Viagra ads, etc, when we search for... pretty much anything.

In privacy (incognito and not logged in) mode, all of the above still applies. What would NOT apply is something like: You are a vegetarian and suddenly all of your restaurant searches rank vegetarian restaurants higher in results while in privacy mode. Unless, of course, for some reason people in your general location happen to mostly eat vegetarian.

In any case, if people don't like it, stop using Google and go use some other search engine; there is absolutely nothing holding you back. More times than not, I think people will switch back to Google because they find the results more useful, even in privacy mode.

[kiriakasis]

> I think people will switch back to Google because they find the results more useful, even in privacy mode.

I now use duckduckgo as default search engine and my experience is mixed.

The problem with google is that sometime you search for something new and then you see the bubble very clearly, which applies non only to search but also to youtube (maybe even more).

The problem with duckduckgo is that you are searching for something specific or something you saw months ago and don't remember well then google's index and tracking can be useful.

[rayvy]

> The problem with duckduckgo is that you are searching for something specific or something you saw months ago and don't remember well then google's index and tracking can be useful.

At this point I don't treat search engines as some sort of dichotomy (Google or DDG or Edge, etc). Rather, I try to use them as a nice blend : Google for when I'm throwing darts at the dartboard and have no idea what I'm looking for, DDG for when I know exactly what I'm looking for (to the point where I can type in the url), so on and so forth.

There's absolutely nothing wrong with using multiple search platforms. Obviously Google is great for when you don't really quite know what you're looking for, but if I want to read Deadspin, typing "deadspin.com" into Google will be the exact same experience on DDG.

Seeing as how most people visit the same websites over and over again, it doesn't make since to just have 1 single search engine (e.g., a Google).

[dhimes]

I use ddg as my primary, and one thing it's brilliant at is Mozilla Dev Network. I append my question with mdn (not !mdn- the bang uses mozilla's search and it's not as good) and I usually get my answer at the top. DDG is so reliable now that when I occasionally do have trouble in, say, a non-mdn related search) I fool around a while before I remember that I can try google.

[Larrikin]

Sounds like it would be nice to have a resurrection of Dogpile with a combination of Google and DDG.

[IIsi50MHz]

searx offers this, has multiple independent servers, and you can even run your own if you don't trust any of them not to aggregate/sell your search data. One is searx.me (which is almost like telling someooe "Search me!" when you mean "I dunno.").

Startpage is basic'ly Google results, but the filter bubble is "all Startpage users". Also brings back some of the search operators that Google disabled.

Qwant is a European search engine that brags about privacy, but the results are hit-and-miss for me so far...haven't used it much.

[striking]

You can use the !g bang to search Google from ddg.

[GiuseppaAcciaio]

I keep on trying to love duckduckgo but find that often even typing the exact title of an article I'm looking for, it's not on the first page of results...

[ergothus]

I have Google routinely deciding that I didn't mean to use ALL three words I searched for and "helpfully" dropping them. Sure, I can tell it "no, I really want those", but the experience is definitely becoming more and more sub-par for me. As bad is when a search doesn't give me what I want, so I narrow it, only to find that Google uses my previous search to decide what I want to see so I still end up finding similar results.

I remember when Google blew us away with Page Rank (goodbye Alta Vista!), but in the last few years Google has gotten so good on providing entry-level information that it's useless for finding specifics, so I expect the next Big Thing in search to come along, though I have no idea how far out it is.

[kibwen]

Fun example of this: last week I was trying to figure out all the floating point operations that can produce NaN. Go ahead and try searching Google for "ways to make nan"; it's going to show you dozens of pages of naan recipes, and there isn't even a link to click to make it actually search for what you've typed (instead there's a link for Did you mean "ways to make naan"?, which shows a different set of naan recipes).

[noxToken]

Although the tailored results have been useful, I think I still like the days back when you needed search operators. I was once looking up stuff about electrons (the particle). A plain query of electron only returned results for the framework on my first page. Understandable.

