It's time to reveal all recommendation algorithms – by law if necessary As it’s been about forty years since I’ve had pimples, it astounds me that YouTube’s recommendation engine recently served me videos of people with some really severe skin problems - generally on their noses. The preview images themselves are horrifying, and should really come with some sort of content warning. I immediately …
If recommendation algos aren’t shared then we need - by legislation, if necessary - a switch that turns the recommendation engine off.
And still, this will only solve part of the problem. Your data is still (legally and illegally) collected, shared, sold and analysed.
You must be able to say no to any data collection and analysis for an effective remedy. Actually, no should be the default and an explicit opt-in with warnings for your health and well-being should be mandatory.
Not good enough. I mean, that's fine for those of us who actually care about such things, but it still leaves the other 98% of the population being manipulated by said algorithms.
They need to be turned off entirely. For everyone, forever. If I want some kind of online service to "recommend" content to me, I'll go to a website dedicated to that purpose, and hopefully I'll go knowing what to expect from it.
totally agree. In the Netflix Documentary about social media "Social Dilemma"; you have the guy who created the YouTube algorithm that gives you the "What to watch next" list and even he said NEVER use it. Always search from scratch.
when the guy who created an algorithm is telling you not to use it, you know that something is wrong.
I think you're right. I have no Youtube/Google account and actively resist data collection generally. This means the Youtube algorithm probably has only the VPN provided IP address and current session history on which to base any recommendations. This can produce odd results at the start of a session but recommendations tend towards similar subjects as the session goes on.
" I have no Youtube/Google account and actively resist data collection generally."
Unless you block javascript, all those silent trackers on the web sites you choose to visit provide the data feed for the ad brokers. I recently downloaded a professional security report and found that the download page included an obfuscated script bot called directly from Farcebook. As the page (actually unlawfully) demanded an email address to allow the download, it's quite possible that Farcebook now have my email address.
Dynamic filtering is what Hill considers "equivalent", but it pales in comparison, it's just an on-off switch for the entire domain. For example, if you need to enable javascript to regain functionality it will enable everything else that comes along with that subdomain (tracking pixels for instance). uMatrix allows you far more granular control.
I agree with the sentiment "If the ads don't appear to match yourself, then you have been successful with hiding yourself from the ad-spionage".
However, while the privacy-minded smart folks exhibit a certain level of smartness; the platforms occasionally exhibit a certain level of dumbness.
I use a browsing setup for YouTube that is mildly good at suppressing tracking, nothing sophisticated but better than out-of-the-box browsers with no precautions.
And then, you watch videos on military history and you get ads for cosmetics clearly targeted at young women.
I guess any old-school advertiser would laugh at such "targeting". If you don't know anything about the eyes watching the video into which you cram your ads, at least learn something about the video you cram your ads into.
Are you suggesting that young women don't watch military history videos?
For what it's worth, some ads are basically throwing money on a fire. The client is after as wide a net as possible and has money to burn on their awareness campaign so you do what the client wants. Each campaign and client is different (and the assumed competency of the people setting up the ads is definitely not a guarantee)
.....primitive "expert systems" were based on sets of programmatic rules. To a greater or lesser extent, the rules were written in pseudo-English, and could probably be understood by a non-expert.
That was then......now we have neural networks providing the "expertise".......and even the creators of these devices have NO INSIGHT AT ALL to the logic which the neural network uses:
(1) Train the neural network on millions of examples ("big data") of the sorts of things you want the thing to do -- this example is "good", that example is "bad" -- over and over
(2) Note that this good/bad classification is arbitrary and may not even be fact checked
(3) Test the "training" to see if the neural network responds "correctly" to inputs which it has not seen before
(4) Note that this testing uses a volume of data MUCH MUCH smaller than the training database
(5) Put the neural network into production
(6) Note that a neural network CANNOT REPORT ON THE INTERNAL LOGIC USED to come to any conclusion
(7) Worse......if the production system is learning as it goes, even the people running the system have no audit ability on what is "learned"
Quote: "...I should not only be able to interrogate how I got a horrifying video of a very bad case of pimples, I should be able to get in there and tune things...."
Dream on!!!! Even the owners/builders of neural networks cannot do that once the neural network is running!!
Isn't progress a wonderful thing.....at least according to the people pushing this technology!!!
