8 ms·

Plausible deniability. Design the system in a way that it'd pick up competitor links as spam (without directly blacklisting them explicitly) using fine-tuned

Plausible deniability. Design the system in a way that it'd pick up competitor links as spam (without directly blacklisting them explicitly) using fine-tuned training data. In case of media attention, tell it's an error and ML algorithms are blackboxes, and take action to "whitelist" those domains only when needed. Yeah, we buy it, Twitter.

tehlike 4y ago
Design a better one and there will be companies looking to acquire you. Problem solved, in a more constructive way.
- can16358p 4y ago
  Well it's not my job, though even by looking at spam accounts on Twitter I can easily pinpoint common patterns that can be extremely successful at preventing spam at Twitter, MUCH better than blacklisting domains. It's dead obvious that they do not want it. The reasons why is beyond me.
  - tehlike 4y ago
    "it's not my job". Someone whose job it is made a few mistakes. you must have made a shit ton of mistake too. Claiming otherwise is stupid. What you and i can pinpoint on twitter spam is independent of "everything is evil". Difference.
    - can16358p 4y ago
      Of course everyone does mistakes, but that's not the point. Twitter is just too powerful with all those engineers and access to ML to have this just as a mistake. Whoever is responsible for that flagging is either: - Incredibly incompetent and not knowing what they were doing (which I honestly don't want to believe) - Knew exactly what they were doing and it's intentional with a benefit of doubt with plausible deniability. Unintentionally blocking a whole domain "by mistake" falls into the first point above. Knowing track record of social media companies hiring good engineers and being evil, also knowing Mastodon being a competitor to Twitter, I don't think I need to tell much more.
registeredcorn 4y ago
I'm not saying your wrong, but that just sounds like an incredible amount of work for something that would presumably be discovered anyway, you know? Wouldn't it be much easier to simply blacklist and put out some bland statement about "not supporting 'hate speech'" or something? It doesn't matter if it's true or not, it just matters that they paint the picture of them opposing something "bad". Lying just seems a lot easier than going to all of that effort.
- can16358p 4y ago
  Maybe I didn't clarify much. It doesn't require more work than they are already presumably implementing for spam detection (unless I way too much overestimate big tech's abilities), it's a little touch to an already-being-implemented system with a bonus of being able to block competitors to some extent, without direct liability.

tehlike 4y ago

Design a better one and there will be companies looking to acquire you. Problem solved, in a more constructive way.

can16358p 4y ago
Well it's not my job, though even by looking at spam accounts on Twitter I can easily pinpoint common patterns that can be extremely successful at preventing spam at Twitter, MUCH better than blacklisting domains. It's dead obvious that they do not want it. The reasons why is beyond me.
- tehlike 4y ago
  "it's not my job". Someone whose job it is made a few mistakes. you must have made a shit ton of mistake too. Claiming otherwise is stupid. What you and i can pinpoint on twitter spam is independent of "everything is evil". Difference.
  - can16358p 4y ago
    Of course everyone does mistakes, but that's not the point. Twitter is just too powerful with all those engineers and access to ML to have this just as a mistake. Whoever is responsible for that flagging is either: - Incredibly incompetent and not knowing what they were doing (which I honestly don't want to believe) - Knew exactly what they were doing and it's intentional with a benefit of doubt with plausible deniability. Unintentionally blocking a whole domain "by mistake" falls into the first point above. Knowing track record of social media companies hiring good engineers and being evil, also knowing Mastodon being a competitor to Twitter, I don't think I need to tell much more.

registeredcorn 4y ago

I'm not saying your wrong, but that just sounds like an incredible amount of work for something that would presumably be discovered anyway, you know? Wouldn't it be much easier to simply blacklist and put out some bland statement about "not supporting 'hate speech'" or something? It doesn't matter if it's true or not, it just matters that they paint the picture of them opposing something "bad". Lying just seems a lot easier than going to all of that effort.

can16358p 4y ago
Maybe I didn't clarify much. It doesn't require more work than they are already presumably implementing for spam detection (unless I way too much overestimate big tech's abilities), it's a little touch to an already-being-implemented system with a bonus of being able to block competitors to some extent, without direct liability.