Post AnSoIsHI3HHhIYKNPM by LovesTha@floss.social
(DIR) More posts by LovesTha@floss.social
(DIR) Post #AnQRXHAvauHbJBKWrw by strypey@mastodon.nzoss.nz
2024-10-27T08:06:36Z
0 likes, 0 repeats
Proof positive that common web #search engines are now shadowbanning from their results. Banning what? Authors Publishers?Whole topics? Not sure.But I used a metasearch engines to search for this subdomain and keyword combo;https://monocles.de/search?q=disintermedia.substack.com+trained+moleThe top results were links to fediverse servers with a copy of my posts linking to the SubStsck post about #MOLE Training I was looking for;https://disintermedia.substack.com/p/invasion-of-the-mole-trainersNot a single link to the piece itself, nor to the original versions of my posts.
(DIR) Post #AnQTI9JUk3rqURFgMC by strypey@mastodon.nzoss.nz
2024-10-27T08:26:16Z
0 likes, 0 repeats
I was looking up my #MOLE training piece to post about a classic example of one of the things it describes;"you could probably rent out your trained moles, as long as you could convince people that the mole's erratic behaviour was the result of inadequate direction from the handler."Sure enough, here's a podcast called;"Artificial Intelligence Masterclass" ... with an episode title that starts;"AGI is Almost Here!"... and the actual topic is a Trained #MOLE. (1/3)
(DIR) Post #AnQTIEKk4qpu3ksFUm by strypey@mastodon.nzoss.nz
2024-10-27T08:26:16Z
0 likes, 0 repeats
David Shapiro, host of the "AI Masterclass" podcast, opens with claims like;"With the release of 01 Preview, we are pretty much a stone's throw from Artificial General Intelligence ..."... and ..."There was 1 story out there on Reddit where someone said 'Strawberry, or 01 Preview, was able to do my entire thesis in about an hour, where it took me a whole year. That's pretty significant."... and ..."This has an IQ of about 120... although the reasoning still fails in some cases."(2/3)
(DIR) Post #AnQTIGPmL63MVxFIci by strypey@mastodon.nzoss.nz
2024-10-27T08:26:16Z
0 likes, 0 repeats
David Shapiro, is either being paid to try to sell you a Trained MOLE, or he's a Useful Idiot for people who are. Do not take anything he says about "AI" seriously.(3/3)
(DIR) Post #AnQiLH0k4shYXpQcHA by strypey@mastodon.nzoss.nz
2024-10-27T11:14:50Z
0 likes, 0 repeats
Me:> Proof positive that common web search engines are now shadowbanning from their resultsIf anyone cares to repeat my experiment at home, I'd be curious to know if your experience differs from mine.
(DIR) Post #AnRlPU2k3mp38t1TOK by strypey@mastodon.nzoss.nz
2024-10-27T23:24:03Z
0 likes, 0 repeats
@tootbrute> i get way way better results from my searxng.arkadi.one instance than google.comI was using a Searx instance, see the OP.> no ai bullshit and other 'sponsored' contentAFAIK Searx is still meta-search that depends on the big engines, so you still get those things, just indirectly ; )
(DIR) Post #AnRloHzO0vzUzrHVgm by strypey@mastodon.nzoss.nz
2024-10-27T23:28:32Z
0 likes, 0 repeats
(2/2)We really need a decentralised search solution that goes deeper than Searx. How come YaCy never took off?Question for the fediverse search haters out there;How would you feel about a fully Free Code web search engine that used the fediverse as a source of links to crawl, to build a freely-licensed search index?#search #WebSearch #Searx #YaCy #AskFedi
(DIR) Post #AnRlxMZeKwvdp4ZPEm by strypey@mastodon.nzoss.nz
2024-10-27T23:30:10Z
0 likes, 0 repeats
(3/3)I've also had thoughts about a system where websites index their own pages, according to their own preferences for how searchable they want stuff on their site to be. They then make that index file available on an API endpoint equivalent to NodeInfo, so any web search tool can add it to its own index.Can anyone tell me why that wouldn't be practical?
(DIR) Post #AnRmLS6vP3ezwccYsq by strypey@mastodon.nzoss.nz
2024-10-27T23:34:28Z
0 likes, 0 repeats
@pre> I think it's more that they promote lots of junk that shouldn't be included than that they demote some stuffPossibly. But I'm assuming that your standard web search engine is just a Trained #MOLE minus the chat front-end. So it's possible that the limits they use to stop the MOLE putting on jodhpurs and goosestepping around also affects other non-mainstream content due to the law of unintended consequences (or centrist biases that conflate radical left with far right).
(DIR) Post #AnRmlrmKaqST55aFCC by 8petros@petroskowo.pl
2024-10-27T08:22:39Z
0 likes, 0 repeats
Substack is a walled garden. Could it be a reason?
(DIR) Post #AnRmlsvELFAecykrho by strypey@mastodon.nzoss.nz
2024-10-27T23:39:16Z
0 likes, 0 repeats
@8petros> Substack is a walled gardenIs it though? Yes, its a monolithic supersite that doesn't respect software freedoms. But they publish on the open web and email, and I think they could be convinced to do what WordPress and Ghost (and Meta's Chains) are doing and publish over ActivityPub too.> Could it be a reason?They make more money the more paying subscribers there are to the newsletters they publish, so why would they limit search?