The other annoyance is the lack of Wikipedia results. For a general topic, I like to have a few pages to chooses from about the topic in addition to Wikipedia. Rarely are Wikipedia results in my organic listing unless I specifically add wiki or Wikipedia.

By the way, this[0] is how to search your query.

[0]: https://www.google.com/search?q=ways+to+make+%22nan%22+-%22n...

[dhimes]

Interesting. Wikipedia what my top result when I DDG'd "floating point operations that produce NaN"

[mda]

Did you try your own sentence "floating point operations that can produce NaN." "Making nan"is a very tricky query, because it has no context that is related to floating point. I think it is not fair to expect anything else. If you give a slight context like "operations that make nan" you will get results.

[barbecue_sauce]

Yeah, "ways to make nan", even being familiar with the domain, is an extremely unintuitive way to phrase that query.

[millstone]

Holy cow that's bad. It says: Showing results for ways to make naan. Search instead for 'ways to make nan'.

Fair enough, so click on the Search Instead and it says "Did you mean: ways to make naan" and then shows a bunch of links about breadmaking anyways.

[Drdrdrq]

You can use: ways to make +"nan"

Disclaimer: didn't test it, but I often use this trick to force G to use some word, and to use it exactly as written.

[dwd]

Google dropped support for + years ago but you could try using -naan and "nan".

https://support.google.com/websearch/answer/2466433

[sanowski]

a search for your question "all the floating point operations that can produce NaN" gave useful results for me.

[ape4]

Its too bad it ignores the case of NaN (and other queries)

[00N8]

idk, but I think they employ multiple tokenizing strategies, including case insensitive, case sensitive,by words, by n-grams of letters...

[hedora]

Can you post an example? I’ve been using DDG for years, and have never encountered this.

Also, the DDG devs are certainly lurking on this thread, and can fix the class of queries in question.

[Kiro]

No, they can't since DDG is a wrapper for Bing.

[therein]

I have been using DDG primarily ever since Google tried to pull that login shenanigan on Chrome users. Switched to FF and DDG and with the occasional !g, I'm content. Try !b and tell me it's the same as DDG's results.

[Kiro]

https://duck.co/help/results/sources

> In fact, DuckDuckGo gets its results from over four hundred sources. These include hundreds of vertical sources delivering niche Instant Answers, DuckDuckBot (our crawler) and crowd-sourced sites (like Wikipedia, stored in our answer indexes). We also of course have more traditional links in the search results, which we also source from a variety of partners, including Oath (formerly Yahoo) and Bing.

What this means is that they use 400 sources for things like Instant Answers and other widgets but Yahoo and Bing for all their organic search results.

[mda]

I think they sometimes mix yandex, yahoo and other stuff as well. There is a reason they never disclose what percentage of queries are served from which source, kinda spoils their magic i guess. Kudos to bing team though, they seem to improved quite a bit apparently.

[i_am_proteus]

This happens to me sometimes, and is why "!g" is DDG's killer app.

[JumpCrisscross]

I use DuckDuckGo as my main search engine. One place it drops the ball is in searching for anything health related. Top of the page: homeopathy, conspiracy theories and supplement salesmen.

[yourbandsucks]

Youtube in particular is terrible for filter bubbles.

In the name of optimizing for 'engagement', my youtube recs are full of politically polarized clickbait. They're not merely reinforcing my existing beliefs, they're actively trying to push me into a bubble.

At least facebook has the excuse of actual people pushing this stuff.

[slededit]

I’ve noticed this too. I’ve never clicked any of them yet they keep getting recommended.

My other favorite example is Netflix and WWII docs/movies. Watch just one and forever onward they will be half your recommendations.

[random878]

I tried duckduckgo, but wasn't pleased with the results. I switched to startpage which has been very good.

[shanty]

This is all well and good for information retrieval, but people are making decisions based on information from search results.

