HA! I use their Opinion Rewards app and it's clear when they are testing new algorithms, and it's also clear that they measure if it's helping users mainly by showing them screenshots and asking...
A spokesperson for Google tells the BBC that the company only launches changes to Search after rigorous testing confirms that the shift will be helpful for users
HA! I use their Opinion Rewards app and it's clear when they are testing new algorithms, and it's also clear that they measure if it's helping users mainly by showing them screenshots and asking which is better for x or y. They often don't ask which one actually answers what you're looking for. It's usually just about which one is more likely to lead you to make a purchase. They often ask you which screenshot of a search you prefer, yet both screenshots are completely identical.
It's very clear that it's just managers manipulating the surveys to make sure their new product is chosen as the "better" option.
I do this with captchas! If I have time and am in the mood, I keep checking bad boxes (yet some that at least resemble what they want like chimney = hydrant, road crossing = stairs, motorcycle =...
I do this with captchas! If I have time and am in the mood, I keep checking bad boxes (yet some that at least resemble what they want like chimney = hydrant, road crossing = stairs, motorcycle = bicycle etc.) and eventually it will accept my bad input. It works, obviously, as I was served many caotchas where there was big picture of motorcycle.divided into squares and I had to check squares where there is bicycle. Or the same with I believe bus and hydrant - one thing on the image but completely other that I had to check.
I'm absolutely certain that some software engineers somewhere love me! :-D
Part of Google's problem has always been giving weights to a site's popularity via quantity of cross-linking, as well as user clickthrough rate. These are easy to abuse, and short of a massive...
Part of Google's problem has always been giving weights to a site's popularity via quantity of cross-linking, as well as user clickthrough rate.
These are easy to abuse, and short of a massive army of trusted moderation to cut the wheat from the chaff, will trend towards garbage SEO content or the biggest commercial players.
So optimizing the algorithm makes sense in that vein, but the problem is that it has no capability (including with AI) an ability to distinguish between high-quality facts and popular nonsense. So a garbage opinion piece in PC World about air purifiers will almost certainly pop above a less-popular blog from a passionate hobbiest.
The only real way to solve this is some degree of trust system, to insure that high uality, factual material is both of those. And that's not the kind of thing easily automated.
It's part of the reason Reddit was such a useful place, the smaller communities and/or the tightly moderated ones, were able to curate (relatively) high quality user-made text content in a way that pre-aggregated it for easier searchability via indexers.
I think there's a triple constraint at work here. Any given ranking mechanism cannot be all three of simple/transparent¹; mechanical; and resistant to gaming. Naïve PageRank more-or-less ignores...
I think there's a triple constraint at work here. Any given ranking mechanism cannot be all three of simple/transparent¹; mechanical; and resistant to gaming. Naïve PageRank more-or-less ignores gameability, and Google's actual less-naïve (but also much less transparent) ranking algorithms are clearly losing the battle; Reddit moderation is explicitly non-mechanical and somewhere between relatively and utterly opaque, but is much harder to game. (At least via first-order methods. Suborning moderation is of course a grand tradition on the site.)
My prediction (with very low confidence, as befits rank speculation) is that we'll end up with islands of manually-curated content tightly-bound by informal, personal trust relationships, floating within a nigh-infinite ocean of complete shit.
¹ These are pretty close to the same thing, because if it's simple enough, people will be able to infer the behaviors even if they're not explicitly revealed. PageRank, of course, has both patents and academic writings about it, so it is explicitly transparent.
I've been looking to start establishing this for my kids, kind of a wall-off, highly curated internet experience with good auditing and best-effort privacy, while having an emphasis on...
My prediction (with very low confidence, as befits rank speculation) is that we'll end up with islands of manually-curated content tightly-bound by informal, personal trust relationships, floating within a nigh-infinite ocean of complete shit.
I've been looking to start establishing this for my kids, kind of a wall-off, highly curated internet experience with good auditing and best-effort privacy, while having an emphasis on auditability and only know-in-person access to thinks like chat and forums.
PMs with kids explicitly not e2e encrypted for that reason. Or perhaps E2E but with a backdoor for all relevant parents,
Can we start one? At least site recommendation? I'm in dire need for clean internet for my kid. They're very good about staying within only allowed sites (trust but verify) and I don't need...
