5 ms·

It's really code red for the Web Google and the Web are tightly coupled. The old contract of if you wrote good content, it would rank highly on Google, was a c

It's really code red for the Web Google and the Web are tightly coupled. The old contract of if you wrote good content, it would rank highly on Google, was a convenient, organic economic one. You would be rewarded with visitors to your site. Google rewarded with better search results. Readers with good content. This unfortunately created a race to the bottom. The intense, years-long effort to DDOS Google search with spammy, low quality, but seemingly good looking SEO-curated content has destroyed the Google and Web experience. I'm not sure Google can ever truly keep up with the volume of spam and low quality. Google tried to react by putting more information on its search results page, but alas, this just meant the content creators valued their destinations that much less. The prevailing digital marketing firm wisdom created a tragedy of the commons and a crisis in quality web content. "Put a pop up here".. "Add 3 pages of boilerplate to mention all the right keywords" etc etc. When people really want to read informational text in a non-obtrusive format. ChatGPT does this simple thing well (informational text without the glaring headaches of random websites). So it wins in these contexts.

seydor 4y ago
The web is bloated. Hopefully openAI will pay directly the information providers. Which also means that the information will not make it to the public web. Which is just as well, after 30 years of giving out our information for free thinking we are contributing to some democratized web, we ended up with the web being 4 monopolies which compete for locking down access
- vineyardmike 4y ago
  > Hopefully openAI will pay directly the information providers It’d be nice if they tell the information providers they used the data. I think expecting them to pay is wishful thinking.
acdha 4y ago
> I'm not sure Google can ever truly keep up with the volume of spam and low quality. Are they really even trying? I see low-quality scraper domains which have ranked highly in their search results for years but never seem to be de-ranked despite just displaying content from GitHub or Stack Overflow. What those sites have in common is that they’re loaded with ad words ads, which suggests to me that there’s less willingness to act against them unless profits dip.
- doctor_eval 4y ago
  Yeah I don’t think they are trying any more. I received a Google search result the other day that was a word for word copy of a SO answer - it even had “stack overflow” in the page title. The spammers weren’t even trying to hide the fact. Nevertheless, the result ranked a couple of lines higher than the SO article it was a copy of. I mean, come on.
  - acdha 4y ago
    The amazing part is that you’ll probably get the same site a month or two from now. I am disappointed that Google doesn’t have a blocklist feature & you’d think they’d want the training data (although I imagine abuse could be a severe challenge).
    - geysersam 4y ago
      They could at least let their users blacklist sites from their personal search results. There are extensions that do this though, thank God.
      - alanning 4y ago
        Which extensions would you recommend for this?
        oefnak 4y ago
        uBlock origin in combination with a list like this one: https://github.com/levymetal/filter-lists https://github.com/levymetal/filter-lists
        numpad0 4y ago
        uBlacklist allows manual blacklisting from Google Search. Somewhat useful for me.
    - wolpoli 4y ago
      One possible reason that Google doesn't offer a blocklist is that Google only provides a limited number of results per query. Given Google typically shows more than one result from the same domain, so with a blocklist, it would actually have to devote resource to find results from different domains.
thrashh 4y ago
I don’t think SEO destroyed the web Rather, it seems people prefer videos There’s now plenty of YouTube videos on a lot of topics. Wikipedia cleans house with the rest Honestly there is access to way more good content today than 10 or 20 years ago. Just maybe not in the same form 10 or 20 years ago
- unsupp0rted 4y ago
  I sometimes have to support Windows machines. When I Google things for Windows that require 3 bullet-point sentences to answer, inevitably I get a 12-minute video of an Indian guy with an accent that prevents me from watching at 2x audio comprehensibly, taking 6 minutes to tell me how common the problem is and 6 minutes teaching me how to download and install spyware that does what I want and much more.
  - P5fRxh5kUvp2th 4y ago
    I love when you google for something and in order to stretch the content out 80% of it is completely unuseful, barely relevant, content. "How do you change the sandpaper in a sander" and most of the article is explaining why sander X is the best sander on the market.
- dageshi 4y ago
  Yeah, the hobbyists that used to make niche sites on subjects they loved have all moved to youtube for publishing their content. They were the people who made the web truly great back in the day.
- ilaksh 4y ago
  I wonder if the OpenAI models have video transcripts in their training. Or if GPT-4 or -5 (whatever upcoming model) will understand video or visual information and it's relation to text. Over the next few years, with grounded language understanding and other capabilities, no one will be able to pretend that these systems aren't intelligent. I mean, some people always will, but it's going to be a very small percentage. Right now I am guessing we are about 50-70% of people convinced that this stuff is cognitive rather than regurgitative. But also I think the abilities of these models clarify the nature of intelligence and the relationship between intelligence, compression and computation.
SkyPuncher 4y ago
Surely ChatGPT will be subject to the same issues that Google is. * Right now, ChatGPT has a disclaimer that it doesn't know much about the world since 2021. This implies much of its training set excludes recent data. * ChatGPT is brand new. Nobody has had a chance to reverse engineer it or game it. * Most "search" is relatively simple queries. There's only so much you can do to differential a bunch of sites that offer the exact same facts. I suspect, ChatGPT will reduce latency in its training data and people will figure out how to rank well against the algorithm. Then, it will be no different than just another search engine.
- JacobThreeThree 4y ago
  Yeah, if anything Google is better positioned thanks to their years of experience dealing with spammers who try everything to game the system.
- nprateem 4y ago
  It's results are based on probabilities. It must surly be much harder to game except perhaps for very new words/slang etc.
- spaceman_2020 4y ago
  It’s a matter of scale. What percentage of market share does chatGPT have to take away from Google for Google shareholders to panic? What does a 10% drop in market share does to Google’s stock price? What does a steep drop do to employee compensation? What does a drop in employee compensation at Google do to the rest of the tech industry?
- mejutoco 4y ago
  That is a very insightful comment. I wonder if openai would offer the different checkpoints for the models, so at least the 2021 would permanently be unaffected by attempts to game it.
amelius 4y ago
Perhaps OpenAI should provide an API such that web-browsers can instantly look up any text and see if it was created by one of their GPT models. Then browsers can highlight and mark the text as such. Perhaps it should even be a law that any AI should record its own output for later queries.
- beagle3 4y ago
  That’s about as effective as checking if a picture was made by DALL-E2 - for a month or two, it might work; but soon enough there will be 5 commercial ones and soon after you would be able to self host.
bottlepalm 4y ago
It’s a death to the part of the web with information to drive ads. It’s a rebirth of a web before everything was crazy ad driven.
- visarga 4y ago
  Interesting perspective.
tangjurine 4y ago
I was thinking this a while back, but Google should have put a significant fraction of its advertising revenue towards the websites it directs to, similar to YouTube. I keep hearing that the web is dying and that organic content keeps being replaced by ad farms. Maybe if it was possible to make a living off of small site content like YouTubers do the web would be more resilient and people would use it more instead of going to wall gardens like Facebook and Instagram, and Google would have had more cash in the long run. The current state of the web is like if every YouTube video needed to have a sponsored ad in it to make money, and Google put its own ad on top of that.
vicentwu 4y ago
ChatGPT trained by current methods as a search tool might have the "spamdexing" problem too. People can still find ways, which might not be as easy as it is now, to generate content and make them be included in the training data. But I think LLMs like GPT can be used as a great weapon against spamdexing because of their ability to "understand" text, which can help improve nowadays search engines, like Google, a lot by applying them as a filter when the spiders are crawling the web content.