(DIR) Post #AnSo07wBbbxmMbW7Bg by LovesTha@floss.social
2024-10-28T11:27:45Z
0 likes, 0 repeats
@strypey That sounds like a reasonable way to use public post data, but it still should be opt in (probably by server and user)
(DIR) Post #AnSo6PJhu0um0CVtM8 by LovesTha@floss.social
2024-10-28T11:28:57Z
0 likes, 0 repeats
@strypey Pre-indexed is just too ripe for abuse. The site map standard gives the best that makes any sense: a list of pages worth indexing and then allow the crawler to index them.
(DIR) Post #AnSoIsHI3HHhIYKNPM by LovesTha@floss.social
2024-10-28T11:31:09Z
0 likes, 0 repeats
@strypey I think the decentralized 'search' that makes sense is more of the Internet index like Yahoo was.
(DIR) Post #AnTky0vlgFveDlkAUq by strypey@mastodon.nzoss.nz
2024-10-28T22:28:29Z
0 likes, 0 repeats
@LovesTha> Pre-indexed is just too ripe for abuseHow?
(DIR) Post #AnTl4lAQgg1bMHomNk by strypey@mastodon.nzoss.nz
2024-10-28T22:29:43Z
0 likes, 0 repeats
@LovesTha> That sounds like a reasonable way to use public post data, but it still should be opt inReally? We need permission to see a link in a public post and add the root domain to a pool of websites to be indexed? Why?
(DIR) Post #AnTlDaIqRwm4GmxiwS by strypey@mastodon.nzoss.nz
2024-10-28T22:31:18Z
0 likes, 0 repeats
@LovesTha> the decentralized 'search' that makes sense is more of the Internet index like Yahoo wasI'm not sure what you mean. From what I remember, Yahoo was just a standard search engine. Are you thinking of something like DMoz, or its successor Curly? In what way is this decentralised?
(DIR) Post #AnTwwqdRfkwRRzNr72 by LovesTha@floss.social
2024-10-29T00:42:43Z
0 likes, 0 repeats
@strypey https://en.wikipedia.org/wiki/Yahoo_Directory before Yahoo search was Yahoo Directory.Decentralized would be a new concept applied to this style of directory/index
(DIR) Post #AnTx9UDKLTt9g1C4FU by LovesTha@floss.social
2024-10-29T00:45:01Z
0 likes, 0 repeats
@strypey Instances are run for humans, not robots. The opt in by instances would be by robots.txt.
(DIR) Post #AnTxDusx7jzVYSmBAe by LovesTha@floss.social
2024-10-29T00:45:50Z
0 likes, 0 repeats
@strypey SEO BS by putting fake data in the index. I'm not sure it would be effective, but it feels like a bad idea.
(DIR) Post #AnU0tKFS1uMnx4jjAe by strypey@mastodon.nzoss.nz
2024-10-29T01:26:55Z
0 likes, 0 repeats
@pre> I hadn't figured the'd have wired those things into their search rankings much yetAFAIK you've got the causality backwards. ChatGPT, and other Trained MOLEs you can chat with, are just a chat interface tacked onto the latest versions of the same old backprop algorithms. The ones that have been giving us search results and 'watch next' recommendations for years.https://betterwithout.ai/
(DIR) Post #AnYii2spQOtfWeQUCm by strypey@mastodon.nzoss.nz
2024-10-31T07:56:45Z
0 likes, 0 repeats
(1/?)Me:> How would you feel about a fully Free Code web search engine that used the fediverse as a source of links to crawl, to build a freely-licensed search index?@LovesTha> That sounds like a reasonable way to use public post data, but it still should be opt inAfter thinking about this some more, you're absolutely right. Opt-in in both directions. Here's what I have in mind.#search #WebSearch
(DIR) Post #AnYikvDuMRWUdZepm4 by strypey@mastodon.nzoss.nz
2024-10-31T07:56:47Z
0 likes, 0 repeats
(2/?)A human @mentions Bob in a post;Hi @Bob, this is Alice from a search engine project called Grassy. You've been nominated as a contributor of links to the Grassy index. Because members of the Grassy community have seen that you consistently post links to useful websites on your account.If you'd like to participate, just follow the Grassy bot at @eater@links.grass.org, and let it follow you back.
(DIR) Post #AnYil0bUMCM9K9jEiu by strypey@mastodon.nzoss.nz
2024-10-31T07:56:47Z
0 likes, 0 repeats
(3/3)If Bob follows @eater and it follows him, the root domain of any links he posts is added to a list for indexing by Grassy. Unless it's already on that list, or a list of sites that are already indexed, or a list of excluded sites (DataFarms etc).
(DIR) Post #AnYl0nILo1pMacQpfs by LovesTha@floss.social
2024-10-31T08:22:34Z
0 likes, 0 repeats
@strypey I like this flow. While it is higher friction, so you'll get less data to work with, you'll get more users too.