"happy users" has not been on google's priority list for a long, long time. 20 years or more.
Where does Google's ad revenue come from if users give up on waiting for their searches to complete before they're ever shown any revenue-generating sponsored links?
Actually, thinking about this more, I fundamentally don't understand Google's search experience strategy over the last 10 years. Even before AI answers, they were heuristic-ing answers into a little box that pushed sponsored ad placements below the fold. Why push sponsored results below the fold? How does Google make money from a search engine that just answers people's questions directly, rather than doing lead-gen for the rest of the Internet (with much of the Internet willing to pay for that)?
They've just done a rather braindead job of it, with the experience of these loss-leader queries feeling perceptually† worse than a regular SERP page would.
It's like a grocery store running a sale on e.g. milk, selling it for below-cost — but then making you have to wait in a line for five minutes in front of the milk cooler to get it. It feels like so much friction that not only wouldn't lure people in; it would push people to avoid that grocer when its milk sales are happening, and shop at other stores instead!
---
† I emphasize perceptually here. Even if the total time spent waiting for an AI answer is shorter than the time it would take to load a SERP page, read and select a link that looks like it'll have your answer, click it, load the resulting page, and read it — the AI experience could still feel longer, in large part because "the ball is in your court" for a lot of the self-navigated query, with you being given things to look at and click on; while "the ball is in their court" for the entirety of the AI response, with just a spinner and no indication of what's happening.
As such, I wonder if this could be solved by just running a cute little animation under the AI prompt window: make it look like it's automating the loading of a [blurred to non-identifiability] SERP page, scrolling down, clicking a result, loading an [again blurred] page, highlighting a few random fragments of it, and then transposing them together to form the paragraphs of the answer — before the answer actually snaps into place on top of that animation. If you can predict how long before the AI answers, you can even use that to determine how long the animation should spend dicking around on the pretend SERP page and answer page.
You serve the ads while the answers spin up.
This is actually a problem I currently have with Ecosia on mobile. I search, get the result, and then the ads pop in, pushing the search results down, causing me to click the ad and not the result that previously occupied that part of the screen.
All human eyeball traffic goes through Google.
It is in their interest to keep destroying the open information marketplace because then advertisers are forced to buy through Google. (Much easier and more profitable than trying to make a good adtech product.)
As others in this thread already point out, it's extremely easy to use DDG or Bing if you want.
As an illusion sure, in practice not really.
I use Kagi, I bought the family plan so I could try convining my family to use it, did they? No.
Did my family want to use Signal for messaging? No.
Do my friends want to use something else than Discord for communicating with me? No.
Competition exists and in plenty but folk are too mind sucked to want to move platform. If its not broke, why change?
People don't want to change. And until such platforms close their gates or whereby something new comes along that isn't a clone; there will be no shift.
If switching cost > perceived benefit why should we expect anything other than status quo? When a truly innovative platform that offers benefit > switching cost, users will naturally flock to it. We're already starting to see that with ChatGPT.
No. Network effects mean that there's not only a high cost of switching away from the largest network, but there's a high cost to use anything other than the largest network. When that bias exists, you cannot infer any preferences from market outcomes.
benefit > switching cost + network cost
This has repeatedly been demonstrated as large networks have been disrupted. Folks switched away from MSN Messenger and MySpace didn't they?
You won’t find me on any social media other than Linkedin and Twitter. And on those, I no longer post or interact with people. I keep Twitter because of announcements and Linkedin as a public CV. I also use WhatsApp because alternatives are not really viable (but I reduced its social aspect as much as possible). You’re free to make your choices. Just don’t complain about it, if you won’t heed my advices.
Google is testing on those as we test on animals. ML for only search? Naw.
If everyone turned off their droid, stopped using their email, switched to Firefox, they would still be on top.
Yes. It's better not to, and if your going to preach, yeah. As I use no services other than Signal but we are still just pawns in this game.
We are a just pittance of income of those who are keeping the empire's on top. And who do you run too? DDG is owned by Apple so your only fuelling their empire.
This goes deeper than you could imagine.
All human eyeballs go through Google's advertising network first. (Unless you're an Apple-only person.)
So for advertisers it makes no sense to budget anything but Google first and foremost. Maybe they'll add some Bing-Yahoo if they have spare money laying around and want to catch the remaining couple percent of the audience.
P.S. Search is only a small part of Google's advertising network, and not the most important one nowadays.
