Women just love it when random internet dudes pull the "if I were a woman you couldn't point out how trite and pointless my complaint is, so let's just assume I am a woman and you shut up now" card. It's so cool for them to be the exemplars of the trite and the pointless.
To spell it out: it takes an increasing amount of effort to avoid any of your interests being tracked, logged, and almost immediately used in ways that you already know will be destructive. This is also an IT professional and a programmer saying this, so they do not want your advice. The fact that they can use many, many countermeasures which require many, many rabbitholes of research, that as a programmer they are able to understand (although they will have to be continually updated) is not a way to fix that fear it's the manifestation of that fear.
After 9/11 and the Patriot Act, librarians fought the government to keep the reading habits of patrons private as a core duty like doctors pledge to do no harm. Now it's difficult to get people to understand why clicking through to an attractive hammock while browsing Amazon registering in 900 databases and manifesting in ads and cold calls selling tropical vacations and being flagged on some government system as 2% more likely to be a flight risk if let out on bond (who am I kidding, AI doesn't give percentages, it just gives conclusions, looking at a hammock might be that stray pixel that can turn an OCR "O" into a "Q") - difficult to get some people to understand why that would make you nervous.
The answer to that isn't "use TOR to get to a VPN to browse amazon, and pay for that with a burner debit card loaded with bitcoin" or whatever works this week. There isn't really a hammock.
edit: also Amazon has that figured out, some researcher there figured out that they can identify you based on your click patterns and timing. Right now you can choose between 1) an app that will just click on everything silently, or 2) another app by a professor at some midwestern university that will jerk around the timing and positioning of your clicks in a way that throws your fingerprint off. The first app has been banned by every store, and triggers 80 warnings and a waiver that has to be signed with two-factor even if you manage to root your phone and sideload it. You have to compile the second app yourself, with a weird toolchain, and it draws in 640Mb of npm libraries. It's already been updated three times in response to Amazon's countermeasures, and the professor just wrote a paper about the entire method probably being ultimately doomed.
> figured out that they can identify you based on your click patterns and timing.
This isn't restricted to just Amazon either; as I understand it, this is the core of the current reCaptcha version (and even more so, its evil twin, reCaptcha V3, which is seemingly so effective it can rely on those mouse events exclusively in most cases)
My experience is that we reached a point where if DuckDuckGo can’t find something, then neither can Google. For local search Google is still a little better, but they choosen to drown out actual results with ads.
Not my experience for dev related searches. I assume this is because Google has learned I always search for JS stuff and I've gotten used to that so I'm not specific enough in my searches on DDG.
Have been using DDG for years, needed to call to Google maybe once per month in a particularly tough queries, on those I get about 50% success ratio - sometimes Google helps, sometimes doesn't. For the vast majority of routine queries DDG is more than enough.
I had a good laugh because that's actually a comment against the mansplaining common narrative, people will jump to explain things everytime they feel it's easy to add their contribute, being the audience a woman or a random guy on the internet or anyone they perceive weak enough to be a good target to show off their knowledge and skills
I think that’s a really cynical way to look at it, especially when the explanations are public. People can explain things to others without there being a hidden power dynamic.
How is an example in another context a "comment against the mansplaining common narrative"? Even your explanation makes it pretty clear that a man-to-woman combination would be one of the most common versions.
He's the one who extrapolated the context, my point is that the explainer does it for reasons that have nothing to do with the gender of the receiver. And this is one example
> Update: random dudes are now saying the unsolicited toxic replies are actually my fault for saying something in the first place. So NOW I feel like I've gotten a little glimpse of what it's like to be a woman on the internet.
I was very curious about how this is evidence that women specifically get berated on the internet rather than evidence that everyone is a potential target. Obviously from this interaction alone you can't tell if it happens more to women or not but... I thought it was kind of funny to say "this thing happened to me as a man and now I really understand that this thing happens to women."
The comment you're referring to didn't occur in the tweet about algorithms - it was a reactive comment that occurred in response to an experience that happened after the tweet. Also, I'm not sure what "to show what a nice guy he is" means here, though I'm guessing it's just being used an insult in the same family as "white knight".
I can't know whether his comment was reasonable without looking through the comments he was referring to (and neither can you), but even given that he was being unreasonable, you chose to be unreasonable in turn.
> I can't know whether his comment was reasonable without looking through the comments he was referring to (and neither can you)
Nah, I can. It seems quite toxic regardless of what has been said to him.
> even given that he was being unreasonable, you chose to be unreasonable in turn
I don’t see what is unreasonable about shaming anti-social behavior. When you see any “unsolicited advice”, even between two same gendered persons, as a manifestation of patriarchy you are bound to come off as toxic.
Is there a term for using Google services less to minimize your risk of inadvertently running into a ban happy algorithm? I realize the thing to do is not rely on their services, but that's much easier said than done
But to what extent would the algorithm know about you if you clear cookies/storage data though? I also imagine changing IP addresses help… does Google use IP as a source for profiling?