seydor 4y ago

The web is bloated. Hopefully openAI will pay directly the information providers. Which also means that the information will not make it to the public web. Which is just as well, after 30 years of giving out our information for free thinking we are contributing to some democratized web, we ended up with the web being 4 monopolies which compete for locking down access

vineyardmike 4y ago
> Hopefully openAI will pay directly the information providers It’d be nice if they tell the information providers they used the data. I think expecting them to pay is wishful thinking.

acdha 4y ago

> I'm not sure Google can ever truly keep up with the volume of spam and low quality. Are they really even trying? I see low-quality scraper domains which have ranked highly in their search results for years but never seem to be de-ranked despite just displaying content from GitHub or Stack Overflow. What those sites have in common is that they’re loaded with ad words ads, which suggests to me that there’s less willingness to act against them unless profits dip.

doctor_eval 4y ago
Yeah I don’t think they are trying any more. I received a Google search result the other day that was a word for word copy of a SO answer - it even had “stack overflow” in the page title. The spammers weren’t even trying to hide the fact. Nevertheless, the result ranked a couple of lines higher than the SO article it was a copy of. I mean, come on.
- acdha 4y ago
  The amazing part is that you’ll probably get the same site a month or two from now. I am disappointed that Google doesn’t have a blocklist feature & you’d think they’d want the training data (although I imagine abuse could be a severe challenge).
  - geysersam 4y ago
    They could at least let their users blacklist sites from their personal search results. There are extensions that do this though, thank God.
    - alanning 4y ago
      Which extensions would you recommend for this?
      - oefnak 4y ago
        uBlock origin in combination with a list like this one: https://github.com/levymetal/filter-lists https://github.com/levymetal/filter-lists
      - numpad0 4y ago
        uBlacklist allows manual blacklisting from Google Search. Somewhat useful for me.
  - wolpoli 4y ago
    One possible reason that Google doesn't offer a blocklist is that Google only provides a limited number of results per query. Given Google typically shows more than one result from the same domain, so with a blocklist, it would actually have to devote resource to find results from different domains.

