xvello

This is the fallacy that LLM-boosters are spreading: "AI can never fail, it can only be failed". But when the fail ratio is so high, it cannot be only an execution problem. It is proof that the LLM technology is over-hyped way over the small niche where it's cost/benefit ratio can be positive.

https://edition.cnn.com/2025/03/27/tech/apple-ai-artificial-...

SEEKING VOLUNTEERS: https://letsblock.it - https://github.com/letsblockit/letsblockit/

A companion project for uBlockOrigin that curates a corpus of content blocking templates, and provides the server to create you personal list of content blocking rules. The official instance just hit 800 active lists and a lot of template suggestions have been filed recently.

The easiest way to contribute is to create new templates, fix or extend existing ones. You need to learn the uBlockOrigin syntax and how to properly target the right elements, happy to mentor! See recent PRs for examples and https://github.com/letsblockit/letsblockit/blob/main/data/fi... for documentation.

The server itself is built with Go and HTMX, it's pretty low-maintenance, but there's interesting improvements if you want to toy with it (need to open issues for these).

Install uBlockOrigin and create custom content blocking rules with https://letsblock.it/filters/search-results

Heya, project maintainer here! Adding Kagi support in https://letsblock.it/filters/search-results was an easy lift, and it allows to use the filtering presets that are available at the bottom of the template. It also allows you to bring your blocklist in and out as you transition between search engines.

Project maintainer here, thanks for the feedback. For context, this template was created two years ago when that creator's videos filled almost 20% of the Nebula timeline and "How do I hide his videos?" questions were a monthly occurrence on forums.

I agree in retrospect that it was a petty move, and I deleted the default values today, as the project now works fine without default values.

Project maintainer here, thanks for the feedback. For context, this template was created two years ago when that creator's videos filled almost 20% of the Nebula timeline and "How do I hide his videos?" questions were a monthly occurrence on forums.

I agree in retrospect that it was a petty move, and I deleted the default values today, as the project now works fine without default values.

Project maintainer here, thanks for the feedback! It's indeed all rendered server-side with progressive enhancement, as javascript was intended to be used. Happy to read that this work is not in vain!

My experience is that these are usually caused by bad-quality frontend code that does not properly handle errors around bad-quality analytics SDKs, and breaks in subtle ways when these SDKs throw exceptions.

The e-commerce industry is obsessed with tracking and optimizing their conversion rate, so a lot of them have a lot of manual instrumentation of their "user journeys". Imagine they emit an analytic event after you fill your address, and before unlocking the next checkout step. If uBO blocks the analytics event submission and the SDK throws an exception, that next checkout step won't unlock unless they implemented basic error handling.

https://letsblock.it : it allows you to create your own content blocking rule list from a corpus or community-maintained templates. It allows you to hide pinterest and stackoverflow clones from search results, remove shorts and upcoming streams from youtube, and many more. The project is now two years old and sustaining a slow but steady growth with an active community.

Agreed! That would be a great template to have in letsblock.it's corpus! Contributions are welcome if you have a ruleset to get us kickstarted.

Heya, you can take inspiration the rules I posted in https://news.ycombinator.com/item?id=35676840 for hiding the second line and the separator. I messed up the last newline, the bottom block is the following two lines:

    news.ycombinator.com##html:not([op="item"]) tr.spacer + tr:not(.athing):remove()
    news.ycombinator.com##html:not([op="item"]) tr.spacer + tr.spacer:remove()

I'll make it a letsblock.it template one day.

Project maintainer here, thanks for posting it!

As described in https://letsblock.it/help/about, this project is a home for rules that cannot be included in the default content blocking lists, because everyone has their own definition of low-value content.

Our goal is to curate a list of templates that allow you to tune up the signal/noise ratio and avoid distractions. You can just copy-paste rules in your uBlockOrigin / AdGuard settings, but the project is best used by signing-up to create your customized rule list. This gives you automatic updates when rules are improved and updated by the community, and the ability to use that list on all your devices.

I self-host several services on a cheap VPS, but I'm paying a yearly subscription to wallabag.it.

I could very easily self-host it, but I chose to help fund its development though that subscription.

Nope, let's not reward them with money. Lobby to get this idiotic TLD shut down. Otherwise someone will grab .pdf and .png soon.