I can’t speak for what data points Google uses for tracking, however... with JavaScript, there is a plethora of information that can be used to uniquely identify you across IPs, and even across browsers. [0]
If you’re logged in to a service on multiple devices you have now linked multiple fingerprints to your account that can be compared against third party data.
It's annoying clicking some youtube link only to find it was something you really don't want more of, and then youtube proceed to recommend much more of it, even though you barely watched any of it.
kinda forces me to watch something else that is new but at least somewhat wanted in an attempt to make the algorithm look at the new shiny i'm interested in and forget the previous one.
i find it funny though when amazon still make recommendations to me based on a purchase in 1998. yes, Jeff, i still want to buy wrestling VHS tapes...
This will kind of clean up your current recommendations but it doesn't reset the recommendation algorithm like it used to. The algorithm still retains memory of your past behaviour.
For example if I consistently listen to 3 unrelated songs [A B C] on youtube together, the algorithm will regularly recommend them to me (because of my unique behaviour, not because they're similar). If I reset my history and then listen to song A, then B and C get immediately recommended even though they aren't similar to A and they don't exist in my listening history which means that the information about my listening/browsing habits is still there in the recommendation model.
If you turn off watch history is basically only suggests stuff based on subscriptions which is much better imo and I never need the history feature anyway.
I wish someone would make an extension that makes this a simple hover+keypress instead of multiple clicks. As it stands right now it is too much of a hassle to manually click 2 times per video. But if you do not then youtube will continue to plaster that video on your homescreen for weeks.
What if it's from a channel I like but don't want to watch that one video...will clicking on this reduce suggestions from this channel, will it do nothing, will it unsubscribe me from that channel(as has happened in the past), ...?
Nobody knows, I only know I don't want to play this game.
Use Firefox multi-account containers and the temporary containers add-on. Now your YouTube starts fresh with every new tab. It's actually nice, it means you get to build a new set of recommendations each time you open a tab, and because YT doesn't have that much data on you yet, the recommendations are directly linked to the thing you're currently watching, similar to how it used to be before the whole tracking extravaganza.
I'm in germany so i always get asinine videos recommended about what americans think about germany and what germans think about what americans think about germany
How Chrome and Chrome-based browsers work with profiles, and push the user to log in, I have a feeling they never fully log out and there's some kind of a supercookie hanging around. Trust issue.
The best way of not playing this game is to, well... not play it. Why should you pay any attention to what YouTube "suggests" you should watch? I just treat YouTube as a service that hosts videos, with a mediocre search feature, and I have no interest in what it thinks I should watch.
If a real human mentions a video that piques my interest, I may watch it. If I'm interested in finding a video on a particular topic, or a specific scene from this-or-that film or TV show, I will search for it myself. If am interested enough in a certain producer's content, I may "subscribe" to them, but I will be the judge if I want to watch the latest video they uploaded.
> Why should you pay any attention to what YouTube "suggests" you should watch?
There is one advantage: some obscure music videos. There are some rare pearls with the comments section almost exclusively thanking YT algorithm for taking them there. Happened to me so many times that I have a separate browser instance and a Google account for YT and I'm very careful what I click when I use it.
True, you do have a point there. I'll admit I have discovered a few bands/musicians by just leaving YouTube on autoplay, after having searched for specific artists I already knew to concentrate while working (before lo-fi hip hop was trending!).
I didn't even know there was a Jazz/Fusion rock scene in Japan in the 70's for instance. I never would have heard of this band without YouTube suggestions but really enjoying their work.
Not playing is even worse, I used to live near a less financially developed area of the city and youtube would promote dreadful videos either musically or societally if not logged in.
I did it for a couple of weeks to see what was being promoted by youtube algorithms in terms of content and ads to the more financially challenged parts of the society. Gangsta and sexually inuendo videos in neighbourhoods that although poor do not have a crime problem and young people don't dress like that. But it seems the algo is working on it. Note: This is in a European country
I’ve been trying it for a few months. I dutifully clicked away every preposterously stupid clickbait title or thumbnail and topic I didn’t care about. It had no noticeable effect.
It doesn't in my experience. I've been telling it I don't want to see: news, politics, music, or reaction videos for months. It still puts them there.
Last week it decided that I wanted to see rap videos and filled my feed with them. I don't listen to rap. I've never watched a rap video. I've told it I'm not interested in every single video, but it still puts them there.
Not sure it'll do much about news videos - Google seems to force the same presumably hand-picked selection of news videos onto everyone's front page, generally covering the kind of political topics tech workers think everyone should hear about.
Try going through your watch history/liked videos and deleting things if it’s really bothering you, I watched a bunch of political videos a few years ago and got burnt out so I deleted every one (took a few hours) and my recommendations instantly changed
Doesn't work well at all. There's a popular garbage news channel here in Brazil.