Can we start one?
At least site recommendation? I'm in dire need for clean internet for my kid. They're very good about staying within only allowed sites (trust but verify) and I don't need technical solutions for online chats or site blocking. Good ad block already achieves what we need.
No social media, no chats or forums, no intra-site search engines and definitely do not click on ads when they slip through from the game wiki type of sites.
Just a different viewpoint: I've always let my kids have full access to the entire internet. With the idea that it's going to be a significant part of their adult lives and the best time to learn...
Just a different viewpoint: I've always let my kids have full access to the entire internet. With the idea that it's going to be a significant part of their adult lives and the best time to learn skepticism and critical thinking is when you are young.
I try to talk to them about not trusting anything at face value and encourage them to talk about what they do online, even if it's not for kids. I try not to judge them for anything they look at but talk about the content in terms of why it exists and what the people who made it stand to benefit, etc. I advise them not to give out any personal information.
We limit their screen time, Homework and chores have to be done. And we encourage them to go out into the world, ride their bikes to parks etc. but when they do have screens there are no limits to what information they have access to.
I've had this policy since before they were old enough to read. With the idea that at some point they would be able to access the entire internet regardless of what I want. My parents had no idea what I was up to online, and could not have stopped me short of locking me in an unwired room. Now devices are even smaller and wireless signals are everywhere. I figured it would be hubris to assume I could control my kids online experience once they have a few years to learn tech.
Basically instead of a walled garden, I tried to have a transparent, peer relationship, where I'm just here to give advice and help them learn how to navigate the ever-changing cesspool that is the internet. At this point they are 11-14, so they are well past the point where I could control their Internet. And we are very much navigating it together. I feel like I would understand the Internet of today a lot less if I didn't have them to explain the tik tok memes to me.
My parents never succeeded at their limited attempts to control internet usage when I was younger. When I was a little kid and needed Mom to sort out dial-up that was one thing. But even by middle...
My parents never succeeded at their limited attempts to control internet usage when I was younger. When I was a little kid and needed Mom to sort out dial-up that was one thing. But even by middle school I was more savvy than them. I doubt it'll be any different for newer generations -- if there's anything teens and tweens have been good at throughout history, it's getting around their parents' rules.
The one thing my parents could have done better imo would have been to educate us on potential dangers like grooming and interacting with strangers beyond kiddie "stranger danger" stuff. Nothing bad ever happened but in retrospect my younger sister was in a host of groomer-y situations that they didn't do enough to prevent imo. But they themselves weren't really knowledgeable enough to do that, which is a big problem even outside online spaces.
I'm envisioning trying to build something in the vein of (or on top of) Yunohost, like a commmunity in a box. The hard part, as always, is community buyin.
I'm envisioning trying to build something in the vein of (or on top of) Yunohost, like a commmunity in a box.
I know many folks may not like to hear this but part of the problem is "Big Tech Capitalism". Google should introspect for a while on why it became Google in the first place - organic and...
I know many folks may not like to hear this but part of the problem is "Big Tech Capitalism".
Google should introspect for a while on why it became Google in the first place - organic and untainted search results that worked on "Content is King" principle. While that principle holds true even today in general, Google algorithms seem to have introduced some gate keeping which override it.
This sure looks like an opportunity for some enterprising entity to just shockingly eat Google’s lunch when it comes to the search business. Honestly, making an algorithm that uses a better...
This sure looks like an opportunity for some enterprising entity to just shockingly eat Google’s lunch when it comes to the search business.
Honestly, making an algorithm that uses a better weighting of “indie” URLs, some qualitative language assessment, maybe linking from reputation-based discussions (a la stack exchange/Reddit-like sites) isn’t impossible to make.
Basically an algo that incorporates already existing human-created recommendation systems instead of pitting ads against them.
The quality content IS out there, and I still maintain that’s what’s gonna win out in the end… there just needs a slightly less greedy tool going mainstream to reinvigorate it.
One of the major differences is being able to add your own custom weightings (or trust factor if you will) to every website. It's stupidly easy to ban AI image websites if you're annoyed by their...
One of the major differences is being able to add your own custom weightings (or trust factor if you will) to every website. It's stupidly easy to ban AI image websites if you're annoyed by their results, and you can uprank fan wikis with better content than Wikia.
