• 0 Posts
  • 1.66K Comments
Joined 2 年前
cake
Cake day: 2023年7月3日

help-circle







  • MacOS by default hides scroll bars. They’re big on form over function which I hate.

    Some people are just like that.

    I knew a couple that mounted their TV in a way that all the ports (eg: HDMI) were inaccessible. They just didn’t care that a big chunk of the TV’s functionality was now blocked. They didn’t want to see wires.


  • A lot of people making decisions are idiots, or are following the whims of idiots above them.

    Back in like 2017 a company I worked for made a mouse tunnel on their web UI. That’s where like you mouse over a menu, and that opens a sub menu. You mouse into that sub menu, and another menu opens. If at any point your mouse leaves this area, the whole thing closes. It’s shit. It’s been a known bad pattern since like the 90s.

    Product guy wouldn’t listen. Not sure if he didn’t care or didn’t understand. Either is bad.

    This happens all over. People don’t care. They don’t understand. They don’t listen to people that do. They have their own metrics and goals that are disjoint from actual value.




  • There are credible allegations that the AI companies are not merely scraping publicly available resources, but are also consuming content in violation of the terms of use / copyright law. Like, a site has a robots.txt file that says “no scrapers” and they scrape it anyway. People would be mad about traditional search doing that as well.

    Secondly, if a search service scrapes your site and then directs relevant users to it, that’s probably fine. Most websites want users to visit. A lot of AI stuff sucks up the content, and then the creators of that content get nothing. No users are sent there. The scraper hitting the site takes resources, and gives nothing back.

    Google has also gotten some flak for putting stuff on their own site instead of sending users to the source. Like you do a search and get a snippet on the google page, and you never click through to example.com/cool-stuff. Well, now the owner of example.com/cool-stuff doesn’t get the click. If they run ads, they get no credit. If they have metrics, they probably don’t see any visitors. If they have like forums, people are less likely to engage.

    If the “AI Search” includes links back to the source, that’s not perfect either. One, it’s kind of excessive to use an LLM to parse text when the origin site is already there and readable. If I search for “population of london”, you can just send me to a census website or even wikipedia. You don’t need to use a whole ass LLM. Two, as I touched on in the previous paragraph, users are less likely to click through if google is putting the core of the information right there (even if it’s not always accurate). It’s still lessening traffic to the origin site, and traffic is often the lifeblood of websites.

    Lastly, a lot of AI stuff is simply inaccurate or misleading. We’ve all laughed at the “use glue on your pizza” stuff or the “there are two Rs in ‘strawberry’” fuckups. If traditional search was really bad, like you type in “cat food” and you got a webpage that was all jewelry and “buy gold” scams, you’d be annoyed, too. That’s more like how search was before old google came about. There were a lot more low effort “SEO” hacks like putting a bunch of keywords in tiny print to fool the search indexer. Now google is the shitty old guard, but they have too much money and power to be easily replaced.

    That’s just off the top of my head. Scraping for AI isn’t the same as scraping to make a searchable index.