Not that long ago, recommender systems weren't ANN-based. Earlier systems typically used mechanisms such as collaborative filtering (CF) and matrix factorization (MF). Using ANNs for recommenders only came into vogue around 2016-2017, generally in an effort to squeeze yet more conversions out of them. In other words, the old methods worked Just Fine for users, but GooFace and the like needed to show growth of some sort, and their researchers had to justify their jobs.
Gao et al in the inaugural issue of ACM Transactions on Recommender Systems (just out a couple months ago, get 'em while they're hot) have a short history in their introduction.
trouble is, there is no one 'algorithm'. Or rather, the algorithms are such a cluster-fuck of algo-cut-and-paste-mix-and-shake-stolen-copied-misapplied-genius-crap-proprietary-algorithm compilation that nobody's able to make any sense of it. Which is pretty convenient ;)
Well yeah. Indeed.
Because of blue/green deployments and DevOps practices companies like Google can push out an experiment to a small number of people (only a few million) and see what keeps them engaged. They can do this many times per day, and the algorithm will get tweaked based on what's best (for Google, not what's best for you, duh!)
Therefore "the algorithm" could be substantially different at 9am, 12pm and 3pm even for the same person, and they could be running several different experiments simultaneously.
Cluster-fuck is the word.
My advice, stay off social media and try to avoid the recommended videos on YouTube if you can. It's designed to suck as much of your time as possible.
I'm fairly sure that there is exactly one algorithm: the adverts promoted to you are the ones from companies who paid the platform to spam their shit.
I'm in full agreement with the person who said that refusing permissions and hiding data leads to you getting random ads, though. It's hardly a shock - the opt outs on platforms like Facebook straight up say "this won't change the number of ads you see, they just may not be relevant to you". To which all I can say is: if you're willing to pay Facebook to have them show irrelevant ads to me, how desperate are you?
I will never do business with a company that spams me on social media. Even if I want what they're offering, I'll go elsewhere. More people need to make this plain, because they're not going to stop hassling us as long as someone sees their ad and goes "Ooo".
The advertisements are certainly using that model, but the recommendations are at least intended to use a better algorithm. This is not for your benefit. They're trying to recommend things that you want to watch so they can keep you there for hours and show you more adverts. I don't know how well it works when you give them a lot of data to go on, because that seems like a better starting point, but if the author does as the article describes and works to destroy that data and block their attempts to track the activity, I'm not surprised that the recommendation system picks some things at random in the hopes of getting something that works.
I don't really understand a person who wants to block the tracking that social media companies do but still cares about the recommendation systems that they're deliberately obstructing. I block the tracking, and I've never clicked on a recommended video except when YouTube decided to autoplay one, at which point I quickly closed it. If they can't identify any patterns about what I want to watch, that's a sign that I'm blocking things correctly (or if we're conspiratorially minded that they've detected that I'm trying so deliberately recommend crap to lull me into a false sense of security).
This, and what the poster just above said. I make some effort to obfuscate my online presence, but not nearly as much as other posters here; I'm fairly sure that, in principle, Google knows more about me than most of my relatives. And I still get tons of utterly nonsensical recommendations.
If I try to use Google News, it still frequently gives me soccer and horoscopes, in spite of the fact that I've never searched for either in my entire life anywhere, and I have in fact explicitly told it dozens of times that I do not want those topics (before giving up on it altogether and switching to another aggregator).
Spotify? I have over a thousand songs in my favorites, and yet if I tell it to play from my favorites at random, there's a half dozen songs that pop up like 10% of the time.
Amazon's recommendations sort-of make sense. Most of the time. Unfortunately, any time I buy something like a vacuum cleaner, it then spends the next few months attempting to sell me dozens of vacuum cleaners. Is it that hard to understand that there are items nobody gets more than one of?
Ads served by Google are - I dunno, sometimes they have something to do with my recent web searches, but most of the time they are... not quite random, but apparently fixated on things that come completely out of the blue. These days it's big in apartments in places I've never heard of, for example, but up until a couple weeks ago it used to show me some kind of manga.
I believe there are two kinds of recommendation algorithms right now: ones that are based on neural networks, and those that are based on a ginormous mess that nobody has understood in years.