Second this. Have a bit of work to transition myself over properly but I've enjoyed experimenting. Even thinking of paying so I can support what they're doing.
Second this.
Have a bit of work to transition myself over properly but I've enjoyed experimenting. Even thinking of paying so I can support what they're doing.
I prefer it for the great majority of my searches. When I first started using it, I felt like it was basically on par with Google, but now it really feels like a big backward step when I open a...
I prefer it for the great majority of my searches. When I first started using it, I felt like it was basically on par with Google, but now it really feels like a big backward step when I open a new browser and accidentally do a Google search when I thought I was doing a Kagi search. I'm not sure if Google has gotten worse, if Kagi has gotten better, or if it's just various customizations I've made to my Kagi search results over time — it's quite possibly all three — but I don't think I could go back now.
Mind you, I am not a subscription kind of person. I don't do Netflix, Spotify, etc., etc., because I am basically allergic to the financial complexity that comes with them; I have to really get a lot of value from a product to be willing to subscribe, and I'm not willing to subscribe to more than about 3-4 services at a given time (I will cancel one to start a new one, even if they are totally different classes of product, just to keep my total subscription number as small as reasonably possible). But Kagi's service is worth it to me and I feel good about supporting them.
That being said, there are a couple of cases where I will use both Kagi and Google in conjunction:
When searching for the cheapest place to buy a specific product, or when searching for a broader selection of products to choose from, I find that Kagi and Google give me very different results, so it's worthwhile to have a look at both sets of results. Usually Kagi will find me more obscure shops, which usually have better deals and a more unusual selection (for example, Kagi recently saved me about $80 on a $175 tool by finding an up-and-coming tool shop that's trying to break into the market with some incredible deals), but not always.
When searching for images, Google usually gives me better results. At this stage, Kagi disobeys quotations marks when I do an image search, so I get more junk results mixed in, and it also returns fewer images. However, it often gives me different results from Google, so I still check both. I hope/expect that Kagi's image search will be improved in the future, but I think regular search is their priority right now (which is fair; I do regular searches way more than I do image searches).
I still use Google only for video searches or map searches.
Of course that depends on what you find wortwhile paying for. It is not like it magically finds stuff I can't find on Google, but the overall experience is just so much more frictionless with...
Of course that depends on what you find wortwhile paying for. It is not like it magically finds stuff I can't find on Google, but the overall experience is just so much more frictionless with features that are focused on the users needs and not the advertisers, that it is easily worth the small price for me.
I frequently google for movies & TV series and look for Rotten Tomatoes pages. I noticed today that the results' descriptions contain irrelevant information, often from other pages, whereas they...
I frequently google for movies & TV series and look for Rotten Tomatoes pages. I noticed today that the results' descriptions contain irrelevant information, often from other pages, whereas they used to contain the excerpt of the critic/audience review summary. I wonder if these changes Google made have screwed up SEO somehow.
Well it has had some decent improvement for my wife's shop. I've always made sure to use decent SEO, good images, alternative tags, etc. It's a small Wordpress/WooCommerce store in the UK. Most...
Well it has had some decent improvement for my wife's shop.
I've always made sure to use decent SEO, good images, alternative tags, etc. It's a small Wordpress/WooCommerce store in the UK.
Most searches for her products are drowned in Temu, Etsy and other company products, but a random UK VPN and Private tab on Google has shown her products with the same searches to be on page 1. Yesterday it was 2 or 3 at best.
Bing is still crud though.
Edit: Just a note to say I do not keyword stuff, no hash tagging and there is not much in the way of off linking and linking back to social media. It's all in long form descriptive English.
HA! I use their Opinion Rewards app and it's clear when they are testing new algorithms, and it's also clear that they measure if it's helping users mainly by showing them screenshots and asking which is better for x or y. They often don't ask which one actually answers what you're looking for. It's usually just about which one is more likely to lead you to make a purchase. They often ask you which screenshot of a search you prefer, yet both screenshots are completely identical.
It's very clear that it's just managers manipulating the surveys to make sure their new product is chosen as the "better" option.
Lol I also use Opinions Rewards and I always try to be as unhelpful as possible.
