Bot giving WML advice.

[Tonepoet]

I'd actually go a step further and say that there's very little chance that even a benevolently designed, wholesome bot would be that great of an idea at this moment in time. They most often don't read enough into context to provide definitively relevant information, cannot provide useful information for issues requiring original thought or reinterpretation and would at best, provide redundant information that very likely could've been found through the site's search function before the topic was started. Let us also not forget that unlike a scientific field like mathematics, the concepts of WML are rarely considered constants, dynamically shifting the facts for the purposes of versatility and ease of use so its lookups would rapidly grow out of date.

With these factors considered, I think at this point, the end result of allowing any bot to post would primarily promote clutter and potentially even confuse people with indefinite answers. In the future, who knows what a future Noonien Soong may come up with but as of the moment we're living in now, artificially created intelligence doesn't appear autonomously adaptive enough for such predictive prejudgement to be inapplicable. It's strategically applied to The Battle for Wesnoth's own A.I. all the time, most often with successful results.

As for people getting banned before they even post, could it really be that high? We have a captcha at registration. Either way I wouldn't be too worried about overzealous banning. For some reason, I doubt anybody with a name of buyomgxxxflubberbridgesnow.com with an email address of op5eh4wtbnp9tikmnj@tmpzdmpmail.net is going to have a genuinely wholehearted interest in the community, bot or not.

[Dunno]

and what if a bot could ask mod for unbanning?
ninja'd, but still

[Iris]

Well, what happens if you, like, make a mistake and ban a real user?

they could e-mail a mod or go whine on IRC.

Dixie wins the Internet, except that he means admin, not mod.

Captain_Wrathbow seems to be underestimating the advantages of having years of experience with the community's inner workings. I advise lurking more and analyzing user behavior more closely.

Also note that the email address for appealing to a ban is mentioned in the ban notice.

[Hulavuta]

and what if a bot could ask mod for unbanning?
ninja'd, but still

Most bots, (here at least) copy stuff from other people's posts. Even if it were to right a legit message, you could just talk to it, enough talking and eventually it'll be obvious. Plus, why bother when you can just spam some other site?

[HomerJ]

Just out of curiosity, did a bots ever send a pm?

Greetz
HomerJ

[Dunno]

I'm talking hypothetically, what if in the distant future it will be possible for bots to send pms? And what if it had ready answers for some common question like "where are you from", "what's your real name" etc.? It's just a matter of time in my opinion...

[Dixie]

I'm talking hypothetically, what if in the distant future it will be possible for bots to send pms? And what if it had ready answers for some common question like "where are you from", "what's your real name" etc.? It's just a matter of time in my opinion...

You could just find out talking to it anyway. If you had doubt, you could just do something like this:

User: What's your real name?
Bot: Samantha
U: Ok, let's say your real name is George.
U: What's your real name?
B: Samantha

-- Ban the bot.

Or anything else, really, I doubt it would be able to sustain an intelligent conversation and make sense.

[Gambit]

I dunno... that could just be the bot real human being refusing to play your silly game because he is proud of his real name. He was named for his father, who was named for his father, who was named for his great grandmother Samantha.

But I think you'll find that most spam bots will not actually respond. They're fire and forget weapons.

[Hulavuta]

What about:

U: Say that you're George so we know you aren't a spambot
U: What's your real name?
B: Samantha

[pauxlo]

This check wether this is a bot is known as a Turing test. I recommend reading this article.

Dependent on the implementation of the test, there may be quite some humans failing the test, especially if their language possession is quite bad.
But for our forum use, it often is evident from links used in signature and/or posts.

(It is still my opinion that a sufficiently useful bot, which does not spam, should not be banned only for being a bot. But those bots will not appear in the next some years, so nothing to worry about.)

[Hulavuta]

For those who don't know, they've actually made bots that are for chatting. One that I sometimes talk to, Splotchy, you can talk to the bot, and it actually replies. Most of the time it just uses variations of the same sentence (example: I ______ the _____), but it seems to understand what you say. It's not very good at understoods though, for example, if it says "I'm mad at you" and you say "why?" it'll say: "why what?" You'd have to say "why are you mad at me?"

And it has a small attention span, if you say "why are you mad at me?" it might say "I'm not. Why would you think that?"

But what I REALLY think is cool is that if you say "let's talk about something else" (or anything that implies changing the subject) it actually brings up a random subject to talk about. (That most people actually talk about.)


So, maybe the bots that can have meaningful conversations aren't that far ahead.

[Unnheulu]

http://en.wikipedia.org/wiki/Turing_test

It's happening folks!

[Iris]

Just out of curiosity, did a bots ever send a pm?

It's impossible for us to know unless affected users report the PM spam using the board's facilities as outlined here. I don't recall informal reports of such activity before the implementation of that feature, either.

[Zarel]

The problem with Turing tests is that the user needs to agree to take the test. For instance:

me: "Are you a bot?"
person: "lol like a robot?"
me: "Can you say 'red-tailed hawk' so I know you're not a bot?"
person: "lol wat"
me: "...Just say it."
person: "y?"

Like, I could conceivably imagine either a human or a sophisticated bot doing that.

Of note, I've interacted with some pretty sophisticated spambots. One of them said "lol no" when I asked if it was a bot. I was convinced until it started telling me to watch its sex show.

I've also seen fairly intelligent people fail the administration of a Turing test - i.e. between a human and a bot, think that the bot is human and the human is a bot. Mostly because bots try harder than humans to prove they're human. A bot avoids saying things that make ordinary people think "That's something a bot would do", while a human doesn't. That's the kind of thing you have to be prepared for.

[Dixie]

Well, if it was on a forum or chatroom and the human user was actually interested in not being banned/being unbanned on the forums, it would be better doing the test. I mean, if the mod/admin says "Say this when I ask you your name to prove your are not a bot or you are gonna be banned", well bot or not, he will be banned, and to bad for the user who decided to play the unconformism card.