My uBlockOrigin rules to filter posts by regexp, to filter out Twitter drama and AI stuff. The last two rules remove the second row + the spacer.

news.ycombinator.com##html:not([op="item"]) tr.athing span.titleline > a:has-text(LLM):upward(tr):remove() news.ycombinator.com##html:not([op="item"]) tr.athing span.titleline > a:has-text(/OpenAI\b/):upward(tr):remove() news.ycombinator.com##html:not([op="item"]) tr.athing span.titleline > a:has-text(/\bstable\Wdiffusion\b/i):upward(tr):remove() news.ycombinator.com##html:not([op="item"]) tr.athing span.titleline > a:has-text(/\bdall-e\b/i):upward(tr):remove() news.ycombinator.com##html:not([op="item"]) tr.athing span.titleline > a:has-text(/\bChatGPT/i):upward(tr):remove() news.ycombinator.com##html:not([op="item"]) tr.athing span.titleline > a:has-text(/\bGPT/):upward(tr):remove()

news.ycombinator.com##html:not([op="item"]) tr.athing span.titleline > a:has-text(/\bTwitter\b/):upward(tr):remove() news.ycombinator.com##html:not([op="item"]) tr.athing span.titleline > a:has-text(/\bMusk\b/):upward(tr):remove()

news.ycombinator.com##html:not([op="item"]) tr.spacer + tr:not(.athing):remove() news.ycombinator.com##html:not([op="item"]) tr.spacer + tr.spacer:remove()

Gandi has been bought out and has started turning for the worse. They already breached contracts with UK customers, so I am moving my domains to the Norwegian company https://domainname.shop/ before they find a way to shaft me.

Source: https://news.ycombinator.com/item?id=35080777

These rules would be a great addition to the low-value content filters we already maintain at https://letsblock.it, and I think a lot of users would benefit from having such a template available.

I don't use booking.com a lot, but would you be interested in contributing to the project and maintain this template when you find rules to update?

Hit me up if you have any question.

One of the points of HTMX is to merge back all of the state to the backend, instead of spreading it between backend and frontend.

For authenticated users, just store a user pref record on your database. For state that is not linked to a user account, use the standard session mechanism of your backend. Alternatively, just use good-old functional cookies, with the Set-Cookie HTTP header. Set them to be available to javascript if needed.

> seems like the multiplier has been getting higher lately

The SSD now being soldered to the system board has of course ZERO correlation with that price gouging.

Agreed! I switched from Spotify to Deezer one year ago.

The UX getting more and more clunky due to the podcast push really soured me, but the fact that part of my monthly subscription went to fund that person was what got me to click on the cancel button.

The switch to Deezer was really positive: the Flow auto-generated playlist is a lot more relevant to me than the Spotify weekly mixes.

Agreed. In the meantime, I use the following uBlockOrigin rules:

    news.ycombinator.com##html:style(filter:invert(100%) hue-rotate(180deg))
    news.ycombinator.com##body:style(background: white)
    news.ycombinator.com##div.toptext:style(color: black)
    news.ycombinator.com###hnmain td[bgcolor="#000000"]

It exists already, as a uBlockOrigin rules generator: https://letsblock.it/filters/amazon-products

You just need to write the regexp.

I have been using Kagi for 6 months now, and really enjoy its clean interface and relevant results. Being able to boost the ranking of relevant domains and downgrade the low-quality ones is a game-changer!

To remove pinterest and slackoverflow mirrors from all search engines, I use https://letsblock.it/filters/search-results

I only play single-player and couch co-op games, so I don't pay for PS Plus. As for the game prices, sales are less massive than on Steam but still frequent. Just create your watchlist on psprices.com.

The only thing I really miss is modding, especially quality-of-life improvement mods.

> I don't understand the market for a device like this at this cost.

I am a software engineer and OSS enthusiast, I operate servers and my home automation system, but I gave up on PC gaming years ago. After a long day of fighting distributed systems, I want my gaming rig to just work. That's why I bought the first PSVR: because Sony offers me a system where I don't have to fiddle with drivers and windows updates.

Price is too steep indeed, but it will go down as the production ramps up. PSVR launched at 400€ in 2016, and I bought it at 200€ two years after that.

While you're in your settings, add these three rules for a quick & dirty dark mode:

    news.ycombinator.com##html:style(filter:invert(100%) hue-rotate(180deg))
    news.ycombinator.com##body:style(background: white)
    news.ycombinator.com##div.toptext:style(color: black)

Hear hear! I use Tiny Tiny RSS and make extensive use of its post filtering abilities. Post with Twitter or Musk in the title go to /dev/null

Since Elastic spitefully patched all of their client libraries to fail if the server is not a "genuine" ES server, I don't see what good a drop-in replacement with protocol compatibility would do.

Go client: https://github.com/elastic/go-elasticsearch/blob/3985f2a1554...

Python client: https://github.com/elastic/elasticsearch-py/commit/e72aa3e24...

FYI, I have such a call to maintainer [1] for a "Twitter cleanup" template similar to the "Google search cleanup" one [2]. There is definitely a need, but it seems over my current carrying capacity.

For Meta's properties, I would wait for the maintainer to have a track record of contributions before being comfortable opening that door.

[1]: https://github.com/letsblockit/letsblockit/issues/269

[2]: https://letsblock.it/filters/google-search-cleanup

I'm interested to understand why you find this provocative.

My ambition is to build a dependable service that will be up for several years to come. For that to happen, I need to limit myself to a scope I am comfortable maintaining in the long run, instead of burning out under scope creep.

Until the maintainer team grows, I would be on the hook for updates and fixes, and I refuse to create an account on any of that company's sites.