Maybe I'm naive but I answer honestly in hopes that they'll actually make improvements.
I do this with captchas! If I have time and am in the mood, I keep checking bad boxes (yet some that at least resemble what they want like chimney = hydrant, road crossing = stairs, motorcycle = bicycle etc.) and eventually it will accept my bad input. It works, obviously, as I was served many caotchas where there was big picture of motorcycle.divided into squares and I had to check squares where there is bicycle. Or the same with I believe bus and hydrant - one thing on the image but completely other that I had to check.
I'm absolutely certain that some software engineers somewhere love me! :-D
https://xkcd.com/1838/
Garbage in = garbage out
Part of Google's problem has always been giving weights to a site's popularity via quantity of cross-linking, as well as user clickthrough rate.
These are easy to abuse, and short of a massive army of trusted moderation to cut the wheat from the chaff, will trend towards garbage SEO content or the biggest commercial players.
So optimizing the algorithm makes sense in that vein, but the problem is that it has no capability (including with AI) an ability to distinguish between high-quality facts and popular nonsense. So a garbage opinion piece in PC World about air purifiers will almost certainly pop above a less-popular blog from a passionate hobbiest.
The only real way to solve this is some degree of trust system, to insure that high uality, factual material is both of those. And that's not the kind of thing easily automated.
It's part of the reason Reddit was such a useful place, the smaller communities and/or the tightly moderated ones, were able to curate (relatively) high quality user-made text content in a way that pre-aggregated it for easier searchability via indexers.
I think there's a triple constraint at work here. Any given ranking mechanism cannot be all three of simple/transparent¹; mechanical; and resistant to gaming. Naïve PageRank more-or-less ignores gameability, and Google's actual less-naïve (but also much less transparent) ranking algorithms are clearly losing the battle; Reddit moderation is explicitly non-mechanical and somewhere between relatively and utterly opaque, but is much harder to game. (At least via first-order methods. Suborning moderation is of course a grand tradition on the site.)
My prediction (with very low confidence, as befits rank speculation) is that we'll end up with islands of manually-curated content tightly-bound by informal, personal trust relationships, floating within a nigh-infinite ocean of complete shit.
¹ These are pretty close to the same thing, because if it's simple enough, people will be able to infer the behaviors even if they're not explicitly revealed. PageRank, of course, has both patents and academic writings about it, so it is explicitly transparent.
I've been looking to start establishing this for my kids, kind of a wall-off, highly curated internet experience with good auditing and best-effort privacy, while having an emphasis on auditability and only know-in-person access to thinks like chat and forums.
PMs with kids explicitly not e2e encrypted for that reason. Or perhaps E2E but with a backdoor for all relevant parents,
Can we start one?
At least site recommendation? I'm in dire need for clean internet for my kid. They're very good about staying within only allowed sites (trust but verify) and I don't need technical solutions for online chats or site blocking. Good ad block already achieves what we need.
Right now the internet is :
Wikipedia with caveats, tvtropes, xkcd, explain xkcd, game wikis, Harvard middle/modern English side by side Canterbury Tales.
No social media, no chats or forums, no intra-site search engines and definitely do not click on ads when they slip through from the game wiki type of sites.
Just a different viewpoint: I've always let my kids have full access to the entire internet. With the idea that it's going to be a significant part of their adult lives and the best time to learn skepticism and critical thinking is when you are young.
I try to talk to them about not trusting anything at face value and encourage them to talk about what they do online, even if it's not for kids. I try not to judge them for anything they look at but talk about the content in terms of why it exists and what the people who made it stand to benefit, etc. I advise them not to give out any personal information.
We limit their screen time, Homework and chores have to be done. And we encourage them to go out into the world, ride their bikes to parks etc. but when they do have screens there are no limits to what information they have access to.
I've had this policy since before they were old enough to read. With the idea that at some point they would be able to access the entire internet regardless of what I want. My parents had no idea what I was up to online, and could not have stopped me short of locking me in an unwired room. Now devices are even smaller and wireless signals are everywhere. I figured it would be hubris to assume I could control my kids online experience once they have a few years to learn tech.