In your vegetarian example, what if 51% of people were vegetarian in an area, and the general population was making decisions off these "localized" search results. We would likely expect that this would influence the minority to the tastes of the majority.

This might be fine for something like vegetarianism, but what about other topics? Should your search results be more racist because you live around a lot of racists? This is best case.

I have tangentially worked with groups that specifically utilize this to provide public opinion sway and consumer capture for their clients.

[alopecoid]

A contrived example to clarify further: Let's say that there is a link, foo.gov/taxes_in_retirement. Locations with a high concentration of retirees might click on the link more frequently compared to other locations. In privacy mode, from this location, a search for "taxes" might rank that link higher in results based on this activity (even though the search didn't contain the word "retirement"). This shows how a link might be ranked differently depending on search location even if the link itself is not inherently location-specific (as oppressed to, say, a local-news link).

Also, search engines can play with inconsistent ranking of results to see how click-throughs might be affected. For example, if moving a link from first to third in the result list has no effect (people continue clicking on the same link even though it's now third instead of first), then it's a pretty strong signal that the link should continue to be ranked first in future results. This experimentation of search results is even more important the more uncommon a search is because there is less confidence in the current ranking until there is more activity to base the ranking on.

Just as stores shift around product placement (front of the store, back of the store, etc), a search engine is free to shift around search results. Keep in mind that product producers might pay for better in-store product placement too, just as customers pay search engines for ad placement in search results.

[rlucas]

You can't just handwave "useful" as an explanation for why location-based SERP discrimination is desirable.

It is "useful" for me to be on the phone with someone in Cleveland and describe how to find something on the Web, expecting that they can follow a similar set of steps at a similar time and get a similar result.

A (sort-of-)deterministic Web can be good and useful. It is a very strong statement of preference and exercise of power to declare that "useful" results must be meaningfully different based on the characteristics of the individual searching.

For whom is that exercise of power most beneficial? I would argue that a rapidly shifting, slippery, personally-dependent presentation of the world's information is extremely useful as a tool of control, but gives only occasional and relatively marginal benefit to individual searchers.

The 2016 US election is a big case in point. Personalizing information delivery, when coupled with asymmetric processing power and data availability, lets you have situations where an atomized polity winds up seeing what suits each individual, but with a radically degraded ability to form collective truths or consensus.

The definition of "useful" is an exercise of power.

[hrktb]

> Google can and should customize search results based on location

I feel more and more these ideas of optimizing for 95% of the use cases give good result on paper but shitty lives for the 5% left.

I understand the good intentions behind that calculation, because making life easier for a huge majority of people should be a good thing.

But for instance boosting local results is one of the way you’ll make people often searching for foreign information miserable. Searching for remote places will most of time be met by random local businesses first. Web based international content will be outranked by local content, and your local newspaper bitching about heat waves when it’s just summer will outrank by far rock bands and manga titles.

Sometimes that’s the wanted behaviors, but for instance currently Google already works with strong preference for localised search, and that’s one of the things that pushed me to DDG.

In a way if Google wasn’t so massively successful I’d root for them to better serve mainstream searches. But in the position they are now I think it’s harder to say they should just care about the vast majority of people. Even 1% of their userbase is an incredibly huge number.

[GordonS]

I completely agree, and often that's how I want Google to work.

But what I would also like is a way to search without using my context, as sometimes I want results that aren't related to my location etc.

[reaperducer]

Yep. Like when I'm using a VPN, but I still want to know about stuff where I am physically located.

[hedora]

For your vegetarian example:

The article runs through the analysis you propose, and (within the limitations of the study) show that Google does apply very similar filter bubbling logged out in incognito mode vs logged in.

In fact in “anonymous” mode, the results are much more similar to the same person’s “logged in” mode than to other randomly chosen people’s logged in or logged out results.

[kgwxd]

What you're saying makes sense for location dependent results (e.g. searching a map of nearby places). For something like "origin of the universe" results would be very different across the world. What happens when I'm in a location I'm not usually in? Do I get the local results or the results for the place I usually reside?

