Home » Open thread 10/23/2024

Comments

Open thread 10/23/2024 — 20 Comments

  1. Try DuckDuckGo. I only use Google for the Maps function, and only because the succession of characters who have owned MapQuest allowed it to rot.

  2. I use DuckDuckGo.
    Advanced Google Search is superior for finding words or phrases from a specific website. But for generic searches, DuckDuckGo is fine.

    Love those milkweed pods. Rural NE memories….

  3. I’m sure that each one of us might have some issue–large or small–which we could be “deeply concerned” about.

    I see that spokesman John Kirby says that the White House is “deeply concerned” about the leak of the Israeli plans to attack Iran.

    In other words, supposed “feelings,” instead of “actions.”

    The real question is, what are they going to do about this leak?

    A leak which, reports say, has very likely come from an Administration official who was already considered pro-Iranian.

    Are they going to pinpoint who leaked these plans, and are they going to arrest, frog march them out of the Pentagon, and to prosecute them with all the vigor and harshness that they have applied to the supposed January 6th “Insurrectionists.”

    Somehow, I doubt it.

  4. Just another open-thread comment about something I read.

    Scott Yenor, who works for the Claremont Institute, has written an article entitled “Destroying the Higher Education Machine” (https://tinyurl.com/2u9vm8hb).

    Well, at least somebody’s talking about it. That’s a start. I still think it’s a fantasy, but most Republican state governments don’t realize that they have the power to drastically reform their states’ universities. They’re afraid to do it, so the thought never becomes fully formed. Who knows, maybe the Claremont Institute can start an intellectual and political trend.

  5. I use brave

    Weve known about tabatatabai for 14 months that tells you are on her side

  6. Mike Plaiss on October 23, 2024 at 10:40 am said:

    So, everyone knows about this, but what’s the alternative?

    https://thefederalist.com/2024/10/22/when-did-google-search-become-totally-useless/

    I’ve heard the competitors are nearly as bad. Anyone have a search engine they like?
    _______________________________________________________________

    I’ve been using Luxxle (https://luxxle.com/) and am pretty satisfied with it.

    Can’t say that I’ve done a thorough analysis of the search engine, but here’s a link to a short review that covers the basics: https://tinyurl.com/338cbjsx

  7. Cornflour,

    Won’t happen until all the current faculty leave and even then there’s been 20 years of new PhDs being produced in the same mold.

    When I started, yes the faculty was predominantly Democrat. However the vast majority of those people were classic liberals. As they retired they were replaced by radicals. Many of those people I’m still in contact with and are as appalled at the changes as I am.

  8. I’ve read that some people have disputed DuckDuckGo’s claims to maintaining users’ privacy, so I posed the question to the Perplexity AI LLM. Below, I’ve copied the most relevant paragraph from Perplexity’s reply, along with citations.

    “However, recent controversies have raised questions about the extent of DuckDuckGo’s privacy practices. Notably, it was revealed that DuckDuckGo allows certain Microsoft trackers to operate on its platform due to a search syndication agreement with Microsoft. This arrangement means that while DuckDuckGo blocks trackers from companies like Google and Facebook, it does not block Microsoft-related trackers, which has led to criticism regarding its commitment to user privacy[2][5]. DuckDuckGo’s CEO acknowledged this issue and stated that the company is working to improve its privacy protections[2][5].”

    [2] https://cyberguy.com/privacy/duckduckgo-privacy-browser-caught-sending-tracking-data-to-microsoft/
    [5] https://www.wired.com/story/duckduckgo-microsoft-twitter-ft-bush-assassination-whatsapp/

  9. Really do like that picture. The way the seeds themselves seem to be floating before you almost in the foreground reminded me of something, but I couldn’t remember what.

    After a little time on DuckDuckGo I realized it was this:

    https://images.app.goo.gl/YQ1eDDcvJD5P4Gz98

    Van Gogh’s Poet’s Garden at the Art Institute in Chicago. Hard to see in a picture of a painting, but when you’re standing in front of it the little white/blue flowers at center/bottom seem very three dimensional, like fireflies hovering off the surface or something.

  10. Am getting a lot of “AI Overview” returns with Google search now. Am no big fan of Google search, but it’s what I mainly use—sometimes resorting to Bing & DuckDuckGo for different results. Sometimes get better results when I change to – say the News tab from the All tab and/or change the Anytime drop down to past Month.