Basically instead of a walled garden, I tried to have a transparent, peer relationship, where I'm just here to give advice and help them learn how to navigate the ever-changing cesspool that is the internet. At this point they are 11-14, so they are well past the point where I could control their Internet. And we are very much navigating it together. I feel like I would understand the Internet of today a lot less if I didn't have them to explain the tik tok memes to me.
My parents never succeeded at their limited attempts to control internet usage when I was younger. When I was a little kid and needed Mom to sort out dial-up that was one thing. But even by middle school I was more savvy than them. I doubt it'll be any different for newer generations -- if there's anything teens and tweens have been good at throughout history, it's getting around their parents' rules.
The one thing my parents could have done better imo would have been to educate us on potential dangers like grooming and interacting with strangers beyond kiddie "stranger danger" stuff. Nothing bad ever happened but in retrospect my younger sister was in a host of groomer-y situations that they didn't do enough to prevent imo. But they themselves weren't really knowledgeable enough to do that, which is a big problem even outside online spaces.
I'm envisioning trying to build something in the vein of (or on top of) Yunohost, like a commmunity in a box.
The hard part, as always, is community buyin.
I know many folks may not like to hear this but part of the problem is "Big Tech Capitalism".
Google should introspect for a while on why it became Google in the first place - organic and untainted search results that worked on "Content is King" principle. While that principle holds true even today in general, Google algorithms seem to have introduced some gate keeping which override it.
This sure looks like an opportunity for some enterprising entity to just shockingly eat Google’s lunch when it comes to the search business.
Honestly, making an algorithm that uses a better weighting of “indie” URLs, some qualitative language assessment, maybe linking from reputation-based discussions (a la stack exchange/Reddit-like sites) isn’t impossible to make.
Basically an algo that incorporates already existing human-created recommendation systems instead of pitting ads against them.
The quality content IS out there, and I still maintain that’s what’s gonna win out in the end… there just needs a slightly less greedy tool going mainstream to reinvigorate it.
https://kagi.com/
One of the major differences is being able to add your own custom weightings (or trust factor if you will) to every website. It's stupidly easy to ban AI image websites if you're annoyed by their results, and you can uprank fan wikis with better content than Wikia.
Second this.
Have a bit of work to transition myself over properly but I've enjoyed experimenting. Even thinking of paying so I can support what they're doing.
Is it really worth it? I’ve seen kagi mentioned consistently
I prefer it for the great majority of my searches. When I first started using it, I felt like it was basically on par with Google, but now it really feels like a big backward step when I open a new browser and accidentally do a Google search when I thought I was doing a Kagi search. I'm not sure if Google has gotten worse, if Kagi has gotten better, or if it's just various customizations I've made to my Kagi search results over time — it's quite possibly all three — but I don't think I could go back now.
Mind you, I am not a subscription kind of person. I don't do Netflix, Spotify, etc., etc., because I am basically allergic to the financial complexity that comes with them; I have to really get a lot of value from a product to be willing to subscribe, and I'm not willing to subscribe to more than about 3-4 services at a given time (I will cancel one to start a new one, even if they are totally different classes of product, just to keep my total subscription number as small as reasonably possible). But Kagi's service is worth it to me and I feel good about supporting them.
That being said, there are a couple of cases where I will use both Kagi and Google in conjunction:
I still use Google only for video searches or map searches.
Of course that depends on what you find wortwhile paying for. It is not like it magically finds stuff I can't find on Google, but the overall experience is just so much more frictionless with features that are focused on the users needs and not the advertisers, that it is easily worth the small price for me.
Have to figure out publicity but yes
I frequently google for movies & TV series and look for Rotten Tomatoes pages. I noticed today that the results' descriptions contain irrelevant information, often from other pages, whereas they used to contain the excerpt of the critic/audience review summary. I wonder if these changes Google made have screwed up SEO somehow.
Well it has had some decent improvement for my wife's shop.
I've always made sure to use decent SEO, good images, alternative tags, etc. It's a small Wordpress/WooCommerce store in the UK.
Most searches for her products are drowned in Temu, Etsy and other company products, but a random UK VPN and Private tab on Google has shown her products with the same searches to be on page 1. Yesterday it was 2 or 3 at best.
Bing is still crud though.
Edit: Just a note to say I do not keyword stuff, no hash tagging and there is not much in the way of off linking and linking back to social media. It's all in long form descriptive English.