Wikipedia:Administrators' noticeboard/Incidents#STBotI just caught my eye. A few recurring debates there (zOMG it's a valid rationale....zOMG read teh policy...), but what really raised my eyebrows was the tone in which ST47 replied to commentary.




If he hadn't developed a few oft-replaced bots, if he was a random non-admin (like, say, me), would we be equally tolerant of this incivility?

No, of course not.

Ah, to be a sixteen year old wikipedia administrator and bot owner. As Duk says; "Is it my imagination or are bot owners more surley than average "

Not really too sure. I don't like the incivility, but it looks like another Beta in the making...

I'm trying to figure out if it's really inaccurate as Duk claims, as the 3 edits he claims, the first doesn't actually have the actual title of the article in question, the second is questionable, and it makes me wonder if it's looking for a wikilink instead of just a string match, and the last actually should be speedied anyway, since it's an orphaned duplicate. It also doesn't give an article name, though it is the same as the image name...

Bleh.

I'd have to say that on average most FairUse bot owners tend to be fairly surley at times. Beta, ST, and Carnildo have had their moments.

Welcome to the world, Duk. And on the freeway you'll notice that the same people drive Ford F-150 pickups and are also rude and inarticulate. There's a certain hormone which causes all this, especially in overdose. 'Bot building, too.

I was gunna suggest we merely have BetaCommand neutered, and see him again in 6 months for behavioral tests. But the world apparently just isn't very rational about these things.

I found this laughable


Duk wants that ST47 not to have any type II errors, but later goes on to block the bot twice for type I errors. I wonder if he realizes that what he's asking for is impossible, unless we're considering something that is 100% accurate, which is pretty much impossible given the diversity of the sample.

I raised a similar issue at Wikipedia:Bots/Requests for approval/John Bot IV and got, well, nothing.

I'd volunteer to take these messages instead of ST47, Betacommand, and the like (I imagine others would do; Charcaroth (sp?) comes to mind), but I doubt this idea would be popular,

Duk can't spell worth a shit (it's surly, not surley) and he doesn't communicate very well, but I think he finally managed to spit it out:

What ST47 refused to state clearly, because he knew that it was wrong but he wanted anyway, was for fair use images to use this particular template. As our conversation continued it became clear that this particular template is not a requirement for fair use, but that ST47 is just too lazy to fix his bot. So then he moved on to criticize everything else he could think of, except his own laziness. --Duk 03:47, 23 May 2008 (UTC)

It wouldn't quite happen unless there's some form of consensus to make the Fair Use policy a bit more user friendly, and by that I don't just mean the guideline, but some sort of committee. I'm not really quite sure how well that would be created, as the Fair Use policy is one of those 800 lb gorillas in the room. Everyone kind of knows about it, but most people really don't like to do anything with it unless it slaps them in the face.

Unfortunately it wouldn't really happen, because most users will funnel to the bot/owner who gave them the message and badger them. Path of least resistance. Giving them some sort of alternative path would be nice, but as I said, I'm not quite sure it could omplemented



Yeah, I saw that too, but I was referring more or less to the Catch-22. (If you set the threshold for a hypothesis too low, you'll invariably get false positives (type I), but if you raise it too high, you'll invariably get false negatives (type II). Essentially, ST would be damned if he let any slip through, but damned if he didn't let any slip through.

regarding the template, I was wondering to the metric that ST was using, and I really don't think that was the correct method to use. What did ST47 promise to check when he got the approval anyway?

Edit: Found it

http://en.wikipedia.org/wiki/Wikipedia:Bot...proval/STBotI_2
http://en.wikipedia.org/wiki/User:ST47/NFCC10c

Quite frankly, I'm appalled that was the rule that got used if there's no check that the images actually use the template before flagging. That's really a horrid design choice, and frankly I'd wonder what the actual false positive rate is.

Bah, you're over thinking this. When a bot screws up too much you block it until it's fixed. Nobody expects perfection.

Calling Duk 'stupid' in ST47's very first sentence to him was definitely not a good move though.

I enjoyed my statistics classes thank you very much

Never said it was a good move. Probably the wrong way to get, but most ImageBot owners take a siege mentality sadly.