    Ran a blog a few years ago (Linux Newbie since 1996) and it was getting to be pretty popular—100-125 visits a day, which was good for me & Linux. Then some elections were coming up, and my visits dropped to 0-10 a day. Posts that had been on the Google Front Page were no longer showing up. Hey, Linux isn’t that popular, and it basically has some 600+ versions (aka Distros) of the Linux OS – so a post on Porteus Linux can certainly end on that Front Page. However, I also started adding the occasional Political post—favoring REPs and bashing DEMs. Got mad at Google after the drops, but still used their search engine even though I knew they favored DEMs.

    Have also started using various AI chatbots for my some of my searches now…

  11. Chatbots can suffer some of the same… ahh… predispositions that search engines can if they’re trained on garbage or biased information. Nothing is immune from human failings if it uses data created and maintained by human beings.

    I hear Tulsi Gabbard finally joined the Republican party. Conversely, Liz Cheney has yet to join the Democrat Party, perhaps in an attempt to fool gullible Republicans (assuming there are many of those left)?

  12. I greatly appreciate this daily interlude of calm and grace amidst the chaos. I have been reading you for many years. Thanks. Mike Roark

  13. Nonapod on October 23, 2024 at 1:26 pm said:

    “Chatbots can suffer some of the same… ahh… predispositions that search engines can if they’re trained on garbage or biased information. Nothing is immune from human failings if it uses data created and maintained by human beings.”
    ______________________________________________________________

    David Rozado has done some analyses of LLM’s political biases.

    Here’s a link to one of his posts:
    https://davidrozado.substack.com/p/the-political-preferences-of-llms

    Rozado uses graphs to summarize his work, so it’s a fast read.

    Recommended for those interested in the topic.

  14. I grew milkweeds in my yard. Beautiful 3 ft plants with good orange flowers.
    So I took some seeds, as pictured, and distributed them in the wild, in sunny spaces near where I hunt woodcock.

    Not one sprouted.

  15. @Cornflour, that’s a great little article and the results are depressingly not very surprising to me. One thing of interest though is that the political preferences of pretrained LLMs seem to generally be relatively neutral and it’s only after the so called “supervised fine-tuning” phase that the leftward leaning biases start to appear.

    I also show in the paper that LLMs are easily steerable into target locations of the political spectrum via supervised fine-tuning (SFT) requiring only modest compute and customized data, suggesting the critical role of SFT to imprint political preferences onto LLMs.

    So what exactly is “Supervised File-Tuning”? I found this here.

    Supervised fine-tuning, involves adapting a pre-trained Language Model (LLM) to a specific downstream task using labeled data. In supervised fine-tuning, the finetuning data is collected from a set of responses validated before hand. That’s the main difference to the unsupervised techniques, where data is not validated before hand. While LLM training is (usually) unsupervised, Finetuning is (usually) supervised.

    “Labeled data” is raw data that has been annotated with tags or labels to provide context and meaning. So this so called labeled data seems to be fairly influential in establishing these biases. It’d be interesting to see what is being labled with what.

    At any rate, David Rozada isn’t willing to conclude that Supervised fine-tuning is the primary cuprit in the emergent biases.

    Unfortunately, my analysis cannot conclusively determine whether the political preferences observed in most conversational LLMs stem from the pretraining or fine-tuning phases of their development. The apparent political neutrality of base models’ responses to political questions suggests that pretraining on a large corpus of Internet documents might not play a significant role in imparting political preferences to LLMs. However, the frequent incoherent responses of base LLMs to political questions and the artificial constraint of forcing the models to select one from a predetermined set of multiple-choice answers cannot exclude the possibility that the left-leaning preferences observed in most conversational LLMs could be a byproduct of the pretraining corpora, emerging only post-finetuning, even if the fine-tuning process itself is politically neutral. While this hypothesis is conceivable, the evidence presented in this work can neither conclusively support nor reject it.

  16. In a NBC News interview VP Kamala asked about religious exemptions in her federal abortion bill she says ” …no concessions…” So every Baptist and Catholic hospital must perform abortions? Every doctor and nurse in the military must be willing to perform abortions or face dishonorable discharge? Will doctors face civil rights prosecutions if they refuse to participate?
    https://nypost.com/2024/10/22/us-news/kamala-harris-doesnt-want-to-make-concessions-to-get-abortion-legislation-through-congress-cannot-be-negotiable/

Leave a Reply

Your email address will not be published. Required fields are marked *

HTML tags allowed in your comment: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>