Unfortunately for the article's author, there's something those two categories have in common. You cannot explain the output just by knowing the input and the algorithm. And you cannot apply effective constraints on their output just by tweaking the input.
Since the current hotness is Graph Neural Networks for recommender systems (see Gao et al in ACM ToRS, as I mentioned in another post), open-sourcing them wouldn't help for most purposes. You could run your own simulations with them, but 1) you'd need the current weights to have any prediction power, and 2) explicability for GNNs is pretty much as bad as it is for other ANN algorithms.
With Twitter we have the unusual case of a recommender system that's loaded up with special-case rules so the badness can be discovered. YouTube's recommender is not going to be amendable to interpretation.
To clarify what some other commenters have said, with "AI", or machine learning, there isn't an algorithm. It really isn't a huge number of if/else statements. Even better, even the people that create/own/manage it have no idea why it does what it does, how it works, or indeed how changing something affects the outcomes.
That doesn't stop modifying the AI's output with things like 'anything Elon mutters is gold', of course.
There is still an algorithm working in a deterministic manner and actually it's fairly simple and easy to comprehend. The difficulty with AI is that the decision making process is ultimately governed by tabulated data. There's nothing wrong with that either per se - it's how e.g. Yacc has generated parsers for nigh on 50 years.
The difference is how that tabulated data is generated. In the case of something like Yacc it can be done longhand, it's time consuming but utterly predictable. The difference with AI is those tables are filled statistically based on what values give the best results. The underlying algorithm and overall behaviour is still utterly predictable - it's still scoring, comparing and if-elsing - but it follows logic never designed, considered, or explainable by a human.
it's still scoring, comparing and if-elsing
Not in any useful sense. It's grinding inputs through a great honking MLP or GNN network stack, and rectifying the output with softmax, probably, or maybe ReLU or some other rectifier. There's some comparison at the end to get the top N results, but ANN recommenders definitely are not just running a series of iterated rules. Even with the data (i.e. the weights) it's not something human-interpretable.
Perhaps you could just decide for yourself what you might like to watch. There is a difficulty in knowing what is available but this is true of all big data repositories and is usually solved in some way by knowing what it is your interested in or just browsing. Kinda like libraries or book shops, although you do have to walk past the algorithmically chosen books on display at the front of the shop.
As for inappropriate pimple related suggestions; could I interest you in:
https://www.youtube.com/user/drsandralee
They seem little more than the following :
1) If you search for / post about thing x, we'll recommend other things that people who also searched for / posted about thing x were interested in, or things that link to thing x.
2) If you search for / post about thing x, we'll recommend other things that advertisers have paid to promote to people who are interested in thing x.
3) What Elon Musk wants promoting (mainly himself, feature exclusive to Twitter).
I've always thought that the reason these algorithms are such closely guarded secrets is not because they are so miraculously clever, but more because they are so embarrassingly simple. It's the sheer volume of linked data that makes them look impressive.
If you searched for X, Y, Z, W...
you are categorized into the group of people that searched for X,Y,Z,W...
we will offer you more of what you have already searched for (eg ads for products you bought last month)
or
we will offer you what others in your category searched for that you did haven't (ads for things they bought that you haven't).
if we don't know enough about you to put you into a category,
we will offer you random shit.
>if we don't know enough about you to put you into a category,
we will assume you're equally interested in everything and send you adverts for categories where we were paid to show it to more interested people than we have in that category.
Since we're dealing with a pipedream where, even if the algorithm became visible, there's nothing ensuring anyone would have the slightest idea of what's going on, it's better to avoid YT.
Invidious lets you do that. Watch YT videos without being tracked, without being served insufferable ads and, depending on instance, download the video. All video recommendations will be based on whatever you're watching, never on your browsing history (since instances don't keep any). On android phones, you can use the NewPipe to the same effect.
Facetious but actually, when you want things to change in America, sueing is a great way of doing it. If youtube actually lost a bit of money for recommending this video you can bet the algorithm would be improved.
(Also, think about the fact that in the USA gun manufacturers and retailers can't be sued for the damage caused by their products. And the USA is the only country where school massacres are a regular thing.)
YT has been sued over this in the past - mostly related to terrorism videos, and then there was the whole thing where kid's content was fundamentally broken, and obviously there's the long sequence of legal fights over piracy that lead to Content Match existing.