It did what it needed to do to realise the same thing and get their fingers in all the other pies.
https://english.almayadeen.net/articles/blog/genocide-in-the...
If the only thing gen AI delivers is killing Google’s search standing, it’ll have served a useful purpose.
https://downloads.webis.de/publications/papers/bevendorff_20...
https://fortune.com/2024/01/18/why-is-google-search-so-bad-s... | https://archive.is/CLO4S
https://searchengineland.com/google-to-cut-thousands-of-sear...
(I use Kagi, DDG, and OpenAI for search primarily, all paid for, so my first hand experience with Google Search is limited, intentionally; are faster poor results better than slightly slower good results? I can spare a few seconds for better, this is not high frequency trading)
How is this in practice? Normal GPT is entirely untrustworthy for a lot of it's information. How can you trust it's "search" results/summaries?
I've said it before, but it's like watching the news about a topic you know a lot about and realizing how wrong they are... yet you continue to believe the same people regarding subjects you know nothing about. GPT feeds you BS, and you have to be knowledgeable about the subject to detect it... yet we blindly trust it's output on subjects we know little about. That's... dangerous.
It frustrates me to no end... and then I go back for more punishment naively believing this time it'll actually be right.
With that said, the newer voice interactive version (using the app), I've found to be pretty slick for debating yourself on technical merits of designs/architecture, etc.
But even if you redo the whole thing and it is efficient, it is still less than some minor performance improvement in Chrome can achieve for the world.
Speaking as a customer, it's slow, annoying, and makes the shopping experience there miserable. If you innovate in that software space, you might be able to make a significant impact on the trade of certified exotic wood and maybe even save a few species of tree.
It is driving people mad via the YouTube algorithms.
And it is destroying democracy.
All Youtube does is feed me engineering and astronomy videos. It's actually improving my life.
Your algorithm is a reflection of you.
This is objectively false. "The" algorithm is designed to maximize engagement - so you will be shown content you are most likely to watch, click, listen to the most. As it turns out, "Outrage Porn" is highly engaging...
Scroll through enough YouTube Shorts, and just like TikTok, your "algorithm" slowly but surely reduces to outrage porn. All paths lead to outrage porn. All of them...
You scroll past, then scroll back because "wtf did I just see?" - and now you're in. The frequency starts to increase and pretty soon you're shouting at clouds feeling like you're the only one that realizes how crazy everything is...
It's hard to keep your feed "clean" and dialed into just things that bring you intellectual stimulation or enjoyment. It's not impossible, but it's quit difficult. You constantly have to realign your algorithm lest it goes off into the weeds.
The confusion happens because this can be very easy. All it takes is rarely watching youtube.
If you only watch a random video here and there, Google won't push all the bullshit at all. So the people watching many videos, getting the shitty recommendations, and complaining that the internet is crazy end-up sounding insane to everybody else.
I have consumed more than my fair share of content on Youtube, Instagram, Facebook videos, and TikTok. My timeline is all engineering, spaceships, jumping puppies, and comedians (and not the negative kind).
It's an algorithm, it wants you to watch more and click on all the ads. It doesn't care if the world burns, but it also isn't going to push you that way.
My statement will only be "objectively false" if the algorithm is somehow programmed for political extremism. In reality, each of us is in the driver's seat.
People have been spouting crap since time immemorial - doesn't mean you have to watch it.
I would suggest a heavy and sustained dose of puppy videos.
Searching then watching is effectively realigning the feed to your latest and most engaged preferences.
Doom scrolling will get worse and worse as time goes on - it appears entirely unavoidable. They pick up on "positive" interaction signals such as slowing down while scrolling past a video, scrolling back to a video, letting it play more than once, even downvoting can be a sign of engagement and a catalyst for more of the same...
You are right - the algorithm doesn't care if the world burns... it just wants to maximize engagement. That, for most of the human population, means stuff that gets immediate reactions. Unfortunately that often means outrageous material.
Yes and no, you're correct in that it feeds you more of the stuff you engage with, but that's not what it should be going if YouTube cared about the mental well being of it's users. The YouTube algorithm is notorious for not responding to deliberate user input. Say I know that the content I engage with is harmful to my mental health, first step in dealing with the issue. You CANNOT tell YouTube to stop enabling you, even if you try. There are things the algorithm isn't smart enough to understand, or if YouTube is evil, understand that it SHOULD NOT respond to.
