Discord: @azalty

Steam: https://steamcommunity.com/id/azalty/

Lemmy is full of dictatorship lovers and political censorship. I’m glad I’ve left Reddit for this 👀

  • 2 Posts
  • 65 Comments
Joined 2 years ago
cake
Cake day: July 21st, 2023

help-circle




  • azalty@jlai.lutoPrivacy@lemmy.ml*Permanently Deleted*
    link
    fedilink
    English
    arrow-up
    0
    ·
    3 months ago

    Again, just relatively common sense.

    Postfix etc… isn’t that hard to run. I selfhost a mailcow installation, which is considered “full” and “expensive to run” and it’s still really fine to run. Not expensive when scaling up to multiple users, the most important is the initial performance cost per instance.

    We’re mostly paying for salaries and R&D I believe. They’re developing other services and stuff, but they could easily charge less and still be profitable, they just wouldn’t have this much reserves.

    Perhaps they don’t realise that by lowering their subscription costs, they could get more users… like me








  • The website’s article completely bases itself on https://duke.hush.is/memos/6/

    Both are spreading FUD about churning. The 2 wallets method is also churning, but “easy” version because you’re separating the results. The end result is still dangerous: you could spend 2 of your “Outgoing” coins in the same transaction, which is really easy to identify, provided both coins are from the same source.

    I’d argue the stastical risk of establishing a trace thanks to a serie of churns is extremely unlikely. Nobody will churn 6+ times anyways, and the advantages of churning far outweighs the ones of not churning.

    Sadly, they also don’t state anything about how that “MAP Decoder Attack” works. After searching, it’s pretty interesting, and we come onto the pretty well known issue that decoy distribution isn’t perfect.

    FCMP will fix this. In the meantime, please churn for any sensitive transaction. Either method (2 wallets or classic churning) works and the effects are always good, even if other mistakes are made (except if sweeping all).

    I have to say I haven’t really checked much of Rucknium’s work, but it’s pretty damn precise. Guy knows what he’s talking about. Our next big threat will be malicious remote nodes, like he said!