Help Me Design a Modbot to Help You Run Your Games

[Psyche]

I want to code up a bot or set of bots that automates the more trivial activities a game moderator does.
However, there are a lots of ways I can do this, and a lot of things I can make it do.
For this reason, I want your help making decisions about the way I'll actually do this.
So that when it's all done you'll actually want to use it.

I want to know...

1.

What would you want a 'ModBot' to do for you?

Can you sort these in order of priorities? How do existing resources for these activities currently fail you? For example, maybe you really want a Bot to do votecounting for you, or maybe you'd appreciate having an easier time assigning and sending out role pms. There's not much that game moderators do that a modbot can't do (besides, like, creating Scummie-winning flavor or setups or stuff like that - though generic flavor and setups are of course possible).

2.

How do you want to

interact

with 'ModBot'?

For example, do you want it to be an app in your desktop? Maybe a Google Form you fill out and edit? Something that controls your account to complete accounts, or a bot account you "co-mod" games with? A unitary resource, or a set of tools? There are many more possibilities besides these.

3.

How should Modbot's particular functions be designed?

For example, there are good ways to post vote counts and bad ones. We should work out all these issues before or as I'm coding them.

In my next post, I'll list some of my own ideas for how we could achieve this, and begin a catalogue of your ideas. Supposing there's enough interest, I'll get to work and hopefully have *key* ModBot functions ready in the next few weeks, if not sooner.

[Psyche]

Some Things ModBot Could Do (in order of when they'd be useful in a game):
- Generate basic setups (given a set of roles, balance constraints)
- Generate basic flavor
- Generate Game OPs
- Assign and Send Role PMs (given a list of players)
- Count/Collect "/confirm" messages, potentially act on them
- Votecounts (better than current resources)
- Coordinate Day/Night transitions
- Compute Night Action Resolutions
- ???

Ways ModBot could 'work' (sometimes many of these ideas could be part of the same implementation):
- A program on a server (or my computer) that people interact with through a web app, google form, private thread, private message or other resource
- A program on moderators' computers that controls their account or my bot account
- an artificial "co-mod" that does selected grunt work for you
- ???

Key Concerns With Other Resources That Modbot Can Improve On:
- current votecounters require exact user naming to work or are tedious in other ways
- activities like vote counting and sending role pms is tedious
- ???

I imagine ModBot working best in Normal, Open, and simple Theme games. But its simpler capacities (eg role pms, votecounting) should have broad generality and ideally be useful for anyone.

[wgeurts]

A good example of one I've seen is keep track of all votes and have the ability to post votecounts whenever players wish to have them. The ability to compile a list of all votes for vote analysis also helps, with additional information such as the ammount on the wagon they joined/left.

Being able to close a thread and post flavour/role PM at deadline/majority would also be good.

Something that automatically follows a specified thread.

[Ircher]

Part of the fun in modding is the fact that there isn't a thing to do everything for you.

We have an automatic votecounter but it's barely used probably for two reasons: 1) Not too many people are aware of its existence 2) There is a lack of interest in such a tool.

[wgeurts]

Part of the fun in modding is the fact that there isn't a thing to do everything for you.

We have an automatic votecounter but it's barely used probably for two reasons: 1) Not too many people are aware of its existence 2) There is a lack of interest in such a tool.

I can assure you that the ability to have a tool automatically post votecounts when players want them will be something very much appreciated. Something that also gathers and provides more data than the average mod does.

[Ircher]

That too -- we already have a way to automatically lock threads at deadline.

Pedit: I'm not saying it won't be appreciated. I'm saying there is a lack of interest in such tools otherwise things like Thesp's Automatic Votecounter would be more prevalent at the moment. Also -- how many mods use the autolock feature at the moment?

[zoraster]

automatic vote counts that post a new post from the bot every page top or X hours from the last vote count.

[Ircher]

That would be useful -- having the bot reserve posts for you.

[wgeurts]

That too -- we already have a way to automatically lock threads at deadline.

Pedit: I'm not saying it won't be appreciated. I'm saying there is a lack of interest in such tools otherwise things like Thesp's Automatic Votecounter would be more prevalent at the moment. Also -- how many mods use the autolock feature at the moment?

Automatic lock at deadline is possible, but a bit could potentially allow a mod to enter flavour ahead of time so that a night can begin instantly. A bit wouldn't have to replace a mod, it would increase quality of life for both players and the mod whilst being an additional tool.

[wgeurts]

The problem with thesps thing is that it was clunky and players didn't work along well.

[Psyche]

Part of the fun in modding is the fact that there isn't a thing to do everything for you.

We have an automatic votecounter but it's barely used probably for two reasons: 1) Not too many people are aware of its existence 2) There is a lack of interest in such a tool.

eh if people are not interested, they can just avoid the thread and i'll get the point

my suspicion is that the votecounter isn't used because it's not designed in a way that mods favor over doing votecounts themselves. I think it's possible to overcome these challenges with a communitarian design process

Something that automatically follows a specified thread.

can you go into detail on this?

[Ircher]

If there's one thing that would be really useful, it would be getting a bot to update the opening post for you. Manually editing it is currently a pain.

