Originally published at: https://boingboing.net/2018/10/12/bing-search-results-are-a-sewe.html
…
I tried to replicate and it looks like MS fixed it pretty fast. I’m not getting any of this. (Of course, now my search history is full of Jews" and “black people are.”)
Isn’t this in part the result of assholes using Bing instead of Google? Or at least the result of Bing not being in the critical spotlight.
Interestingly enough a Google image search for “jews” isn’t terribly kind either:
It’s probably that so few people use Bing that it’s easy to skew the results.
I thing we should highlight the importance of the end of the article
" it is instructive to see such a huge-scale mirror image of what … we are are searching for."
Getting Bing or Google to alter their algorithms specifically so that I don’t see this garbage is nice. Doing so can interrupt the virality of the ideologies involved, but…
But you know what would be more valuable? Actually knowing that these things are trending and being able to confront these things in the PUBLIC sphere.
“Nazis were bad, mmmkay” produces a whole different reaction than the visceral realization that you should have said “Nazis ARE bad”
Also, and I just…it’s Bing so it is ok, but are we criticizing a search engine for showing people things they are actually looking for? That is their job. And search recommendations are a modern tool used to further that goal.
I don’t want to be in a situation where information is being hidden because of ideology. This isn’t a situation where I couldn’t let my nephew search for the my Little pony movie.
In fact, by uncovering through merely accurate search the connections between things that seem innocuous and the place those things hold in a community can interrupt the process of recruitment of some of these groups by revealing the ends before you’ve invested enough into the process to alter your psychology.
Imagine a world where you aren’t allowed to know that the neo-nazis made Pepe the frog into their mascot for a while…but you are allowed to accidently use your kid as a signal to other people that you are down for a little racial purity when you search for Pepe memes to print onto your son’s lunchbox.
Do you work for Microsoft?
Bad-a Bing!
This isn’t an issue of showing people what they are looking for. It’s an issue of prompting them on what they should look for next. A person searching for Michelle Obama isn’t served by the search engine showing, as one of its first recommended searches, a racist conspiracy theory rather than something factual. If they were actively searching for that racist conspiracy, then yes, the engine should bring that up (though again, ideally showing factually accurate pages above those that promote the conspiracy). But our tools shouldn’t actively be egging people further into these sorts of delusions.
I assume that’s got to be a factor. I’ve noticed for a while that the alt right has a weird preference for bing. Seems to be a weird (but expected) combination of its ability to find porn and “google is biased” nonsense. Gotta figure if a largish group of heavy internet users is focused on bing. Both using it frequently, and doing SEO junk specifically targeting it its gonna mess with the algorithms.
No, I don’t think that’s the criticism at all. I don’t think anyone would complain if a search for “racist things about black people” turns up racist things about black people as the primary result. But if a generic search about “black people” leads predominately to racist shit, that’s a problem. Right?
God, if there was a machine I could be plugged into to accomplish just that thing, I’d gladly pay for it.
Exactly. Garbage in, garbage out.
. “But if a generic search about “black people” leads predominately to racist shit, that’s a problem. Right?”
Yes that would be a problem. That would be big problem. I’m not about to go use Bing to check, but I thought that the author was describing that they followed a series of suggested searches to find the bad stuff. Since suggested searches are (I think) based on search sequences done by past users, I’m not sure I have a problem with it.
I’m ideologically prone to believing that more unbiased factual information improves outcomes, so I feel like having a public reflection of searches based on a sociologically representative sample is always good.
Discourse is important, politeness and civility less so. We can’t have conversations if things are never brought up in the public sphere. Do those conversations involve broken noses? Sometimes, but hey, it’s worked for 20,000 years
When I try to replicate the results for search hints I don’t get any of the stuff that guy got. I suppose it’s possible that microsoft has applied an insta-fix. But there’s another more interesting possibility.
An obvious methodology problem: Search results are tailored to what you have searched for previously.
The researcher who originally published this result says that he had previously been searching for horrible stuff on the internet. Is it possible that he contaminated his search shortcuts with searches that he had previously performed directly? Or more damningly, that the enormous variety of search shortcuts that he got to underage porn reveal something about searches that he had performed for personal reasons?
Well, I’d argue that it hasn’t yet managed to go so completely wrong that everyone is dead. Yet, anyway. That’s a pretty low bar for ‘working’.
One of the central problems we’ve been having with search companies and social media is that they tend to prioritise the biased, inacurate information. Promoting it to the top of results, distributing it faster and further. And putting it in the suggested search results. Because they operate on non-curated algorithms, they’re relatively easy to manipulate. And fundementally based around maximizing ad revenue, and built on things user input and link volume as opposed to anything like an objective standard for things like unbiased and accurate. Or even usefulness.
And that’s exactly what the complaint here is. Biased, inaccurate, information is being given priority over actual information. Because the system for determining priority and validity is borked.
I thought the central complaint was a personwell versed in the topics knowingly clicked on a series of suggested results and found the terrible things he was looking for.
The bar is that the process brought us from scattered groups of Hunter gatherers to the vast interconnected civilization we have today. I am often myopic enough to call that a low bar, but not today.
The bubbling of search results (and the association of those searches with your MAC address, your IP address, and/or your personal profile) is a legit problem for researchers. DuckDuckGo and creating new/clean profiles for particular research projects are only partial solutions for an issue I struggle with all the time.