Every week I have to do that, it never works.
I never click on the videos, so no idea why they think I'm interested.
Yes, I always hesitate to mark YT videos as unwanted because often it's from a channel I otherwise like a lot and I have no idea how YT will behave after that. Considering the site is known for even auto-cancelling subscriptions sometimes it feels like a UX minefield.
YouTube's personalization recommendation is a nightmare. Just the other day, YouTube played "Billie eilish -copycat" for me literally after every other song.
I pretty much only YouTube in Incognito mode these days. Everytime I forget, YouTube manages to annoy me so much within half an hour that I switch to Incognito again.
It's frustrating that nowadays YouTube seems like it can only keep 3 or 4 types of videos that you like in its head. I feel like I'm watching the same thing over and over whenever I click on recommendations. I totally agree with you.
Exactly! Sometimes my recommendations are good, with long form tech and history content I actually want to watch. But click on one Top Gear clip and that's all out the window, and I have to actively recall what these channels were called and try to build them back into my recs.
There was a time a couple years ago I watched one episode of Arthur for nostalgia purposes and it completely obliterated my recommendations. I scrolled for hundreds of videos and didn't see a single non-arthur one. It took months for the site to become usable again.
I continually get suggested videos that are all already watched. I can't think of the last time I've gotten video suggestions that don't already have a red line at the bottom. At least it has stopped queueing up the same 3 videogame walkthrough videos I always click out of the minute they start.
About half of the videos Youtube throws at me are ones I've already watched. While I will occasionally rewatch something, it's nowhere near half the time. I thought that it was because I use Youtube without being logged in, but I guess not?
The fewer categories they use, the less manpower and computer power they spend on classifying and censoring content. They serve millions upon millions of queries a day with millisecond response times, and they don't get paid for providing value to you, so that's what we get. I hate what the internet is these days.
It’s better if you click that stuff, it causes the algorithm to recommend stuff you don’t care about, which means less YouTube holes to get sucked into
> amazon still make recommendations to me based on a purchase in 1998. yes, Jeff, i still want to buy wrestling VHS tapes...
It’s funny you mention that.
For many years I would get recommendations for Latino lgbt books. Weird as it’s not really my topic. I found a way to look at how they generate recommendations, and it turns out that I bought a book about a gay Chicano growing up for a college class in 1998 or something.
He found out he bought a book about “chicanos” which I assume is slang for a gay or lesbian Latino/Latina person which the user takes as confirmation that the purchase they made in 1998 is the explanation for why they are receiving recommendations for Hispanic LGBT literature now in 2021.
I don’t think the user meant to imply they had gained any special insight into the algorithm other then personal confirmation that purchase as far back as 1998 are still being incorporated.
If you spend a minute too long looking at the menu for a movie it will not only auto play, but for the foreseeable future Netflix will assume you’re watching it, want to finish watching it and watch similar stuff.
More than most other services Netflix doesn’t trust your ratings or your lists when making up recommendations.
It feels like Netflix knows what I like and throws things I don’t like at me on purpose, just to see whether I’ll change my mind. (So I unsubscribed)
The discovery is specifically what makes Youtube, Netflix or Spotify sticky beyond just content hosting, and even at suggestions, pirate websites are better at it.
You can disable autoplaying of next episodes in a series that you're watching, and you can disable autoplaying of previews, but I don't see any way to disable autoplaying of movies and shows in general when you're looking at their info pages.
Fundamentally, tools should serve their masters. When it's the other way around, it's time to find a different tool.
And yes, they had to get rid of ratings, because some people's feelings got hurt because they were being rated so low. Can't have that pesky objective reality rearing its ugly head.
I don't know how common this is, but I've gotten into the habit of creating "personas"-- different users that have different habits and are interested in different things, yet all me.
It's insane that I feel it helpful to take on "multiple personalities", but there it is.
Part of it is that these algorithms are fairly one-track. They can mix it up a bit, but it's always too much of one thing and too little of another. They can't truly comport with the reality that someone can have multiple interests and tastes.
another solution is to use 2 instances of chrome or firefox with different --user-data-dir which can run in parallel . It's actually very convenient for checking up stuff without 'messing' the main setup
I was being flippant, but isn't that true of anything though? How do you know Linux isn't secretly reporting all your activities to Big Penguin?
You should be able to trust a popular open-source project, because you trust that there are folks who will go in and take a sticky beak at these kinds of things. This why a lot of security folks prefer open-source software over close-sourced solutions.
Surely, if Chromium were cross-correlating profiles, then it'd be on the front page of Hacker News in short order.
can you use them in parallel? in any case i don't want any correlation between the two (other than the IP i guess). in my case i often log in to the same website from different accounts for testing purposes
You can you them in parallel and they share no extensions, sessions, website settings, or other local data. You can even have separate themes for each profile.
