Open rlemon opened 10 years ago
I don't think the bot tweeting things using a command is an important feature. I wouldn't care for it, personally. However, it could be a feed for starred messages. Thoughts? On Aug 22, 2014 6:33 PM, "Robert Lemon" notifications@github.com wrote:
So the bot has a twitter account and it so far has done nothing. I'm thinking of ideas on how we can integrate twitter (tweeting) as a command.
Maybe an owner only command? then it doesn't get overly abused. !!tweet O Hai! Look at this tweet!
and the bot would post to twitter, then reply with the link to the tweet.
looking for alternate ideas and discussion. What does everyone think?
— Reply to this email directly or view it on GitHub https://github.com/Zirak/SO-ChatBot/issues/211.
Perhaps tweet starred comments, along with a link back to them.
the bot can also tweet the "pinned" messages automatically
We all just wanted the same, I guess that should be implemented
So there's a chatbot for a particular chat.
And it has a twitter account.
Followed by people from the chat.
Where it posts messages from the chat.
What?
So we can retweet certain starred messages instead of having to type them out ourselves :P On Aug 22, 2014 6:40 PM, "Zirak" notifications@github.com wrote:
So there's a chatbot for a particular chat.
And it has a twitter account.
Followed by people from the chat.
Where it posts messages from the chat.
What?
— Reply to this email directly or view it on GitHub https://github.com/Zirak/SO-ChatBot/issues/211#issuecomment-53058080.
Also get updates on the happenings of the chat while not actually there
This just in:
Haha, I wouldn't mind, but I do see the point. Either way, I'm ambivalent to the idea On Aug 22, 2014 6:43 PM, "Zirak" notifications@github.com wrote:
This just in:
devdocs.io
— Reply to this email directly or view it on GitHub https://github.com/Zirak/SO-ChatBot/issues/211#issuecomment-53058366.
Maybe have a command where you can see who's in chat via twitter. It'd be a direct tweet
I like the feed for starred messages idea. What room though :P?
If this were a feature, it should only tweet pinned messages, IMO.
Tweeting starred messages could be easily abused and lead to spammy tweets.
An owner only !tweet :messageID
could be pretty nice.
Starred messages are too bound to abuse, pinned messages don't always reflect exemplary messages from the room.
Yeah, pinned messages are usually announcements - not something to tweet.
I occasionally login to chat just to check if anyone's posted interesting links that have been starred, so I like the idea of tweeting starred messages, but yeah, it's hard to do that in a way that doesn't seem abusable. There's a lot of starred messages that I don't think we really want to record for posterity. (Maybe if they get a certain number of stars? Or is it possible to tell who starred a particular message and limit it to messages starred by particular users?)
Maybe it could be tied to a voting system to explicitly vote to tweet certain starred messages? (And people who abuse that voting system could have their voting permission revoked?)
If we only tweet pinned messages and 3-times starred messages, I don't see what kind of abuse there could be. Really.
Think of the out of context LOLs that Cap can share on twitter. The world must know the tale of the two slides..
Like, for example, we would have just tweeted "I think defecating on park equipment is awesome". Is that really the sort of thing we want to preserve for all of the internet?
Retsam, is that even a question?!
It could also tweet once a day basic stats, like how many messages sent, max amount of users chatting at the same time, longest inactive periods... Would just seem like something easy and cool to do, even tho it wouldn't have an actual purpose other than... coolness? But do half of her commands do anything but only-for-coolness stuff? Also, she could have 2 twitter accounts, one accessible by everyone 'trusted' on chat (low level of trusting, for example even I could use it), maybe just once par hour par user and can be abused/overused, while the other just for room owners to display more 'almost-serious' stuff. Also, I agree on the stars thing
I like the stats idea. They're usually interesting to see
On Fri, Aug 29, 2014 at 11:31 PM, Copot Matei notifications@github.com wrote:
It could also tweet once a day basic stats, like how many messages sent, max amount of users chatting at the same time, longest inactive periods... Would just seem like something easy and cool to do, even tho it wouldn't have an actual purpose other than... coolness? But do half of her commands do anything but only-for-coolness stuff? Also, she could have 2 twitter accounts, one accessible by everyone 'trusted' on chat (low level of trusting, for example even I could use it), maybe just once par hour par user and can be abused/overused, while the other just for room owners to display more 'almost-serious' stuff. Also, I agree on the stars thing
— Reply to this email directly or view it on GitHub https://github.com/Zirak/SO-ChatBot/issues/211#issuecomment-53909863.
I would love to have the bot tweet starred messages on a dedicated twitter account.
I don't see too much trouble on the abuse part, because the worst that can happen is that stupid shit gets posted on Twitter, which is exactly what most people post on twitter anyways.
Also re to the fact we don't want stuff to be preserved for the entire internet: [tag:pro-tip] chat already gets indexed by search engines. If you don't want something public don't put it on the web...
This is more a gimmick than anything else, but at times there are some gems getting starred in certain rooms (amongst the crap) and a twitter feed would be nice to easily aggregate stuff what goes on in different rooms.
My €0.02
All the points that I thought of on this have already been covered, but to briefly summarise:
Sounds good, messages should be tweeted on these occasions:
x
(3? or 5?) stars!!tweet <message>
or !!tweet <:messageId>
I would not tweet pinned messages as many pinned messages can tend to be informational ones posted multiple times. I also think there should be a delay in when the bot attempts to tweet the post. The post should be active for x-amount of time before it's considered. This will prevent it from tweeting a post that was quickly starred, but then pinned and/or unstarred for moderation concerns. I personally think option 3 should be the only one, maaaaybe 2 provided there are time limits and the ability for room owner to delete tweet if necessary with a command.