[kiriakasis]

But their searches were (at least partially) political in nature, so a result of a national news site talking of a local news would be relevant. "gun control" is location dependent is there is a protest at your city

[kgwxd]

I guess it comes down to what a search engine should assume your reason for searching is based on what you enter. If the query doesn't strongly imply I'm looking for "news about X"; "local opinion on X"; or "closest place related to X", I'd like it to assume I'm doing "research on X" so I get results that do not take my arbitrary location into account.

[saalweachter]

How would you like Google to decide what the correct answer to [origin of the universe], [gun control] and [vegetarianism] is?

[throwawaylolx]

>Google can and should customize search results based on location

Should there not be an opt-out option though?

[alopecoid]

There is: Use something else.

I don't mean this in a snippy way, but truly. If it's that bothersome, why not try something else? It seems that most people instead think that they have the privilege to change the product to their desires.

[throwawaylolx]

> why not try something else?

Because the "something else" lacks other features. Are you suggesting that users shouldn't have "the privilege" to suggest new features or discuss already existing features?

[malloryerik]

Better than having to go to Privacy Mode (which can feel a little creepy with the spy icon and even just the name), it would be easy for Google to make a clear On/Off toggle for personalized vs raw results.

For the sake of education and not being evil (or "doing the right thing"), it would be nice to be able to view results from other typical profiles' points of view.

[thecleaner]

The arguments for personalisation make sense. But the other story is sometimes we need unbiased results. Like when reading about politics. You need the dose of "whats are the facts" rather than "What are the facts for me"

[Kye]

Replacing an algorithm that bombards people with propaganda with unfiltered propaganda doesn't seem like an improvement. Someone who, for example, mistakes a map of policing for a map of crime won't do well in either scenario.

[w323898]

The point of the article was that, even when there is no reason to do so because location and identity are identical, Google STILL gives differently ranked results. Please read more carefully.

And I've been using DDG for a bit now and have found it perfectly useable.

[alopecoid]

Sorry, but I think you are missing the point (please see my follow-up post). The point is, search results are not static even for the same inputs (search string, location, etc), even in privacy mode.

That's great that you use DDG and find it useful! If Google was a true monopoly, as the current media blitz would have you believe, then you would not have been able to so easily switch to DDG (or Bing, or...).

[LMYahooTFY]

Criteria for monopolies aren't static and unchanging. The switching cost for energy or telephony services is not what it was in 1950.

I can't say with confidence how far along Google is toward the threshold of "Monopoly" and have yet to hear an analysis that I would consider definitive in any way.

[hedora]

The point the rest of us are making is that even when you opt out, which google supposedly supports, it still bubbles the results based on your profile.

This is shady, at best. It probably contributes to the US’ current political instability (different propaganda / news in red states), and is also probably an unauthorized use of personal information in places like Europe that have laws about such things.

[iscro]

People wanted and used google precisely because it was "unbiased" and "unlocalized". It's why it became so popular.

The only reason for the biased localized results is due to corporate pressure from media and news industries.

Also, you are conflating localized and unbiased results with spam and scam. Nobody is saying google shouldn't remove scams and spam. People are just saying they want unbiased results.

[SquareWheel]

>People wanted and used google precisely because it was "unbiased" and "unlocalized". It's why it became so popular.

That's contrary to everything I know about Google's history, including the origins of Backrub and PageRank. Please provide citations in your comment.

[nofunsir]

Found the Google employee.

[sctb]

Could you please stop doing this? It breaks the guidelines.

> Please don't impute astroturfing or shillage. That degrades discussion and is usually mistaken. If you're worried about it, email us and we'll look at the data.

https://news.ycombinator.com/newsguidelines.html

[alopecoid]

You should see how I defend Facebook or Apple.

[mempko]

What's good for facebook and apple is good for google?