As yet, they've not had to reveal their algos, but they have modified them - mostly to support YT's bespoke form of IP law, and fence off content aimed at kids. Lawsuits can get results.
Maybe I've misinterpreted you but you seem to be a bit upset about being shown pimples. Might I suggest that people suffering from skin conditions have as much right to be represented on the internet and share their experiences of their 'own story' as you put it without framing that experience as 'marginal or unimportant' or to suggest that they are so unpleasant that you should be protected from them. Just because you don't like it doesn't mean you can demand the right to be shielded from it.
I don't like the mass surveillance any more than most people in the know, and whilst I don't really use any social media, I do spend time on YouTube and find it quite good at matching what I'm interested in viewing. At least where it works in my favour I'm happy to tolerate the intrusion, perhaps you should try not interfering with it and it might work better at helping you avoid 'unpleasant skin conditions'.
There's a subgenre of online video content that's about 'pimple popping' - mostly sebaceous cysts, but occasionally you'll run into parasites, insects in ears, and the like. It's gross-out content, fascinating and/or horrifying depending on how squeamish you are around yellow/grey/brown goo being expunged from people's skin.
I have another solution : don't use it.
Instead, there is the old-school method. It's called bookmarks. In my browser bar, I have a Youtube folder. All the channels I follow are in that folder. At the bottom of the bookmark bar folder, there is an option "Open All in Tabs" (at least, there is in any proper browser). So, when I feel like exploring my channels, all I have to do is go to my bookmark folder, open everything I have chosen, and wait for the refresh (on a 1Gbps fiber line, it's okay).
YouTube recommendations ? I don't need stinkin' YouTube recommendations. If one of the hosts on a channel I follow recommend that I go look at another channel, then I go check it out.
Bah, humbug.
I thought there was something wrong when youtube started recommending them to me.
I know what it is. There are 1.4 billion in India and they're skewing the recommendation engine by sheer numbers. The same goes for the Muslim Indian separatists and Indian government videos...
Whoever it was at Google that thought THIS was a good idea really needs to be minced, and the change reverted.
I immediately tell YouTube “I don’t want this” and “never recommend this channel.”
I don't get it, if you're pulling all the levers you can to increase your privacy level, a sign that it is working properly is that you're getting ads/reccomendations that do not match your situation or preferences. If you're a crunchy liberal and your ad feeds are full of bug out bags and links to buy gold, that's a good sign. Same if you're a staunch conservative and you get... oh whatever they're selling to liberals these days.
If anything, you should click on the recommendation (feel free to close immediately) just to help pollute the dataset.
If, OTOH, you want the machine to give you accurate reccomendations, then you should accept all cookies, enable javascript, create (and stay signed in to) a Facebook account, do the same for Google, always browse with Chrome, never use incognito mode, and never use privacy focused browser extensions.
“ an algorithm that promoted owner Elon Musk’s tweets above all others, promoted specific political interests - and that specifically will not promote tweets directly pointing to LGBTIQ+ words, concerns, or themes”
Can the author of this article provide any solid references to these serious accusations?
So the belief that Elon is deliberately changing how Twitter shares tweets etc. is fairly based on evidence. Of course, Twitter already did this, but as Musk is directly involving himself now and owns the company it is "his fault".
The Twitter code release may have spurred this particular suspicion re LGBTIQ+, however Musk has been cited as pushing himself. There was a period mid-February when lots of folks complained they were being shown tons of Musk's tweets. Articles were published stating that Elon was agitated his tweet during the Superbowl received less engagement than President Biden's. Apparently the twitter core team of about 80 employees was summoned by Musk to "fix this" which resulted in the flood of Elon tweets everyone saw. He also reportedly fired one of Twitter's principle engineers for suggesting his tweets were less interesting as interest in him was declining. https://www.platformer.news/p/yes-elon-musk-created-a-special-system?r
Hang on. Musk sets the rules here. He happily tweets "Pedo guy" at someone with no evidence beyond the country they're in, that's a serious accusation. Sounds to me like Musk thinks it's fine to make accusations with fuck all evidence.
Why are you so keen to white knight for him? Hoping some of that cash will rub off?
Many lawyers are now going to court over the use of algorithms to violate discrimination laws (racial, economic and gender) and regulations.
The same old games are being played on the people, just being hidden behind algorithms now.
