Anti-Spam Questions

[trooper6]

It's still time spent trying to figure it out. If you hire a human to translate a captcha, then it works because it only takes like 10 seconds each. If you have to google search something, then it starts taking 2-3 minutes.

If the question is difficult enough that it will take 2-3 minutes of google searching to figure out the answer, that is going to shut out a lot of noobies who aren't spammers. There are members who are 12 years old, who don't have English as a first language, or who are sensitive to places not seeming new-person friendly. Questions that would stump a spammer for 2-3 minutes could easily drive them away. And if the answer is googleable, and the spammer is getting paid, they'll find the answer if LSF truly is valuable enough of a target to hire actual humans to defeat captchas in the first place.

[fleet]

I don't use renpy, but I make visual novels, so any questions that are specific to ren'py might not be a god spammer deterrent.

[PyTom]

If you looked at the question I picked above, I'd say it's someone who has played a visual novel in any engine could answer.

[trooper6]

If you looked at the question I picked above, I'd say it's someone who has played a visual novel in any engine could answer.

That is true, but wouldn't a person being paid by spammers also be able to answer that question?

[PyTom]

My theory is know. Visual novels still aren't particularly mainstream, so it's unlikely that the person could answer it without searching - and that searching is likely to be uneconomical. The goal here is to increase the cost to the spammers to make us a less-tempting target.

[Taleweaver]

To advance the text in a visual novel, _____ the left mouse button. (5 letters.) There are 3 valid answers, beginning with c, p, and t.

Too easy. If you pay someone to solve captchas, said person will also be able to answer that question. I mean, there are only that many things you can do with a mouse button.

Maybe something like "Visual novels are the Japanese equivalent of ______" (adventures / adventure games)

Or "Visual novels is a genre of games originating from ________" (Japan)

[trooper6]

Maybe something like "Visual novels are the Japanese equivalent of ______" (adventures / adventure games)

Or "Visual novels is a genre of games originating from ________" (Japan)

When I first game here, I would not have been able to answer those questions. I found Renpy through Digital: A love Story. And being from the art game/EVN side would make those two questions too hard.

[Taleweaver]

Maybe something like >The abbreviation "VN" stands for...?<

[Kailoto]

Hmmm... some ideas that aren't obvious but aren't esoteric either:

What does the Py in Ren'Py stand for? (Python)
What script statement in Ren'Py creates a choice? (menu)
Who is the creator and developer of Ren'Py? (PyTom/Tom/etc.)

Might come up with more later. I also liked these questions that others have posted (slightly rephrased), as they land in that in-between area as well:

What year was Ren'Py initially released? (2004) (Also my favorite one so far - not common knowledge but very easy to look up.)
What tutorial game comes with every download of Ren'Py? (The Question)
What key skips text in Ren'Py? (Control/ctrl/etc.)

My rule of thumb for these types of captchas is that they can't be answered by common sense (i.e. "How many letters are in Ren'Py), but have to be found with a quick search. Questions that only have few permutations of answers are also preferable.

[fleet]

In my experience, captchas don't work. We get enough traffic that it's worth it to just pay a person to sit there solving captchas. I did come up with a new question, which I _think_ everyone should get:

To advance the text in a visual novel, _____ the left mouse button. (5 letters.) There are 3 valid answers, beginning with c, p, and t.

Of all of the questions posted, this is the only one I could answer when I first started coming to lemmasoft, since it works in VNs I had previously played.

[Rinima]

At risk of over-simplifying it, couldn't we ask "What does VN stand for?"

[Tempus]

To advance the text in a visual novel, _____ the left mouse button. (5 letters.) There are 3 valid answers, beginning with c, p, and t.

Of all of the questions posted, this is the only one I could answer when I first started coming to lemmasoft, since it works in VNs I had previously played.

I agree that this is probably the only appropriate question so far. If we agree that a.) visual novels are a niche and b.) we would like more people to create and experience VNs, we don't want the first thing someone has to do upon trying to join the community be an arbitrary research project. Nothing is obvious, but PyTom's question seems the most inviting. That being said...

My theory is know. Visual novels still aren't particularly mainstream, so it's unlikely that the person could answer it without searching - and that searching is likely to be uneconomical. The goal here is to increase the cost to the spammers to make us a less-tempting target.