I'm unaware of any potential problems relating to fingerprinting and so on, but for your use case it doesn't sound like it'd be a problem.
Yes buts he’s code switching in order to fool an automated panopticon that’s integrated with our most advanced communications technology. That’s a bit unique.
It infuriates me that we basically had this as a feature, lost it, and have to sort of reimplement it with an almost illicit feel.
I can recall in the mod-late 1990s when Mainstream America was starting to go online for the first time, everyone was training their kids "never give out your real name." From that simple bit of stranger-danger paranoia, we built a lot of communities as psuedonymous by default-- your AIM "screen mame" was rarely your given name, you could have different usernames on each forum, your email address probably referenced your favourite sports team or anime character.
This inherently constrains aggressive "passive" personalization. Without an obvious canonical identity, you don't want to try to cross-profile too aggressively, because user "hakfoo" on site A may well be different from user "hakfoo" at site B, and you had to assume that any ID you tracked was limited or transient: when you go off to college or apply for a job, you're probably not going to want to be slapping "PonyGirl1987" on your resume.
I wonder if it's that the algorithms are limited to being one-track or if they're overoptimized to being one-track though. I assume they have a profile somewhere that looks like "12% coin collecting, 31% travel to Paraguay, 9% 2004 Ford Focus Repair, ..." and then they offset that data with what content produces the best revenue/engagement/metric of the week. In the process, many secondary interests simply get demoted to the pont you see nothing but tire change tutorials.
What you're describing is the "real" Internet, which peaked somewhere around 2007. Everything was pushed about as far as it could be in that direction.
The problem was (almost) no one was making money off it. Online advertising, as it was originally done, was a joke. Most of the early internet was based on ideals of quality, community, and freedom. None of these make much money.
So people started hunting for what DID make money, and they discovered data harvesting and targeted advertising. And like a cancer, that unholy pair devoured most of the web. Information really WAS power, like all those breathless articles and hacker manifestos in the 90s said. Power and MONEY. And the new masters of this realm have dedicated themselves to taking as much of YOUR information as possible. They need it, like Elizabeth Báthory would have needed the blood of virgins if she was an actual vampire and not just insane.
Same, but I did this long before the current tracking nightmare. That's the whole FUN of "cyberspace": you can create and destroy identities at will, which means you can experiment. Even better, you can live out multiple lives simultaneously to some degree, like multi-threading code.
Realistically, on the modern net you need to: Use a VPN, kill cookies and ads aggressively, have multiple accounts (or avoid logging on at all in some cases), and never, ever use 2FA except with financial institutions. That's the bare minimum. No point in just complaining that tech companies are evil all the time. The smart organism adapts to its environment, and uses the tools at hand.
I'm mildly curious about the whole Jordan Peterson anti-woke movement. Mildly, as in, I'm interested in the arguments but I often don't agree with them. I once made the mistake of clicking on a Peterson YouTube video and subsequently got only obscene alt-right "Peterson KILLS stupid feminist reporter!!" type of edited content. Like at some point 80% of my suggestions was crap like that, drowning out all the lovely nerdy Tom Scott stuff I come to YouTube for.
I had to reset all suggestions to get away from it. Whenever I now see a video like that about the culture wars (or about covid for that matter), I open an incognito browser and search for the title. It's nuts that this is necessary.
It's come to a point that I now hate suggestion engines with a passion. I wish I could simply only see the tweets/videos/whatever from people I follow/subscribe, chronologically. I can't figure out whether the people making these services are simply incompetent or downright evil.
I think that in the future, having a way to tell the algorithm "I didn't like this" will be common courtesy. Especially as the algorithm becomes stronger, I.E. companies offload more and more curation to ML.
On YouTube, this exists. On the Home screen where you get video recommendations, you can mouse-over each video title, click the little three dots, and mark that you're not interested in a video or channel. They'll never show up again in your recommendations.
It's a consequence of the algorithm optimizing for engagement. YT obviously wants people to stay for as long as possible, to interact, write comments, upvote/downvote and follow the chain of recommended videos. The longer they stay on YT and the more content the consume, the more ads YT can show and the more money they make.
Anti-woke/alt-right type content is provocative, controversial and makes people engage with. Either because they want to enthusiastically join in or because they want to dunk on it for how ridiculous it is. And the people who are into it are generally really into it, so they'll keep watching hour after hour of it.
YT doesn't care, as long as they get as many people watching videos and looking at ads as they possibly can.
This supposedly won't happen anymore because YouTube changed the recommendation algorithm a year or two ago to not show hardcore politics as a suggestion for everything.
Personally I find it easy to dismiss bad suggestions by looking at animal videos or vtubers, those reliably displace everything else. And that everything else was just strange Minecraft roleplays for small children, not racism lately.