Example: I wasted a lot of time on Shorts, and videos less than 2 minutes, junk content. So I start hitting "Not interested", the algorithm WILL NOT pick up on the thing these videos have in common, that they are short. You can block channels, that works, except that there are thousands of similar channels which YouTube will then just push instead.
What I would like to see is YouTube have settings for: What kind of person do you want to be? And then just push the hell out of relevant content, ignoring any level of engagement. I think everyone knows why YouTube won't be doing that.
YouTube is a company, they do what's good for the stock price / profit margin. If it hurts the users, they don't care. It's not that they are evil, they just don't give a crap.
To give a specific example, something that honestly kind of baffles and bothers me is the amount of firearms related videos in my recommendations despite the fact that I never watch them, I don't own any firearms and I keep clicking "Don't Recommend Channel" and they never seem to go away. It has to be because the topic is tangentially related somehow to something I've watched, or maybe people who are into firearms overlap with some of the content I watch. But it's the volume and the non-responsiveness to my feedback that annoys me.
Another example: my wife and I enjoy watching the Eurovision Song Contest every year. So naturally we've watched a lot of that content in the days and weeks leading up to it. Obviously the algorithm is going to recommend content related to Eurovision, but my home page is completely flooded with these recommendations to the seeming exclusion of most other stuff that I want to watch. And I keep clicking "Don't Recommend Channel" on the various "reaction" videos that are just feeder content featuring people reacting to the songs and performances ... because I CAN'T STAND reaction videos in general and NEVER watch them. But they keep appearing in the recommendations nonetheless.
Recommendation algorithms are a hard problem to solve, for sure. I'm not trying to be unfair to YouTube because I know these recommendations are somehow related to what I do watch. I just know that it feels like 90% of those recommendations miss the mark regardless. YouTube is starting to feel like cable TV these days ("200 channels and nothing to watch") where my wife and I often find ourselves just doom scrolling and refreshing the home page (we watch on our TV) hoping that something we're interested in watching at that moment will appear. Often times nothing does and we go do something else.
I don't watch news on YouTube, ever. If I see a video from a news outlet I automatically click "Don't Recommend Channel." And I've done that with all of the big major news channels like CNN, Fox News, MSNBC etc. But that doesn't stop videos from local affiliates showing up in a never ending game of whack-a-mole.
To be fair these do not tend to "flood" my recommendations. But I want them to go away all together and I can't seem to make that happen.
The fact that it has been reported by groups like Amnesty International that Facebook helped to enable the ethnic violence in Myanmar makes arguing this sort of point much easier. People want to treat complaints about attention economy companies like they are coming from whiners who just don’t like ad profiles. It is useful to have specific information to remind people of the actual stakes of uncontrolled, personalized propaganda and social manipulation that these companies can enable.
So if you have one for Google I’d definitely be interested in adding it to my collection. (I’ll admit some bias, I think they are a little better than Facebook, but that bar is so low, let’s be careful not to trip over it).
The best arguments for Google's algorithms making people crazy for lay people would be Aaron Swartz like individuals who just go unknown.
I can think of atleast 4 individuals new to me the last 6mo ive discovered who used to cover court cases, ran NGOs, or worked as journalists that have been seemingly swept up in what looks like delusions.
Lots of typos, egegious oversights, etc#. Its hard to personify these individuals as theyre hard to reach and anyone w ptsd or whom knows people w ptsd know how hard it is to provide support to someone locked in a fight-or-flight reflex. After a terror/fear/paranoia becomes familiar anything else feels foreign. Human connection becomes scary.
Read up on bateson-esque schizomogenisis and reflect on the CIAs relation with the "unabomber".
Many of the comments here exaggerating the evil of the algo and claiming that it ruins people's minds unless they're "shown better" reek of elitism and an implicit desire to control other's belief/feedback choices for their own good. Aside from this being a morally weak viewpoint, it's also bullshit in a practical sense. The majority of such people with an intellectual superiority complex are little better at avoiding intellectual bias than the average person, they just sometimes learn to disguise it with more sophisticated looking justifications.
Companies are not democracies. If every single dev thinks that the user experience matters, but the CEO wants more ads, you will get more ads. It's literally a tyranny.
Or, as I like to say, Google hired some of the best, and promoted all of the worst…and after 25 years here we are.