VNs aren't really mainstream, true, but that's sort of irrelevant since it's easy enough to answer the question you proposed even if the part before the comma were removed like so:

• "_____ the left mouse button. (5 letters.)"

Those three VN-specific operations have 100% overlap with generic mouse usage. In other words, you don't need to know anything about VNs to solve it—it can be easily intuited.

[Kailoto]

My theory is know. Visual novels still aren't particularly mainstream, so it's unlikely that the person could answer it without searching - and that searching is likely to be uneconomical. The goal here is to increase the cost to the spammers to make us a less-tempting target.

VNs aren't really mainstream, true, but that's sort of irrelevant since it's easy enough to answer the question you proposed even if the part before the comma were removed like so:

• "_____ the left mouse button. (5 letters.)"

Those three VN-specific operations have 100% overlap with generic mouse usage. In other words, you don't need to know anything about VNs to solve it—it can be easily intuited.

Which is why we need to stop thinking about "questions that everyone could answer" and shift to "questions that people can find the answers to easily." There's nothing wrong with a question that people don't automatically know the answer to. I can think of two other forums that have done this, and in both cases, they asked questions that I had no way of knowing. The thing was, though, they were both questions along the lines of "what year was X released," the answer to which could be found through a ten second Google search.

I mean, if you're looking at anti-spam questions because you're worried that someone could simply be paid to solve captchas, wouldn't that hypothetical person also be able to easily solve a question like the one written above? They wouldn't even have to look it up - as Tempus stated, it's easy to intuit it.

[Hazel-Bun]

Which is why we need to stop thinking about "questions that everyone could answer" and shift to "questions that people can find the answers to easily."

I agree. Something that can be searched up in less than five seconds isn't exactly that hard, but can help. Plus, less spammers and less work on the moderators and users. Also, things related to VNs can be good things just to know coming into the community despite what source drew you in.

[Tempus]

"_____ the left mouse button. (5 letters.)"

Those three VN-specific operations have 100% overlap with generic mouse usage. In other words, you don't need to know anything about VNs to solve it—it can be easily intuited.

Which is why we need to stop thinking about "questions that everyone could answer" and shift to "questions that people can find the answers to easily." There's nothing wrong with a question that people don't automatically know the answer to. I can think of two other forums that have done this, and in both cases, they asked questions that I had no way of knowing. The thing was, though, they were both questions along the lines of "what year was X released," the answer to which could be found through a ten second Google search.

I mean, if you're looking at anti-spam questions because you're worried that someone could simply be paid to solve captchas, wouldn't that hypothetical person also be able to easily solve a question like the one written above? They wouldn't even have to look it up - as Tempus stated, it's easy to intuit it.

The person paid to solve captchas is going to be totally thwarted (not just slowed down) by a question that can be easily Googled but the newbie isn't going to be inconvenienced? How are we so certain of this asymmetry? That sounds like total nonsense, honestly.

IMO the question system sucks in isolation. It'd be better to have an easy question that a human can intuit to halt primitive bots and then develop a system that uses a heuristic to identify potential spammers when they register and restrict their posting abilities if they match that pattern. Unlike a question, you can't scrape a server-side heuristic and pay someone to solve a bunch. If it's well-written and based on actual data there'd be a low amount of false-positives.

In my observation, most egregious spam (the stuff that links offsite, offers you various crap, etc) is made as a new topic. Spam elsewhere, such as posts in existing threads or profiles is relatively innocuous. In existing threads it's usually "wow I really like this! I'm totally a genuine human!" — harmless and short. Given those content-less posts few will visit their profiles. So, test the combination of a user's first post that starts a thread against their registered details and if they match the pattern prevent them from starting new threads (or from posting a new thread if that was their first post). Anyone who's restricted gets placed in a simple list for review by mods.

The heuristic might be something like (keep in mind this is an example, I'd need to actually look at commonalities between spambots and differences between real humans to develop something suitable):

• Has a website listed on profile.
• Has a number in username.
• Does not list a deviant art, github, or soundcloud profile.
• Uses certain words or symbols in a post starting a new thread ("$", "£", "buy", "sale", etc.)

Anyway, I didn't suggest this initially because I know phpBB can't do this. LSF would be better off migrating to entirely different software IMO, but that's getting even more off-topic.