Open oksushi opened 9 years ago
People used to get crazy about channel logging, I wonder if they still will now that it's kind of "out there" as a thing?
There was one bot that just joined every single large freenode channel and tried to be the "Google of IRC". That was eventually klined and had to negotiate with the ircops in order to continue operation. But that was years ago now...
My plan has been to do anonymized metrics along with logs for anyone who can demonstrate that they were in the channel at the time (bot tracks join/parts, bot exchanges login token with website for security).
I know #html5 and #web have public logs. Most project channels on Freenode do too (#whatwg, #node.js, #jquery, etc)
I'm not for or against it, but I don't think people are as crazy as they were about it
Quite a few channels do. As far as I know, the only policy from freenode is that if you log, you must say so publicly. It's usually in the topic with a simple "This channel is logged" or similar.
Honestly even if some people are "crazy"/paranoid about being logged, then joining a well-traveled IRC channel is not the best way to stay private. There are plenty of people who have their own personal logging. So if it is stated in the topic they can just leave before saying anything if they don't like it.
Anyway, that all said I haven't typically found logging to be useful to me in a channel. I might occasionally check it but not for any real reason. The only benefit would maybe be to mods/ops when not online with their own logs to see if there's a bad past history with a person if they're currently exhibiting bad behavior.
So I'm not really for or against this.
@preaction yeah, I think that company eventually died out if we're thinking of the same thing, since their major business plan was kinda nuked. Their problem was more that they didn't ask channels if they wanted them to be there. Iirc they were trying for more than freenode, to make a "graph" of sorts of IRC. A big undertaking even if you ignore getting permission, etc.
@AMcBain I remember talking to someone about that bot after I banned it. The line they were giving me was about all the knowledge that was being lost or locked up in useless personal vaults. And, like you, I've never looked at public logs for anything useful, and only have ever seen them used to win an argument about what said in the past. So, they didn't very much convince me (context is everything, and StackOverflow is starting to see why every question isn't necessarily useful for people other than the petitioner). I could swear I remember hearing about their monetization strategy, and being more determined to ensure they were nowhere near the channel.
But, putting together a simple channel log would be the first step in mining it for metrics about user behavior, including specific users. Logstash + ElasticSearch would make for a quick, useful tool (rofl, logstash can even do this out of the box). The benefit being the ES database, rather than being locked up in someone else's database.
3 minutes of using botbot.me, and I can do: "nick:name" to see all the chat that user made. While doing just "name" will give me chat directed at that user. I can't find any kind of advanced API for any kind of crazy features like "everything a user said and everything anybody said in the channel 10 minutes before and after", which sounds like the kind of query I'd want to do for disciplinary purposes.
Additionally, being able to objectively find "Frequently Asked Questions" would point to improvements that could be made in our infobot, in our procedures, in our own knowledge, in our website, and in global CSS documentation. In short, we could mine user data to improve CSS itself (if you want to get grandiose with a slight hint of megalomania ;) but I am surprised that the spec authors aren't in the trenches dealing with people attempting to use their Frankensteinian creations)
So: What are the use-cases and how can we best satisfy them? Then, what's the minimal amount we need to do to start on that path?
I think it wouldn't be too challenging to set up logstash + ES on a server somewhere. We could make a logs.hashcss.com repo with a UI for custom queries.
No, it wouldn't be, but why?
Some use-cases:
But none of those things require public logs.
I'm tempted to set up a SurveyMonkey or something for a democratic process around public vs. private logs. We're already being privately logged by everyone in the channel. But to my knowledge nobody has posted those on a website anywhere.
I see it mostly as an easy way to get everyone looking at the same thing. Examples:
I realize that all of these could be done with private logs, and it would work fine, but public logs may make some things a little easier.
Also, side note, freenode doesn't have half-ops, only ops and voices. And I wouldn't be a fan of people having voices for the same reasons the ops don't wear their +o all the time, it draws unwanted attention to them and creates a visible hierarchy in the channel.
1 can be done with private logs, 2 and 3 can't I don't think. And 2 is, I think, enough of an issue to warrant it. For 3, if it's frequent enough, there are always adding facts to the bot. For 4, well, that's one of the fun bits of anonymity online.
And, sure, if someone wants to make the magic private-ish log thing I mentioned... But I don't think people are that concerned about logging anymore, not since the issue was cracked open a few years back. It truly is a new era of privacy on the Internet...
I agree, I don't plan to give out +v, they're pointless unless the channel is +m or +R (and I also don't plan to do either of those). I'm not looking to create a hierarchy, I need a way to give +o on a probationary basis, lest I screw it up again. If there were a +h, it'd be obvious which people were probationary and which were not, so the +o could monitor them effectively. I'd be less hesitant to give out +h than I am +o.
Though the #nodejs channel has a policy of "if you complain about harassment, you're given a +o bit", which I'm on the fence about... If I gave out +o to some of the people who've asked...
For 3, if it's frequent enough, there are always adding facts to the bot.
Or adding the link of the conversation to the bot, if we find that it's helpful.
I need a way to give +o on a probationary basis
There is the +f flag that allows modifying the channel access list. You could have trusted ops with +f as well, and probationary ops with just +o, so a trusted op can remove them from the access list if they deem necessary.
Though the #nodejs channel has a policy of "if you complain about harassment, you're given a +o bit", which I'm on the fence about... If I gave out +o to some of the people who've asked...
That's a larger, more active channel with a different culture than here. If there's always one op at least casually monitoring #css and #css-ops, it should be enough.
https://botbot.me/ is a nice option worth considering