It's important to note however that alt-right content is just one particular example. Often people encounter this algorithmic behavior and ascribe certain motives to Youtube etc. And while that probably IS happening to some degree, it's also true, and probably more common, that the system is just showing you more of what it thinks you will like. There are plenty of stories of this happening in a more innocuous (but still obnoxious) way: click one car repair video, everything is car repair videos. Click one clip from an old animated show for nostalgia, EVERYTHING is now clips from old 90s shows.
And to counter all this, often the system is actually surprisingly insightful in digging up obscure crap from god knows where that's really neat, and right up your alley, that you never would have seen otherwise.
And then it goes back to filling your whole front page with Simpsons clips.
Im the opposite. I very ocassionally watch a JP video which I find interesting and it means Google no longer fills my feed with woke stuff which is nice.
Sure lots of negatives, but in a few cases, it can also be useful. I have carefully used a YouTube account that's specifically trained for just the right music recommendations that the auto-play just works mostly right.
I use stimulating music (not relaxing) for work mainly, playing in background, and don't want to keep finding new music. Novelty is an important factor in stimulatory music. So this has been great for me.
It's crazy just how much I go out of my way to avoid things that "personalize" my experience.
I don't use online music services, I discover music on various platforms and and download mp3s and keep a local library.
I avoid Youtube at it's defaults, I use 3rd party apps and VLC to do most of my watching, other than my subscriptions I tend to skim the Home page very rarely.
I do not use Netflix or other streaming services, I try to hunt down DVDs/Blu-Rays and prefer ripping them for my personal library.
My only problem is exclusives, as a fan of The Witcher series, I do feel like I am missing out, but if I feel a really strong urge I can always borrow an account from a friend, create a temp profile, watch the series and delete it.
Their convenience features just add more inconvenience to me.
You should abstain from googling politically incorrect ideas.
It is possible that the state will require searches for for politically incorrect content be logged and used to create a profile on possible dissidents.
American. Programs like this almost definitely exist in America.
Even if they didn‘t, it would only be a matter of time for them to be implemented since it is in the interest of the state to monitor its population for behavior that could be dangerous to it.
That isn‘t even necessarily bad! It‘s just problematic that it could easily be used for immoral surveillance.
It‘s imo a good thing to identify likely future terrorists.
It‘s not a good thing to surveil members of e.g. sexual minorities or peaceful dissidents.
I‘m a recent immigrant, so I am sorry if my English sounds weird.
Not an unreasonable concern. Remember Google's Project Dragonfly? They were building and testing a system which would not only censor Google Search in mainland China, but provide Communist Party authorities with unlimited access to search logs.
They only stopped because they got caught, and enough Google employees caused a stir about it internally. The executives who supported this project and kept its existence secret within Google are still with the company. Who's to say they won't try again, in China or any other country?
>Locating core parts of the search system on the Chinese mainland meant that people’s search records would be easily accessible to China’s authoritarian government, which has broad surveillance powers that it routinely deploys to target activists, journalists, and political opponents.
The dumb thing about recommendation algorithms are that they assume that I’d like the same things, regardless of my mood or current situation. YouTube Music at least try to make guesses based on the time of day, but fail to take the actual music into account and just focus on the bands, and not the tempo, lyrics and overall sound of a track.
The only “algorithm” that sort of work is Amazons book recommendation, but I’m not sure that not just based on what others have bought.
In the novel, "Feed", one of the characters has a hobby of searching for all sorts of random things, in order to confuse the algorithms. She would add absurd items to her shopping cart, without buying any of them.
It's so weird, across all Google's products I occasionally get these weird Arabic and sometimes even Islamic content suggestions.
I've never opened any of these links, am not behind a VPN, am very much not Arabic and do not understand the language, and - with all due respect - am completely uninterested in Islamic holy scripture.
Why, even after all these years, Google thinks this is content I'd want to see is a complete mystery to me. Something I'm going to have to live with for the rest of my life it seems.
I can think of several explanations. The least likely is that for some reason your IP (or one of the IPs) is misattributed to another country by their geolocation mapping system. Another is that some of your property have been used by an Arab (you bough a second-hand device, lost a phone etc). More probably ones: one of your devices might have been infected or you might have clicked an ad that was disguised by something else but was linked to products/services offered in Arabic.
Yes those are the only explanations I came up with as well, but none of that applies.
I have kept a log of all IPs I regularly connected from, and only a single one of those originated in another country - another European one.
Never lost a device (jinxed now) or bought or even temporarily used a second-hand one.
An ad or intrusion do seem to be the most likely culprits, but ads are blocked everywhere, and I believe my security hygiene to be pretty decent.
This has been going on for over a decade now. Due to the sporadic nature if these happenings, I'm almost starting to think there's some old (but still wrong) data stuck on some edge node somewhere - or perhaps someone at Google is actively teasing me for being so critical of the company :)
the thing I hate the most are personalized ads and amazon suggestions that keep pestering you for weeks about an item you just bought and don't need anymore
You bought a cable for your phone yesterday. Why don't you buy another one? Don't you wanna buy? You sure? Here are the best single cables in your area ready to be hooked.