thrashh 4y ago

I don’t think SEO destroyed the web Rather, it seems people prefer videos There’s now plenty of YouTube videos on a lot of topics. Wikipedia cleans house with the rest Honestly there is access to way more good content today than 10 or 20 years ago. Just maybe not in the same form 10 or 20 years ago

unsupp0rted 4y ago
I sometimes have to support Windows machines. When I Google things for Windows that require 3 bullet-point sentences to answer, inevitably I get a 12-minute video of an Indian guy with an accent that prevents me from watching at 2x audio comprehensibly, taking 6 minutes to tell me how common the problem is and 6 minutes teaching me how to download and install spyware that does what I want and much more.
- P5fRxh5kUvp2th 4y ago
  I love when you google for something and in order to stretch the content out 80% of it is completely unuseful, barely relevant, content. "How do you change the sandpaper in a sander" and most of the article is explaining why sander X is the best sander on the market.
dageshi 4y ago
Yeah, the hobbyists that used to make niche sites on subjects they loved have all moved to youtube for publishing their content. They were the people who made the web truly great back in the day.
ilaksh 4y ago
I wonder if the OpenAI models have video transcripts in their training. Or if GPT-4 or -5 (whatever upcoming model) will understand video or visual information and it's relation to text. Over the next few years, with grounded language understanding and other capabilities, no one will be able to pretend that these systems aren't intelligent. I mean, some people always will, but it's going to be a very small percentage. Right now I am guessing we are about 50-70% of people convinced that this stuff is cognitive rather than regurgitative. But also I think the abilities of these models clarify the nature of intelligence and the relationship between intelligence, compression and computation.

SkyPuncher 4y ago

Surely ChatGPT will be subject to the same issues that Google is. * Right now, ChatGPT has a disclaimer that it doesn't know much about the world since 2021. This implies much of its training set excludes recent data. * ChatGPT is brand new. Nobody has had a chance to reverse engineer it or game it. * Most "search" is relatively simple queries. There's only so much you can do to differential a bunch of sites that offer the exact same facts. I suspect, ChatGPT will reduce latency in its training data and people will figure out how to rank well against the algorithm. Then, it will be no different than just another search engine.

JacobThreeThree 4y ago
Yeah, if anything Google is better positioned thanks to their years of experience dealing with spammers who try everything to game the system.
nprateem 4y ago
It's results are based on probabilities. It must surly be much harder to game except perhaps for very new words/slang etc.
spaceman_2020 4y ago
It’s a matter of scale. What percentage of market share does chatGPT have to take away from Google for Google shareholders to panic? What does a 10% drop in market share does to Google’s stock price? What does a steep drop do to employee compensation? What does a drop in employee compensation at Google do to the rest of the tech industry?
mejutoco 4y ago
That is a very insightful comment. I wonder if openai would offer the different checkpoints for the models, so at least the 2021 would permanently be unaffected by attempts to game it.

amelius 4y ago

Perhaps OpenAI should provide an API such that web-browsers can instantly look up any text and see if it was created by one of their GPT models. Then browsers can highlight and mark the text as such. Perhaps it should even be a law that any AI should record its own output for later queries.

beagle3 4y ago
That’s about as effective as checking if a picture was made by DALL-E2 - for a month or two, it might work; but soon enough there will be 5 commercial ones and soon after you would be able to self host.

bottlepalm 4y ago

It’s a death to the part of the web with information to drive ads. It’s a rebirth of a web before everything was crazy ad driven.

visarga 4y ago
Interesting perspective.

tangjurine 4y ago

I was thinking this a while back, but Google should have put a significant fraction of its advertising revenue towards the websites it directs to, similar to YouTube. I keep hearing that the web is dying and that organic content keeps being replaced by ad farms. Maybe if it was possible to make a living off of small site content like YouTubers do the web would be more resilient and people would use it more instead of going to wall gardens like Facebook and Instagram, and Google would have had more cash in the long run. The current state of the web is like if every YouTube video needed to have a sponsored ad in it to make money, and Google put its own ad on top of that.

vicentwu 4y ago

ChatGPT trained by current methods as a search tool might have the "spamdexing" problem too. People can still find ways, which might not be as easy as it is now, to generate content and make them be included in the training data. But I think LLMs like GPT can be used as a great weapon against spamdexing because of their ability to "understand" text, which can help improve nowadays search engines, like Google, a lot by applying them as a filter when the spiders are crawling the web content.