Pedit: True, it would be optional, but it's still hard to say how much support there will be for a bot. Also, some of this stuff such as VCs (see mastina for a great example) help define the mod.

[Psyche]

I've been wondering for a while what the best strategy for designing a PageTopBot might be. I could always have it constantly check the thread (or a subforum), but implemented at a large scale I worry that could put an unnecessary load on servers.

I could instead do some research first, and try to measure the average amount of time (and also the standard deviation) it has taken before for page 1 to become page 2, and page 2 to become page 3, all the way up to page 120 becoming page 121 or something. I could then use these statistics to time threadchecks by the PageTopBot.

But that might be an unnecessary amount of work.

If PageTopBot were a unitary process, pagetopping threads across the forum, then maybe just having it check all the threads at once through bookmarking every X minutes would be acceptable performance?

[Backhand]

automatic vote counts that post a new post from the bot every page top or X hours from the last vote count.

Def this. Could do nothing but this and would be great.

[Psyche]

ok then i'll just do the votecounter for now and branch from there

now what i would like help with is:
- proper votecount formatting (with help also designing "Hey the target of this vote is ambiguous" indicators)
- deciding how often a bot should check a thread for pagetopping (this has to do with a - tolerance for repeated pageloads, and b - tolerance for missed pagetops)

[zoraster]

Perhaps have a mod enter the players into the vote counter for the thread and have them able to list several aliases. But overall if the tool is useful enough, mods can force their players to use it. Whether it's using the [/vote] tag always instead of [/b] bold ones or using a player's full name, etc.

[Psyche]

I actually have a lot of strategies for handling ambiguities that should make the votecounter superior to previous renditions.

For example, partially typed usernames like "zor" can be easily matched to zoraster if no one else's user starts with zor (in these cases, the vote shouldn't be assigned in the first place). Similarly, acronyms like AP can be easily matched to AngryPidgeon (though perhaps not "BH" to a username like Backhand) supposing no other username can be similarly acronymed or begins with "ap". Finally, partial mispellings of usernames like zoratter can be easily matched to the right username, supposing someone the misspelling could equally apply to isn't around (e.g. zorotter). This strategy (rooted in computing the Levenshtein distance between strings) can also be used to detect mispellings of partially typed usernames (eg. "zer") or acronyms (though I don't think that's safe).

All this together should make aliases only occasionally necessary (especially if rules are clear and reasonable), as success identifying voter targets should be close to human-level performance. But there'll be ambiguities sometimes anyway (for example, if someone tries to vote Backhand with

vote BH

), and mods/users ought to be alerted to them if a bot is handling votecounts.

[Human Sequencer]

i have ideas this is an ego post for now

[Aristophanes]

i have ideas this is an ego post for now

[Ircher]

Allow it to catch the ends of names as well, esp. for those users like LlamaFluff.

[chamber]

I actually have a lot of strategies for handling ambiguities that should make the votecounter superior to previous renditions.

For example, partially typed usernames like "zor" can be easily matched to zoraster if no one else's user starts with zor (in these cases, the vote shouldn't be assigned in the first place). Similarly, acronyms like AP can be easily matched to AngryPidgeon (though perhaps not "BH" to a username like Backhand) supposing no other username can be similarly acronymed or begins with "ap". Finally, partial mispellings of usernames like zoratter can be easily matched to the right username, supposing someone the misspelling could equally apply to isn't around (e.g. zorotter). This strategy (rooted in computing the Levenshtein distance between strings) can also be used to detect mispellings of partially typed usernames (eg. "zer") or acronyms (though I don't think that's safe).

All this together should make aliases only occasionally necessary (especially if rules are clear and reasonable), as success identifying voter targets should be close to human-level performance. But there'll be ambiguities sometimes anyway (for example, if someone tries to vote Backhand with

vote BH

), and mods/users ought to be alerted to them if a bot is handling votecounts.

Thesps tool already used levenshtein distances, as a note.

[Aristophanes]

Would we be able to select which functions are used for our games? Like, making Votecounts (especially as Sir suggested), generating Role PMs (with input), counting confirmations, and locking a thread at hammer (maybe after an hour of twilight or whatever, as I think twilight is an important part of the game) would all be things I'd like to use it for. But generating flavour, as an example, would not.

Also, I don't use Auto VCs that exist currently because I don't trust them. I have broken them before and I think they are too touchy in how you have to use them.

How would the VC generate a doublevoter's vote? Could that be an option programmed into it?

Also, I like to link the votes to the post in which they were made in my VCs. It is very helpful in making sure they are counted correctly as well as being a nice reference for players as it provides context for the votes and limits mod errors in VCs. This could be a helpful feature as it provides the Mod a way to easily double check the Bot's work. Would this be a possible feature as well?

[Psyche]

hm ok

[Kison]

IMO the coolest new tool to add to a mod's arsenal would be automatically posting vote counts, not just generating them on command. Getting to a point where we trust the tool enough that we let it handle that aspect of moderating would be a great step forward.

[zoraster]

Totally agree with Kison. That one thing I think would decrease the burden of moderating to a degree that might get more mods putting up games. Last I checked (Which was months ago to be fair) we were still at a mod deficit.