A digest every few hours, also use a db with re-used urls and don't repost any message containing the url in db. Post TOP X messages, maybe a little algo on comment length and starCount
? things.
Sounds complicated and needless On Sun, Oct 26, 2014 at 22:48 ᵠ notifications@github.com wrote:
A digest every few hours, also use a db with re-used urls and don't repost any message containing the url in db. Post TOP X messages, maybe a little algo on comment length and starCount
? things.
— Reply to this email directly or view it on GitHub https://github.com/Zirak/SO-ChatBot/issues/211#issuecomment-60531633.
After thinking for a bit maybe you're right (@KyraD) about pinned messages, we should keep them separate from tweeting because if someone wanted to tweet a message they pinned, they can just do that with the command - assuming that a command is implemented - but even if it isn't they still shouldn't be explicitly tied into each other.
Also +1 @CS- RE: A digest of the past "few" (3 - 6?) hours most starred posts or something. Also strongly agree that we should make sure things are not tweeted more than once (if we have another automatic tweet method such as pinned messages or even manual commands), Most likely using a db.
Re: Anyone who says this is complicated/needless/unnecessary etc @benjamingr :
So? :laughing: As @towc said, half of her commands are unnecessary anyway, it's just because we can!
How about stopping the talk and one starts writing the code? Le 26 oct. 2014 22:10, "Billy Mathews" notifications@github.com a écrit :
Re: Anyone who says this is complicated/needless/unnecessary etc @benjamingr https://github.com/benjamingr :
So? [image: :laughing:] As @towc https://github.com/towc said, half of her commands are unnecessary anyway, it's just because we can!
— Reply to this email directly or view it on GitHub https://github.com/Zirak/SO-ChatBot/issues/211#issuecomment-60532440.
Oh, @Ralt, I genuinely wish I had the time for that
@Billy- your time isn't more valuable than my time, or @Ralt's, or any other maintainer's time. I could've been swayed to say tweet every message with 3+ stars, but as @benjamingr pointed out, this does seem more complex than necessary. Having said that, if you really want it, then make it and PR.
I see a lot of potential for abuse here. How about just using the account for tweeting bot status updates so that the regs don't have to ask every time Cap goes down?
What about we get something running and tweak it according to experience? Le 26 oct. 2014 23:05, "Josh Moncrieff" notifications@github.com a écrit :
I see a lot of potential for abuse here. How about just using the account for tweeting bot status updates so that the regs don't have to ask every time Cap goes down?
— Reply to this email directly or view it on GitHub https://github.com/Zirak/SO-ChatBot/issues/211#issuecomment-60534353.
@Zirak I never said my time was more valuable than anyone else's, and I'm not asking anyone to put in time either.. If someone wants to do it and has the time then they can. I'm just giving my ideas and opinions as asked for..
As for that being too complex I have little idea how she works although I'd imagine it couldn't be too difficult to determine how many stars a message has? There must be an event when a message gets starred so you could just do (messages[id]+1 > 3)
? Again I am pretty clueless just speculating.
I think @monners point is good as well - and to branch slightly off-topic from this idea - could get her to pin a message in the room when she goes down and even unpin it when she's back up?
@Billy- I suggest you become familiar with the internals of the bot before suggesting that things are "easy".
I'm closing this issue - there's a good consensus that we don't want the bot to tweet anything.
@SomeKittens I never said it was easy. I said that I imagine it couldn't be too difficult. Check yo' facts before you misquote me :-)
And a few people are concerned about the feature being abused and others think it may be too complicated but I wouldn't say there was a consensus that it's not wanted? At least it should be discussed further I think until there is a consensus that we shouldn't go ahead with this feature.
Collaborator powers! woo! comment! woo! < moving on >
@SomeKittens and I agreed that we kinda deviated from the point.
So far, most people agree that they want the bot to tweet...something.
Opposition to each point:
I'm a bit biased towards this being a bit pointless, so convince me otherwise.
I'd like to oppose @Zirak's Opposition point 1: likely maybe, but who knows what the cat drags in... if the feed ends up being useful it might be followed by a lot of programmers/IT people and even get traffic to the JS room (best case scenario maybe). and 2: how hard would it be, it's about 140chars or less :)
Indeed all of this is a glorified filter over star hist. and dumped on Twitter, but having this data we can analyze it: are most of the 3+ stars messages interesting/valuable? is this a good enough filter?
Abuse needs to be taken into consideration.
If it would also ignore one-box-ables (img especially) this could start making some sense, useful things to share would be those created in the room (eg. insight, advice) instead of shared: http://i.imgur.com/eEMa2Dn.gif
While we were discussing this in here it looks like at least some people don't mind about starred messages and context:
http://stackapps.com/questions/3826/shit-se-says-stack-exchange-chat-stars-out-of-context
So the bot has a twitter account and it so far has done nothing. I'm thinking of ideas on how we can integrate twitter (tweeting) as a command.
Maybe an owner only command? then it doesn't get overly abused. !!tweet O Hai! Look at this tweet!
and the bot would post to twitter, then reply with the link to the tweet.
looking for alternate ideas and discussion. What does everyone think?