This is also at the forefront of AI issues. Just a new way to discriminate without being caught.
With all your efforts to avoid data collection, even if successful, Google/Youtube might have categorized you somehow, vaguely. E.g., do you google? Have you set the search language(s), how many results you want per page, or anything else? I don't think you can do that without a persistent cookie (or a whole jar) that may then be available to Youtube. And you say yourself that you give Youtube directions - I don't suppose those will work without a cookie, either. Nor will VPN keep that jar closed. And if you don't block Google Analytics and a few hundred other creeps the rest of the Intertubes may be as capable as Youtube as far as you are concerned.
Now, let's say Google/Youtube/Chocolate Factory pegged you, with some probability, as being of age at which you might have pimply-faced kids (age being determined as a target criterion by some advertiser). Or your VPN presence point is right in the area where someone runs a campaign of some kind (which would be a coincidence). Or whatever. Those may be the only "algorithms" there are, and knowing them will not necessarily make you wiser. What the algos certainly are not is real time surveillance capitalism nightmare like "Mark's second cousin posted a picture from a birthday party tagging Mark and his teenage son, AI analysis determined that the boy has pimples, the cousin's GMail history connected Merk to a particular device with Google's super-cookie on it...." The algos are simply not that smart or complex. This must be distinguished from the capability of crunching an awful lot of collected and stored data if you are specifically targeted for some reason, but pimply-faced Youtube recommendations are simply not that important.
Point is, it may be business, not personal. I suspect that the real reason those algos are kept secret is not to deprive you of you peace of mind but to prevent advertisers from realising those algos are not as sophisticated as... well... advertised.
I used to work in advertising. It's one big scam. .... ecofeco
Yes, and in the greater schema of Internetworking Things you are fated/destined to be treated as no more than a number and excluded club member to be serially used and constantly cynically abused as seen fit by quite effectively anonymous and/or alien sources and forces?
And yes, that is one of those rhetorical questions you can deny is true until you are blue in the face and fully exhausted ..... and y'all be thanked greatly for that pathetic and apathetic response in reply to an unpleasant revelation you yourselves alone are unable to change and remedy.
You might to consider though, that your present gaolers are absolutely terrified by the arrival and deep and dark embedding and shady and shadowy development of Leading AIDEntities and Virtual Machines exercising Large Language Models in Novel Fields of Enterprising Endeavour, as is evidenced by the mad rush to try to curtail and ban their free use and your curious enlightening engagement with them.
Some folk, and invariably always those with everything and a great deal to lose, are just not bright enough to know whenever everything is already lost and they be flogging a dead rotting horse which is going nowhere incredibly fast.
Pity them not, for they care not a jot about and are not worthy of your concern as they battle to survive in the prisons they have created for you and run for themselves.
When I used one of the social media sites, every time I saw something I did not want to see ever again (partiular subject, or message type), I would click the "do not show me, never want to see again, blah, blah" buttons, but still got them.
One a few occasions when I have got something in my feed about a particular subject that I don't want to see as I have no interest, I have chosen whatever followup / more info option there is and said things like "I am a recovering abcd, please do not show me these again as it is too tempting", or "I am suffering with depression, and this kind of item is pushing me back in there".
Still showing even though if I was recovering from an illness, addiction or depression this would really worry me that they ignore your requests, and if you are in a dangerous place, this could be the tipping point.
I use a plugin 'Enhancer for You Tube' which has a tonne of options for blocking ads, removing recommendation videos (including the ones at the end of videos), hiding comments, setting dark mode among many others. This way I only see what I've searched for. Maybe I've reduced the usefulness of the site to me, but I'm happy with that. I can switch browsers if I want the full horror / experience.
The recommendation and advert algorithm may well be trying to build a picture about me but has little to work with and I never see it's hard work anyway. Everyone wins.
I'm currently migrating from Twitter to Mastodon and, while Twitter's algo is (a) poisonous & (b) shit, I find myself missing algorithmic help in Mastodon. (Mastodon is proud of having no algorithm.)
I'd like to be recommended stuff I might like (I can always say no), I'd like the system to understand that I'm not really interested in certain subjects (which probably won't respond to text filters).
I'd even like to construct my own algo - which would require, of course, that the algo is transparent.
Algo working as intended.
