User talk:ClueBot Commons/Archives/2023/February
 | This is an archive of past discussions with User:ClueBot Commons. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Battleground Mentality?
What's up with the United States bomber images? 2600:1003:B842:B0DA:6461:7AE3:F56B:4C92 (talk) 05:20, 5 February 2023 (UTC)
Can you please provide more on what you mean by this. This appears to have nothing to do with ClueBot Esolo5002 (talk) 19:39, 5 February 2023 (UTC)
- Please take a look at the user page for any of the Cluebot accounts. There you will find images of United States bombers on each of them which are captioned as the bot being in action. Also, "Enduring Encyclopedia" mentioned in the captions is a play on operation "Enduring Freedom" which many people outside of the United States consider to be prolonged unjustified aggression. 2600:1003:B847:B16C:DA8:609A:1959:B83C (talk) 19:45, 5 February 2023 (UTC)
- As I've explained back in 2015, the use of military vehicles is a metaphor and a sort of long-running tradition for anti-vandal bots on Wikipedia—in fact, the tradition is as old as anti-vandalism bots themselves, with Tawkerbot2 back in 2006 being the one that started the tradition. —k6ka 🍁 (Talk · Contributions) 21:52, 5 February 2023 (UTC)
How does ClueBot do what it does...
Ok... ClueBot NG came along and reverted some vandalism on the Rebel Wilson with this edit. YAY ClueBot! So, my question it...How does it *know* to actually do that? heh, and explain the how of it all as simply as you can if you don't mind. Thanks, Shearonink (talk) 23:10, 5 February 2023 (UTC)
- You might find the FAQ for all of the Cluebots informative. The TL:DR version is datasets and filters, if I understand it correctly. The folks who run the bots are playing with stuff way over my head though, so don't quote me on that. Millahnna (talk) 23:22, 5 February 2023 (UTC)
- @Shearonink and Millahnna: The TL;DR is: machine learning! All the hype over machines learning and becoming as smart as humans? ClueBot NG was doing that way back in 2010. The long and short is that ClueBot NG learns what is vandalism and what isn't from a database of edits that have been manually marked as either vandalism or not vandalism. In recent years, you've probably heard a lot of controversy over companies collecting large quantities of data in order to train their bots—this is for the same principle that makes ClueBot NG work, actually: ClueBot NG's database contains a lot of collected edits in order to train the bot on how to identify vandalism. If the database is small, then the bot will have a much harder time learning what is and isn't vandalism, so the idea behind machine learning is that you get as much data as possible so the bot can learn from it. The net result is that ClueBot NG is a much better vandal fighter than all previous anti-vandal bots, because ClueBot NG is basically trying to think more like a human rather than just using a simple set of filters, and will get smarter as the database of edits expands. —k6ka 🍁 (Talk · Contributions) 02:07, 6 February 2023 (UTC)
- Oh that sounds familiar. I think I might have actually been around for part of the manual marking. I was very active in 2010. Millahnna (talk) 02:21, 6 February 2023 (UTC)
- K6ka So it just runs and no one specifically tells it at one time or another what to do, the operators might turn it off or turn it on...it's not like some of the other tools around here that are enabled and run by editors on one edit or another then... Thanks Millahnna for the FAQ, very helpful & greatly appreciated. Shearonink (talk) 02:45, 6 February 2023 (UTC)
- Oh that sounds familiar. I think I might have actually been around for part of the manual marking. I was very active in 2010. Millahnna (talk) 02:21, 6 February 2023 (UTC)
- @Shearonink and Millahnna: The TL;DR is: machine learning! All the hype over machines learning and becoming as smart as humans? ClueBot NG was doing that way back in 2010. The long and short is that ClueBot NG learns what is vandalism and what isn't from a database of edits that have been manually marked as either vandalism or not vandalism. In recent years, you've probably heard a lot of controversy over companies collecting large quantities of data in order to train their bots—this is for the same principle that makes ClueBot NG work, actually: ClueBot NG's database contains a lot of collected edits in order to train the bot on how to identify vandalism. If the database is small, then the bot will have a much harder time learning what is and isn't vandalism, so the idea behind machine learning is that you get as much data as possible so the bot can learn from it. The net result is that ClueBot NG is a much better vandal fighter than all previous anti-vandal bots, because ClueBot NG is basically trying to think more like a human rather than just using a simple set of filters, and will get smarter as the database of edits expands. —k6ka 🍁 (Talk · Contributions) 02:07, 6 February 2023 (UTC)
Archived a section that should not be archived
Why did User:ClueBot III archive this section when it's marked not to be archived until 2030? Vanjagenije (talk) 18:36, 15 February 2023 (UTC)
- @Vanjagenije: (talk page stalker) From User:ClueBot_III#How does ClueBot III know when to archive my page?, "ClueBot III works based on the revision history of the page in question, and not on the timestamps associated with signatures." -- John of Reading (talk) 18:50, 15 February 2023 (UTC)
- This has nothing to do with the timestamps and/or signatures. This is about a sections that should never be archived. And, the bot has always understood this until now. Vanjagenije (talk) 19:08, 15 February 2023 (UTC)
- @Vanjagenije: OK, I've poked around in the history more deeply. Until this edit in October 2021, the page was archived by bots following the instructions at User:MiszaBot/config. Those bots look at the timestamps in signatures, and the section was protected from archiving by a hidden signature with a future timestamp. That was added by this edit in 2012 and modified by this edit in 2021.
- But after the switch to ClueBot III, those embedded timestamps had no effect, since Cluebot III looks at the page history, not at the signature timestamps. The section became eligible for archiving about 90 days after it was last edited. I have de-archived the section again, and applied a new {{Do not archive until}} instruction that adds a different invisible code that Cluebot III does check for. -- John of Reading (talk) 08:28, 16 February 2023 (UTC)
- This has nothing to do with the timestamps and/or signatures. This is about a sections that should never be archived. And, the bot has always understood this until now. Vanjagenije (talk) 19:08, 15 February 2023 (UTC)