It's funny, just today I noticed how much worse is Google Now feed (or whatever it's called today) because it keeps repeating news items and category that I have very little interest in. It's almost like Google is trying to save money by computing less accurately for my interests.
214 comments
[ 4.9 ms ] story [ 234 ms ] threadhttps://twitter.com/jjcollinsworth/status/139066623945075507...
After 9/11 and the Patriot Act, librarians fought the government to keep the reading habits of patrons private as a core duty like doctors pledge to do no harm. Now it's difficult to get people to understand why clicking through to an attractive hammock while browsing Amazon registering in 900 databases and manifesting in ads and cold calls selling tropical vacations and being flagged on some government system as 2% more likely to be a flight risk if let out on bond (who am I kidding, AI doesn't give percentages, it just gives conclusions, looking at a hammock might be that stray pixel that can turn an OCR "O" into a "Q") - difficult to get some people to understand why that would make you nervous.
The answer to that isn't "use TOR to get to a VPN to browse amazon, and pay for that with a burner debit card loaded with bitcoin" or whatever works this week. There isn't really a hammock.
edit: also Amazon has that figured out, some researcher there figured out that they can identify you based on your click patterns and timing. Right now you can choose between 1) an app that will just click on everything silently, or 2) another app by a professor at some midwestern university that will jerk around the timing and positioning of your clicks in a way that throws your fingerprint off. The first app has been banned by every store, and triggers 80 warnings and a waiver that has to be signed with two-factor even if you manage to root your phone and sideload it. You have to compile the second app yourself, with a weird toolchain, and it draws in 640Mb of npm libraries. It's already been updated three times in response to Amazon's countermeasures, and the professor just wrote a paper about the entire method probably being ultimately doomed.
This isn't restricted to just Amazon either; as I understand it, this is the core of the current reCaptcha version (and even more so, its evil twin, reCaptcha V3, which is seemingly so effective it can rely on those mouse events exclusively in most cases)
As a dev, I google a lot to figure out how to do X.
My experience with DuckDuckGo was that it added 2-10 minutes of filtering for usable results every time.
Do you have an example of a search you've done where the results were much better from Google Search than from DuckDuckGo?
Due to the very topic this HN post is about, I don't think this is a good method - your results might be vastly different from mine.
But let's look at the number of relevant results among the top 5 results for some thing I recently wanted to learn about.
"Comprehension categories" - a rather specific term from category theory
https://www.google.com/search?q=comprehension+category https://duckduckgo.com/?q=comprehension+category
Google: 5/5 (for me) DuckDuckGo: 1/5 (for me)
"comprehension categories category theory" 5/5
I'd hardly say it's 2 to 10 minutes extra work, especially for such an obvious example
Here you go: https://imgur.com/a/CssDXS3
Without the protective ideological shield, that reaction would have been deemed "hostile" and "toxic" by the twitteria.
I was very curious about how this is evidence that women specifically get berated on the internet rather than evidence that everyone is a potential target. Obviously from this interaction alone you can't tell if it happens more to women or not but... I thought it was kind of funny to say "this thing happened to me as a man and now I really understand that this thing happens to women."
I can't know whether his comment was reasonable without looking through the comments he was referring to (and neither can you), but even given that he was being unreasonable, you chose to be unreasonable in turn.
Nah, I can. It seems quite toxic regardless of what has been said to him.
> even given that he was being unreasonable, you chose to be unreasonable in turn
I don’t see what is unreasonable about shaming anti-social behavior. When you see any “unsolicited advice”, even between two same gendered persons, as a manifestation of patriarchy you are bound to come off as toxic.
On desktop, set Firefox to purge everything when closing the browser. Use Chrome only when necessary to do strictly Google-related things.
But to what extent would the algorithm know about you if you clear cookies/storage data though? I also imagine changing IP addresses help… does Google use IP as a source for profiling?
If you’re logged in to a service on multiple devices you have now linked multiple fingerprints to your account that can be compared against third party data.
A VPN will not protect you from this tracking.
[0] https://amiunique.org/
(I’ve plugged this site before, so I feel like I should say I am not affiliated with it)
Gives you the entropy your browser is giving out in number of bits.
I am not afraid, but annoyed to not have "manual mode" internet where everything I do has no hidden consequences and is simply what I asked.
kinda forces me to watch something else that is new but at least somewhat wanted in an attempt to make the algorithm look at the new shiny i'm interested in and forget the previous one.
i find it funny though when amazon still make recommendations to me based on a purchase in 1998. yes, Jeff, i still want to buy wrestling VHS tapes...