There were 5 pair of decrepit eyes on the board. Talking rando nonsense about the weather and carburettors.
There is a zero chance of those eyes engaging to view/click/buy. Solution : Inject some rando, irritating disturbance, bearing no relation to meteorology, or the finer points of stochastic venturi effects.
(waits for wry smiles at the setup pun...no? Please yourselves then)
You now have 20+ pair of eyes on the board engaging to vent their ire. One pair will be a Goldeneye.
Hello Mr Product seller, have a look at our metrics this month - user engagement is up across the board - sign here...and here.
You will get bonus points for propagating engagement beyond the platform.
Yaaay now we're all playing. Algo working as intended. Well done. Have a cigar,
I'd recommend a nice Cohiba, form Havana House and a glass of Tamnavulin Sherry Cask (Tesco £45) have sit down in a darkened room with a paper bag over your head (preferably bespoke, recycled, available from Rocaba - now £10 off your first purchase) at least until the palpitations subside - Fibricheck do some nice heart rate monitoring solutions which can help with that.
This post brought to you by Google Sponsored Ads and the letter "A"
Have to go now - according to my Tag Heuer Formula 1 x Gulf, it's time to check in on the Mayo clinic about this worrying growth I've been reminded about...somehow.
A number of Youtubers I follow have started moving to other video streaming providers, such as Rumble. YT's algorithms are incredibly aggressive at classifying content as something they aren't, and de-monetising said channel. YT will still post ads and earn revenue of course.
They even terminated Adrian's Digital Basement because it classified his own videos as Impersonating him! (There's a video on ADB II about it).
I still like YT for certain channel's content but I have absolutely zero loyalty to the platform. Post it elsewhere and I'll follow.
Hey listen, let's have a bit of a thicker skin in life. This has gotten out of hand where folks are indignant that they be subjected to something they don't want to hear. Whether it's a college class or a movie in a movie theater, they now have to have disclaimers, lest someone be offended. This cancel culture attitude needs to stop. Suck it up buttercup, and just go with it, celebrate opposing views and defend their right to have one. Make your point with facts and logic, not insisting that someone be sued or DOX'ing them because you don't like what they say.
I have a FB account but use it very little. I have certainly not told FB what I would like to see. What I get on the video feed is therefore presumably a vanilla set of guff
Number 1 theme is diggers and big machinery
Number 2 theme is porn
Number three theme is comedy
Sometimes I might look at the comedy, but I am not interested in the rest.
I also get a lot of ads, and I normally say less of these. However, FB keeps giving me ads for Marsh Fuels, who supply types of fuel I cannot use. In spite of repeatedly saying less of these, FB keeps giving them to me. Marsh Fuels presumably pay for each ad presented, so this means FB are deliberately defrauding Marsh Fuels.
Every time FB present an ad from a company that has been rejected by its recipient this surely counts as fraud. I doubt FB are the only company defrauding their sponsors, but this does not make it legal. The sooner FB and others face the courts for this type of fraud the better - it might persuade them to stop doing it.
... that really needs to see Google's algorithms for ad placement is the advertisers. Personally, I don't care if I'm shown pointless, unrelated adverts. But I'm sure the people who are paying per view generated would like the question answered: "What the h*ll made you show that person this ad?"
And, "Can I get my money back, please?"
You would have thought so, but the evidence is that mis-targetted internet ads have been a thing for at least 20 years and the advertisers continue to throw good money after bad. These days it is pretty much a cliché that they are randomly distributed. Private Eye has had a running gag about it for years now and the brainlessness of Amazon's "I see you bought a wahing machine, would you like to buy ten more?" is also widely mocked. *Everyone* knows that the system doesn't work. The web giants are not deceiving the advertisers, the advertisers are not deceiving the companies who place ads, and the population don't believe that the ad is aimed at them.
There is no fraud here. Just a huge waste of time and money by people who have realised that they can inflate the price of absolutely everything by a tiny amount and then cream that off to pay their salary.
From the ad space, Google is actively working to make their tools black boxes and we're supposed to just trust their algorithms. It's happening in Google Ads and Google Analytics 4 where they push their algorithm-powered tools while removing existing tools that we have more control over.
It's pretty frustrating to be the person who has to explain why two numbers don't match up.
The fundamental problem is that FB etc are ad-supported.
So you're going to get ads.
