This is a detailed consideration of the Edit filter. It's long. If allergic to Abd Walls of Text, consult your medical professional, do not read.
QUOTE(EricBarbour @ Mon 16th May 2011, 2:19am)
![*](style_images/brack/post_snapback.gif)
QUOTE(melloden @ Sun 15th May 2011, 7:43pm)
![*](style_images/brack/post_snapback.gif)
There are plenty of more important edit filters to discuss. Like the hidden ones.
Care to share some hidden filters with the rest of us miserable wretches?
Filter 407 is hidden. The only reason that we know it is an "Abd" filter is that I reported it and I and others, deliberately or accidentally, tested it. The filter produces a report, "Disruptive block evasion," and there is a bot set to copy the report to IRC.
I'll agree that this particular "incident" is minor. However, often abuse is exposed through minor activities, one reason being that major ones are often confusing because of POV conflict, and it's easily taken with "a pox on all your houses."
Th edit filter log, by the way, is not visible from blocked IP or by any blocked editor. That, alone, is an interesting factoid. It is visible to unblocked IP. I know of no other pages at Wikipedia -- besides the edit screens -- that are not viewable based on blocked status. So
if you are blocked because of an edit filter report, you can't see the entry in the log.For reference, the filter log.
Log for Filter 407. If you are sitting on a blocked IP and don't have an account to bypass it, you can see
a copy of the filter log, as of May 15. I'm maintaining this page every few days, and it shows if IP was me and if it was blocked.
Recent filter changes does not show hidden filters like 407.
History of Filter log 407.To answer another question, edit containing Paula Abdul. triggered the filter on 08:41, 9 May 2011. The filter was edited after that, but, of course, we can't see the edits. Hidden filter. That edit didn't add "Abdul," which was already in the edited section. Naughty, naughty, Timotheus Canens!
Edit filters should be run in log-only mode until and unless it is established that false positives are rare. Since blocks are not automatically issued, none of the exclusion action of the edit filter is actually effective for anything, since I don't need to mention my name directly, or at all, often. This filter triggered 11 times over a 12 hour period, as I probed it. It then has triggered 17 times over the next week. I believe that none of these were me.
Watching that filter is wasting the time of many admins.If they are on IRC, they will see the "disruptive block evasion" report, and, as in at least one case, they may knee-jerk block an editor, particularly one who has read these reports and who decides to test the filter, by making a harmless contribution. "Deliberately triggering the edit filter" -- which isn't an accurate description of testing behavior, by the way -- is a New Block Reason.
Testing is finding out what triggers the filter, and if the edit is one that should be legitimate, there is a positive purpose in it. Deliberately triggering the filter(s) would be disruptive. I could do that, if I were so inclined. Not yet. I have seen nothing that rises to the level of disruptive triggering.
Each one of the false positives does damage, since the filter is operating on block-edit mode. The filter message gives the editor no clue why the attempt is being rejected. Some false positives are not resulting in any sanction, but, in one case, I attempted to add text that an IP had added, rejected because of a spelling error (abd for and) and it was reverted and my IP blocked. (This was also reported by me to the False Positives page, and the report was reverted and the IP blocked. It used to be that helpful positive actions by a blocked/banned editor would be iallowed, but Our Will Be Done has become paramount, and what's insane here is that there seems to be no concern over collateral damage. After all, these are "only" IP editors.)
There is no policy requiring the reversion or automated blocking of edits by blocked editors. It is
allowed and I have supported that policy. I was even blocked once on Wikiversity for routinely reverting the edits of a blocked editor, my goal in reverting was to:
(1) Enforce the block (which was for serious disruption, involving outing, if I'd been an admin, some of the edits would have been revision-deleted -- and some later were).
(2) Insure that the reversion was "per block," not "disruption" because this establishes no position on the content itself, thus allowing reversion back in of some of the edits. Historically, with this editor, most of the edits were good, I'd done this quick reversion, with later review and restoration. Most edits were restored.
(3)Self-reversion was suggested to this editor, but, in fact, the editor took an opposite tack: he revert warred over the removals. For him, it was His Will Be Done. I was blocked because, given the lack of custodial attention, I was repeatedly reverting, and that irritated a few editors. (Self-reversion shows cooperation with a ban or block, it is merely a cooperation that keeps limited communication open, instead of Go Away Forever "cooperation."
If they wanted to use the Edit Filter to enforce a block, there would be excellent ways to do it that don't censor content, but that result in bot reversion all contributions of a blocked or banned editor, automatically, with an informative edit summary and reference to a page that tells users that they can revert this material back in (or restore the removal that the banned editor did), by taking personal responsibility for it. This would be automated self-reversion. (The editor allows it by using an account that is identified or by using IP on a list.) Far more flexible and this could be -- should be -- fully open.)
The basic problem isn't Timotheus, would that it were that simple! The problem is lack of coherent and intelligent and efficient structure, designed to foster genuine consensus.Efficient structure designed to exclude participation entirely, rather than channeling it into efficient consideration, is fascist, corrupt, it's predictable that it will ultimately fail. The edit filter is being used for this kind of efficiency, to enhance the personal power of the admin who created the filter, and those also inclined to rBI, yes, emphasis on "block" and "ignore." Problem is, intensive work on the Edit Filter, and blocking for any testing of it, is not Ignore, it's Attention. "Efficient" block enforcement, meaning strict enforcement, requires continuous attention.
The biggest problem with self-reversion, practiced alone, is that the edits under ban may be ignored, even if they are excellent, even when they correct obvious errors. The ignoring of these edits represents a common Wikipedia problem, that frequently nobody is watching, or, if someone sees the edit, they don't have time to review it, and it then slips unnoticed into the landslide of History. So, anyone practicing self-reversion may need to develop connections with other editors who remain aware of the contributions.
This requires cooperation. That's one reason why I claim that
self-reversion develops cooperation, and the history of the practice shows that.
Cooperation scares the hell out of certain administrators for reasons that I won't explain today!