[-] kevincox@lemmy.ml 44 points 1 week ago

They added telemetry. 100% of responses had internet access.

[-] kevincox@lemmy.ml 49 points 2 weeks ago

I am a touch screen enjoyer. At least in theory. I like having time to browse, look at pictures, easy access to customization options and most importantly no feeling of pressure. I am not spending a cashier's time and potentially blocking someone behind me (at least there is usually less of a line for the self-ordering).

However there are negatives for sure. My biggest annoyance is that these devices are often annoyingly slow and unresponsive. They just display a tiny bit of text and images, they should switch between screens at 60fps, not 2s per click. Also if I know what I want it is often faster to tell the cashier and let them enter the order (on their more expert-optimized and less laggy keypad).

[-] kevincox@lemmy.ml 44 points 1 month ago

Because some people want to filter it out. So it gets a label.

[-] kevincox@lemmy.ml 38 points 3 months ago

This is pretty clever. As I understand it.

  1. Because LLMs are slow most of them stream the response to the user.
  2. The response is streamed as text, but generated in tokens.
  3. This means that each "chunk" leaks the length of the text corresponding to the token.
  4. You can then use heuristics to guess the text of the response based on the token lengths.

This is a good reminder any time you are sending content in small chunks over an encrypted channel, many encrypted channels don't provide protection against size leaks by default.

It seems there are a few easy solutions to this:

  1. Send the token IDs (as fixed-size integers) over the network rather than the text.
  2. Pad the text representations of the tokens to a fixed length.
  3. Batch the tokens more (and maybe add padding) to produce bigger chunks and obscure individual token size.

These still all leak the approximate length of the response, but that is probably acceptable.

[-] kevincox@lemmy.ml 41 points 4 months ago

Yeah, "physical" in that the bits live under my control. Not like a separate disc per movie.

[-] kevincox@lemmy.ml 45 points 8 months ago

Yeah, people are getting really upset at Google/Mozilla here but SafeBrowsing is actually a very good service. I legitimately believe that it frequently prevents malware infections and phishing on a regular basis. It is also architected with a privacy-first approach that reveals very little data to Google. And the SafeBrowsing privacy policy is actually one of Google's very tight ones.

I think Mozilla made the right choice to enable it by default. They also make it fairly easy to disable this for advanced users under the "Deceptive Content and Dangerous Software Protection" setting. (No need to crack open about:config, disabling it is fully supported.)

I understand that this may be a controversial opinion.

[-] kevincox@lemmy.ml 56 points 9 months ago

The rant comment will be forever changed.

And dare I say improved.

[-] kevincox@lemmy.ml 49 points 10 months ago

This is https://www.hyrumslaw.com/.

Basically there are two types of breaking changes:

  1. The change may break something.
  2. The change breaks a contract of the code.

What you are experiencing with debugRepr() is that you have triggered 1. You have made a chance that may break a user. But you have not triggered 2 because the new output is still within the previous contract. What level of stability you want to uphold is up to you.

[-] kevincox@lemmy.ml 48 points 10 months ago

Not only that but the EU doesn't want to make it seem like people can come and go as they please. So they will make serious demands for rejoining.

[-] kevincox@lemmy.ml 37 points 10 months ago

The problem of which instance to host a community on is a big problem for Lemmy. Grouping is an interesting idea but it causes problems as now there are different mods and admins that control subsets of the community.

Picking a single "winner" and letting the others wither seems like the right approach and will probably happen naturally but if the original instance ever shuts down or struggles under the load you will have a mess to migrate to a new instance.

If Lemmy communities were decentralized it would make a huge difference. You could just have a single community but it could survive instances coming and going (as well as many other performance and resiliency benefits). But that would be a huge change to the underlying implementation of communities.

[-] kevincox@lemmy.ml 52 points 11 months ago

This proposal absolutely infuriates me. This is making it so that you won't be able to browse the web unless you are using "approved" hardware on an "approved" OS with an "approved" browser. You will have no freedom to control your computing. Even if your browser is open source it will barely matter because you won't be able to patch it, you will need to run the approved binaries.

Fuck off and let me use the software I want.

This is SafetyNet from Android. You won't be able to access your bank, your movies, your anything unless you are using hardware and software that is controlled by billion dollar corporations.

[-] kevincox@lemmy.ml 43 points 11 months ago

A lot of people don't understand that there is nothing magical about a written contract with a signature. If you agree to something you have a contract. It doesn't matter if it is written, spoken, gestured or anything else. Written contracts with signatures are often preferred because it is very clear that there was an agreement and what was agreed to. But just about any method of agreeing is just as binding.

2
submitted 2 years ago by kevincox@lemmy.ml to c/rss@lemmy.ml
1
submitted 2 years ago by kevincox@lemmy.ml to c/decentralized@lemmy.ml
1
submitted 2 years ago by kevincox@lemmy.ml to c/rss@lemmy.ml
1
submitted 2 years ago* (last edited 2 years ago) by kevincox@lemmy.ml to c/rss@lemmy.ml

I know the Email isn't everyone's favourite RSS reader but it works really well for me. I wasn't happy with any of the existing services so I started my own.

https://feedmail.org is a low-cost RSS-to-Email service with nice clean templates. I'm happy to answer any questions.

1
submitted 2 years ago by kevincox@lemmy.ml to c/web@lemmy.ml

This is a service I created to consume RSS feeds via email. This has been my preferred way to consume RSS for a while but I never found a service that I was really happy with and no self-hosted tool easy enough to manage.

So I created FeedMail mostly for myself but decided to share with others. I would appreciate feedback and any questions you have.

2
Backups with IPFS (kevincox.ca)
submitted 3 years ago by kevincox@lemmy.ml to c/ipfs@lemmy.ml
3
submitted 3 years ago by kevincox@lemmy.ml to c/ipfs@lemmy.ml
4
submitted 3 years ago by kevincox@lemmy.ml to c/nixos@lemmy.ml
1
submitted 3 years ago by kevincox@lemmy.ml to c/openstandards@lemmy.ml
view more: ‹ prev next ›

kevincox

joined 3 years ago
MODERATOR OF