Re: Regarding the Use of Third-Party Programs and Player Safety

[Lemage]

Developers exposed something to the client that could be abused/exploited with any middleman tool, the third party program itself is an issue but devs need to learn to take accountability and fix their sh*t, and people need to stop absolving them of the blame for making archaic choices and poor decisions with their developments in the first place. They're not an indie dev.

[Exmo]

But since the blacklist functions also prevent characters from being displayed (to stop in game 'following', poses, emotes, etc) there would need to be more information sent (or omitted) from the servers responses, e.g. the server wouldn't send 'char abc is standing at x.y' type data for blocked chars. Someone could probably view the network traffic and compare that against network traffic when the blacklist is not enabled and use the difference to determine alts. More time consuming and possibly not as 100% as the existing stuff, but still possible. We also don't know the performance impact of all traffic having to be ran through that kind of filter. I'd like to say it wouldn't be much but this game and network perf ... Well, who knows

100% this. You're still communicating character ABC is the alt of someone on your blacklist, either through a isBlocked flag or through omission of that character. The only difference is you're now performing an additional N^N logical checks every 100ms.

[BigCheez]

100% this. You're still communicating character ABC is the alt of someone on your blacklist, either through a isBlocked flag or through omission of that character. The only difference is you're now performing an additional N^N logical checks every 100ms.

Not this.

You would essentially already need to know that character B was an alt for character A to be able to track them down and confirm that they're also blacklisted.

Again, it isn't O(n^n). I don't think you even know what that means. And again, you only need to run the check once and then cache the result for as long as you're receiving data for the player.

Posting "you're clearly not a developer and that's ok, but then keep your opinions on software architecture to yourself" and then this is wild. Zero self-awareness.

[Exmo]

"Just cache the result" while forgetting you need to check the cache on every request, which is the point of the cache. Checking the cache is a logical check. Making a cache isn't a magical thing that solves all problems. It's still N^N logical checks, but also you need to maintain the cache every time someone changes their blacklist or enters or leaves the instance (or did you want a memory leak that kills the entire server?).

[BigCheez]

"Just cache the result" while forgetting you need to check the cache on every request, which is the point of the cache. Checking the cache is a logical check. Making a cache isn't a magical thing that solves all problems. It's still N^N logical checks, but also you need to maintain the cache every time someone changes their blacklist or enters or leaves the instance (or did you want a memory leak that kills the entire server?).

That depends how they're handling the data and how you cache the results. Yes, they need to check to see if data for a player should be sent but that isn't an expensive operation. Yes, they need to update/clear the cache every once in a while. No, that isn't a big deal either.

[ValynS]

If it's something thats done incredibly frequently, like updating information on other characters on screen, even a tiny change could have a non-trivial impact. It also doesn't really prevent people on figuring out alts, it just shifts the effort of doing it around. A malicious person can stand in limsa, grab some network data, add "Char X" to their blacklist then re-grab the data and they'll see "Char Y" no longer in the response. It'd probably be an improvement over the current one, but the tradeoff of added computation/latency on every single update from the server may not be worth it. I don't think there's a way to have account-wide blacklisting that doesn't inherently have some degree of can-find-alts-if-dedicated-enough problems.

[BigCheez]

It'd probably be an improvement over the current one

The word "probably" doing some real heavy lifting in this sentence.

[AmiableApkallu]

If it's something thats done incredibly frequently, like updating information on other characters on screen, even a tiny change could have a non-trivial impact.

I've worked with folks who were concerned about a program needing another ~750KiB of resident memory because the typical use case was to run tens of thousands of instances of it on a single host, and it was a big ask of the typical admin to "just add some 10s of GiB of RAM."

It also doesn't really prevent people on figuring out alts, it just shifts the effort of doing it around.

On the other hand, shifting effort around, or increasing the amount of effort required... that can be enough. After all, much of modern encryption and hashing isn't based on making it impossible to recover the original text. It's based on making it infeasible for some (large) collection of adversaries within some (large) amount of time and resources.

[Exmo]

On the other hand, shifting effort around, or increasing the amount of effort required... that can be enough. After all, much of modern encryption and hashing isn't based on making it impossible to recover the original text. It's based on making it infeasible for some (large) collection of adversaries within some (large) amount of time and resources.

This is totally true. But I think what ValynS is saying it's that, although replicating a huge database with everyone's alts would certainly be more difficult, it would still be relatively trivial for, for example, a single stalker with a single victim in mind to discover their alts. And it's cases like this that are the heart of the conversation.

[AmiableApkallu]

It would still be relatively trivial for, for example, a single stalker with a single victim in mind to discover their alts. And it's cases like this that are the heart of the conversation.

I mean, yes, if you give the someone the ability to "blacklist" not just a single character but every character tied to an account, then they will be able to discover someone else's alts with enough effort. After all, the so-called "victim" here blacklisting "their stalker" can learn all of the stalker's alts. Swap around the words "stalker" and "victim," and nothing changes.

So the question you have to ask is: Do you want the ability to block every character on an account in one easy go, or would you rather prefer to do it one character at a time?