Which they defend by pointing out the amount of crap they get. Which is rebutted by pointing out the unsatisfactory image deletion messages. Which are defended... actually, I'm not sure, how do they defend the crap messages? I suppose "it links to NFCC" is considered good enough by some, maybe it's just me?

The completely ridiculous thing is that these bots can easily fix the majority of non-orphan issues on their own, and the powers that be specifically don't want them to. It's absolutely crazy.

Can you elaborate on that? The argument I have heard advanced is that a bot can't write a justification since that requires judgement. Which powers that be are you referring to? BAG?

typically the rationale goes as that it's impossible to write a more custom message for every violation and what they write ends up being sufficient.

The validity of that defense is in the eye of the beholder though.

The vast, vast majoirty of fair use images come from identification photos - book covers, album artwork, corporate logos, etc. No judgement is required on these on the image end - the information is easy to replicate for each image (i.e., the source is always from the distributor of the material, the copyright is always owned by the creating artists, etc). That most of the "problems" come from that area, it could be fixed by a bot in a matter of days, and then there wouldn't be all this drama about it, especially when people know full well that there's no fair use problem with these images and choose to be combatitive about it anyway.

The concept that you'd need some "fair use rationale" for these types of images is ridiculous anyway, but since that's what the project wants to do, you'd think you'd automate the process a bit. I mean, the silly placeholder ones from the dropdown box are more than good enough - tweak those for the future and you're all set - but since they want to make it difficult...

I'll always defend use of an album cover to illustrate the article about the album, or a corporate logo to illustrate the article about the corporation. Until and unless fair use is banned completely (as it was on es:wp) those are absolutely solid uses.

But how's a bot to know that the illustration's use is for that specific article and not for, say, a discography (where it seems album covers are not allowed, I don't agree but that seems to be consensus) or an article about a rival corporation? I'm not trying to be argumentative, if a bot COULD know that it's on the "right" article, I agree it SHOULD. I just don't quite see how, absent tagging that gets you most of the way there already.

I do think that some of the bots are a bit nitpicky about stuff that is "close enough" (tags being there and a human knowing things are right) but when a bot tags an image, I just go fix it... if I can figure out what the issue is.

If the image is being used in a {{infobox album}}, {{infobox book}}, etc., a bot programmed to use WP:FURME would be excellent. I use that, and the judgement required is minimal if anything; furme fills in all the required parameters from the infobox, and a bot could double check it.

As for orphans, it could check history of the article linked to on the image page and see if the image was recently removed, and then at least leave a message to that extent somewhere where people check.

The bots aren't checking for that now. I could upload a copyrighted coca-cola picture and throw it in Pepsi tonight, and no bot would pick up on it.

All the bots are designed to do is check for fair use rationales. Instead of making 1000 messages indicating a poor rationale, it should be making 1000 rationales and allowing the editors to adjust/remove as they see fit. A much better task for fair use patrollers to be looking in articles as opposed to the quixotic quest they're on now.

Only problem I would see with that Jeff, is that by this stage i the game, as I've gathered, is that most images have had some form of a fair use rationale added. The only images as far as I can tell that might not have them are the ones from the past 2 or so months. There might be older ones, but Beta's bot pretty much killed that category in time for the March 2008 cutoff.

At this point most images have a rationale, rather than a lack of one. While creating a rationale isn't too hard for a bot, the big (size-wise) issue in terms of the fair use policy is the strength of the rationale, so focusing the bots on the strength is understandable.

So if someone actually wrote a bot that could write and do proper fair use rationales, would that be a bad thing?

I feel from reading this thread I'm missing some Great and Special Insight on that.

Oh no doubt that it would be a great thing. It just happens that to write a bot to do everything would be impractical and probably down right impossible.

Aside from logos/covers, most images fall out of "easy rationale" range (images to illustrate something) , as there's no way for a bot to write an accurate rationale. (Nominally, I'd consider profile photos for bios in the easy rationale category, but unless they're dead or a hermit, the policy is to get free images of them.)

It would be a great thing, but the ignorant masses who have a fixation on eliminating fair use altogether will never let it fly. I half expect them to come up with a new hoop to jump through and require new tags on the already tagged images at some point.

Well, it'd put a lot of people out of business.

Or maybe they could go write articles?

One can hope!