For example if I consistently listen to 3 unrelated songs [A B C] on youtube together, the algorithm will regularly recommend them to me (because of my unique behaviour, not because they're similar). If I reset my history and then listen to song A, then B and C get immediately recommended even though they aren't similar to A and they don't exist in my listening history which means that the information about my listening/browsing habits is still there in the recommendation model.
Seems to work reasonably well for me.
Nobody knows, I only know I don't want to play this game.
If a real human mentions a video that piques my interest, I may watch it. If I'm interested in finding a video on a particular topic, or a specific scene from this-or-that film or TV show, I will search for it myself. If am interested enough in a certain producer's content, I may "subscribe" to them, but I will be the judge if I want to watch the latest video they uploaded.
There is one advantage: some obscure music videos. There are some rare pearls with the comments section almost exclusively thanking YT algorithm for taking them there. Happened to me so many times that I have a separate browser instance and a Google account for YT and I'm very careful what I click when I use it.
https://www.youtube.com/c/TerminalPassage/videos
I didn't even know there was a Jazz/Fusion rock scene in Japan in the 70's for instance. I never would have heard of this band without YouTube suggestions but really enjoying their work.
https://www.youtube.com/watch?v=20eVtt1VJyg
I did it for a couple of weeks to see what was being promoted by youtube algorithms in terms of content and ads to the more financially challenged parts of the society. Gangsta and sexually inuendo videos in neighbourhoods that although poor do not have a crime problem and young people don't dress like that. But it seems the algo is working on it. Note: This is in a European country
Last week it decided that I wanted to see rap videos and filled my feed with them. I don't listen to rap. I've never watched a rap video. I've told it I'm not interested in every single video, but it still puts them there.
I pretty much only YouTube in Incognito mode these days. Everytime I forget, YouTube manages to annoy me so much within half an hour that I switch to Incognito again.
There was a time a couple years ago I watched one episode of Arthur for nostalgia purposes and it completely obliterated my recommendations. I scrolled for hundreds of videos and didn't see a single non-arthur one. It took months for the site to become usable again.
It’s funny you mention that.
For many years I would get recommendations for Latino lgbt books. Weird as it’s not really my topic. I found a way to look at how they generate recommendations, and it turns out that I bought a book about a gay Chicano growing up for a college class in 1998 or something.
I don’t think the user meant to imply they had gained any special insight into the algorithm other then personal confirmation that purchase as far back as 1998 are still being incorporated.
If you spend a minute too long looking at the menu for a movie it will not only auto play, but for the foreseeable future Netflix will assume you’re watching it, want to finish watching it and watch similar stuff.
More than most other services Netflix doesn’t trust your ratings or your lists when making up recommendations.
The discovery is specifically what makes Youtube, Netflix or Spotify sticky beyond just content hosting, and even at suggestions, pirate websites are better at it.
And yes, they had to get rid of ratings, because some people's feelings got hurt because they were being rated so low. Can't have that pesky objective reality rearing its ugly head.
It's insane that I feel it helpful to take on "multiple personalities", but there it is.
Part of it is that these algorithms are fairly one-track. They can mix it up a bit, but it's always too much of one thing and too little of another. They can't truly comport with the reality that someone can have multiple interests and tastes.
You should be able to trust a popular open-source project, because you trust that there are folks who will go in and take a sticky beak at these kinds of things. This why a lot of security folks prefer open-source software over close-sourced solutions.
Surely, if Chromium were cross-correlating profiles, then it'd be on the front page of Hacker News in short order.
I'm unaware of any potential problems relating to fingerprinting and so on, but for your use case it doesn't sound like it'd be a problem.
https://support.mozilla.org/en-US/kb/containers
https://support.mozilla.org/en-US/kb/profile-manager-create-...
https://www.huffpost.com/entry/data-and-goliath-digital-surv...
I can recall in the mod-late 1990s when Mainstream America was starting to go online for the first time, everyone was training their kids "never give out your real name." From that simple bit of stranger-danger paranoia, we built a lot of communities as psuedonymous by default-- your AIM "screen mame" was rarely your given name, you could have different usernames on each forum, your email address probably referenced your favourite sports team or anime character.
This inherently constrains aggressive "passive" personalization. Without an obvious canonical identity, you don't want to try to cross-profile too aggressively, because user "hakfoo" on site A may well be different from user "hakfoo" at site B, and you had to assume that any ID you tracked was limited or transient: when you go off to college or apply for a job, you're probably not going to want to be slapping "PonyGirl1987" on your resume.
I wonder if it's that the algorithms are limited to being one-track or if they're overoptimized to being one-track though. I assume they have a profile somewhere that looks like "12% coin collecting, 31% travel to Paraguay, 9% 2004 Ford Focus Repair, ..." and then they offset that data with what content produces the best revenue/engagement/metric of the week. In the process, many secondary interests simply get demoted to the pont you see nothing but tire change tutorials.
The problem was (almost) no one was making money off it. Online advertising, as it was originally done, was a joke. Most of the early internet was based on ideals of quality, community, and freedom. None of these make much money.