The 'recommendation systems' they use are clearly shit.
So you're going to get shit ads.
One day, the advertisers will realize that they're paying waaaay over the odds, and things will calm down.
Then, perhaps, folk will remember that once upon a time they used to pay for a daily newspaper. It, too, had ads, but they were passive and could be ignored.
Even today, some folk subscribe to paper magazines (examples: Stereophile and the Spectator). These too have giant ads. But they're preselected to be relevant (in the case of Stereophile) and mostly harmless (in the case of the Spectator).
Bottom line, if you pay for a 'feed', you'll get less crap ads. Might be time to start re-contemplating the "you get what you pay for".
Bottom line, if you pay for a 'feed', you'll get less crap ads. Might be time to start re-contemplating the "you get what you pay for". .... BlokeInTejas
That does not logically compute on a number of levels, BlokeInTejas, suggesting as it does ....
1) No matter what you pay for a "feed", you'll always get some crap ads which you would not wish to be paying for.
2) If you don't pay for a "feed', you'll always get some crap ads which you would not wish to be paying for too.
.... with both polar opposite decisions resulting, nevertheless, in the supply of you getting so much of what you are not wanting to pay for, and nobody really wants or needs?
It happens I occasionally drink with the guy who devised the 'explain' algo used by Google.
Two things stand out;
0: How laughably simplistic it is
1: You won't understand it.
Most people count from 0, like they did when they were 5.
A decent degree in maths or postgrad in econometrics and off you go, apparently there are now fast track courses only two years.
Also software is complex.
Yes, really.
Causality is really quite hard to determine in large code bases and since *of course* you've studied Boolean networks you may well feel that even if your program is as simple as a collection of deterministic finite state automata that you reach the point where it is stochastic.
Oh, but you want to inspect my code to see if it is naughty ?
Seriously dude ?
You speak C++, Python and VBA ?
Good because your mortgage rate, something you people whine about a lot these days is based upon my code.
Ah ? But you can cut & paste code snippets ?
Well done. and you're a "graduate" of a coding boot camp, hired because you're far more attractive than me.
Being prettier than me really ain't an exclusive club.
Sure you can look at my code and I can see how much patronising you can take on your maths before you tell me to just fuck off.
But you've read on Wired that algos use data ?
You can't look at the data because GDPR, oh how we laugh.
Please don't tell me you believe it is possible to genuinely anonymise data.
Seriously, you do ?
Would you like to buy some of my crypto currency ?
Long ago I worked at IBM Labs, it ended badly, actually really badly and it was mostly not my fault.
So did Mandelbrot.
Yeah, you've got a t-shirt with a Mandelbrot set on it.
You hardcore geek you.
Sadly Benoit is no longer with us but a crucial pat of his work was "sensitivity to initial conditions" or as the pop-Science you've read in New Scientist calls it "chaos"
Even as a coding boot camp 'graduate' you must have noticed that even quite small changes to code have huge effects or none on what it does.
So even if you can understand what my code does today, which you casn't, by the time you've pretended to understand it, the system no longer behaves in the same way.
Or it does.
Code is like that, be it it a neural network or the codec that plays your porn.
The problem with disliking a video is that the system sees that dislike as engagement, and it is this engagement which spurns more of the same shit.
There is a reason these systems are referred to as echo chambers.
The problem with the modern world, is there is this mistaken belief that all human problems can be solved with an algorithm, which is so far off the mark it’s actually scary.
Go to Youtube in a fresh browser, and watch in horror what the moron masses enjoy, apparently the moron masses enjoy “goofy face thumbnails” and “puerile shit”, because that is what the main page will display.
The 2 channels it seems to recommend the most are ssniperwolf and mr beast, both of which contain goofy face thumbnails and puerile shit.
Also for a lot pf parents, Youtube is the new babysitter, and this definitely affects their algorithm
Sometimes, letting them track you means you don’t have to be constantly reminded how stupid the average humans is
The thing about algorithms is that they can be subverted.
As a thought experiment, what would happen if:
[1] you created an account which spoofed the identity of <politician> (not necessarily from your own country);
[2] you created a bot which used the account from [1] to watch YouTube continuously, creating clicks as it went, and deliberately chose the most disgusting and illegal content available;
[3] You timed how long it took for <politician> to get the message.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