So people started hunting for what DID make money, and they discovered data harvesting and targeted advertising. And like a cancer, that unholy pair devoured most of the web. Information really WAS power, like all those breathless articles and hacker manifestos in the 90s said. Power and MONEY. And the new masters of this realm have dedicated themselves to taking as much of YOUR information as possible. They need it, like Elizabeth Báthory would have needed the blood of virgins if she was an actual vampire and not just insane.
Realistically, on the modern net you need to: Use a VPN, kill cookies and ads aggressively, have multiple accounts (or avoid logging on at all in some cases), and never, ever use 2FA except with financial institutions. That's the bare minimum. No point in just complaining that tech companies are evil all the time. The smart organism adapts to its environment, and uses the tools at hand.
This is a quarter step away from voodoo and south seas cargo cult rituals.
I had to reset all suggestions to get away from it. Whenever I now see a video like that about the culture wars (or about covid for that matter), I open an incognito browser and search for the title. It's nuts that this is necessary.
It's come to a point that I now hate suggestion engines with a passion. I wish I could simply only see the tweets/videos/whatever from people I follow/subscribe, chronologically. I can't figure out whether the people making these services are simply incompetent or downright evil.
The AI future is now and it's awful.
I find Spotify similar. I try a suggested playlist. I check out after 2 or 3 songs. It doesn't get the hint.
Anti-woke/alt-right type content is provocative, controversial and makes people engage with. Either because they want to enthusiastically join in or because they want to dunk on it for how ridiculous it is. And the people who are into it are generally really into it, so they'll keep watching hour after hour of it.
YT doesn't care, as long as they get as many people watching videos and looking at ads as they possibly can.
Maybe that's the effect and not the cause.
Personally I find it easy to dismiss bad suggestions by looking at animal videos or vtubers, those reliably displace everything else. And that everything else was just strange Minecraft roleplays for small children, not racism lately.
And to counter all this, often the system is actually surprisingly insightful in digging up obscure crap from god knows where that's really neat, and right up your alley, that you never would have seen otherwise.
And then it goes back to filling your whole front page with Simpsons clips.
I use stimulating music (not relaxing) for work mainly, playing in background, and don't want to keep finding new music. Novelty is an important factor in stimulatory music. So this has been great for me.
I don't use online music services, I discover music on various platforms and and download mp3s and keep a local library.
I avoid Youtube at it's defaults, I use 3rd party apps and VLC to do most of my watching, other than my subscriptions I tend to skim the Home page very rarely.
I do not use Netflix or other streaming services, I try to hunt down DVDs/Blu-Rays and prefer ripping them for my personal library.
My only problem is exclusives, as a fan of The Witcher series, I do feel like I am missing out, but if I feel a really strong urge I can always borrow an account from a friend, create a temp profile, watch the series and delete it.
Their convenience features just add more inconvenience to me.
https://www.nbcnews.com/tech/security/can-government-look-yo...
That isn‘t even necessarily bad! It‘s just problematic that it could easily be used for immoral surveillance. It‘s imo a good thing to identify likely future terrorists. It‘s not a good thing to surveil members of e.g. sexual minorities or peaceful dissidents.
I‘m a recent immigrant, so I am sorry if my English sounds weird.
They only stopped because they got caught, and enough Google employees caused a stir about it internally. The executives who supported this project and kept its existence secret within Google are still with the company. Who's to say they won't try again, in China or any other country?
https://theintercept.com/2018/11/29/google-china-censored-se...
>Locating core parts of the search system on the Chinese mainland meant that people’s search records would be easily accessible to China’s authoritarian government, which has broad surveillance powers that it routinely deploys to target activists, journalists, and political opponents.
The only “algorithm” that sort of work is Amazons book recommendation, but I’m not sure that not just based on what others have bought.
The book, written in 2002, was very prescient.
(this is a HN joke)
(but I did start building my own music player)
https://twitter.com/FPresencia/status/1379765168482172940
I've never opened any of these links, am not behind a VPN, am very much not Arabic and do not understand the language, and - with all due respect - am completely uninterested in Islamic holy scripture.
Why, even after all these years, Google thinks this is content I'd want to see is a complete mystery to me. Something I'm going to have to live with for the rest of my life it seems.
I have kept a log of all IPs I regularly connected from, and only a single one of those originated in another country - another European one.
Never lost a device (jinxed now) or bought or even temporarily used a second-hand one.
An ad or intrusion do seem to be the most likely culprits, but ads are blocked everywhere, and I believe my security hygiene to be pretty decent.
This has been going on for over a decade now. Due to the sporadic nature if these happenings, I'm almost starting to think there's some old (but still wrong) data stuck on some edge node somewhere - or perhaps someone at Google is actively teasing me for being so critical of the company :)
yep the hyperparameters for the search suggester model need a lot more tuning