Misplaced Pages

This is an old revision of this page, as edited by Hesperian (talk | contribs) at 00:12, 9 October 2011 (→Arithmetic error?: so write a bot that checks for arithmetic errors in conversions, and fixes the ones that are wrong.). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

Russian battleship Sevastopol (1895) again at Featured article candidacies

Hey Lightmouse, I see you were a reviewer at one of Sevastopol's many reviews. As it's last FAC was closed due to low participation, I"d like you to come and review it for it's current FAC, in order to get a better picture of its current situation. Thanks, Buggie111 (talk) 02:17, 4 October 2011 (UTC)

Dynamic conversion deemed bug.

I question the benefit of using {{convert}} to do dynamic conversion of (e.g.) km to miles (as you have done in various articles), especially where the equvialent has already been provided. Such conversions need to be done only once, not every time an article is generated. And if an editor feels a "round" number is appropriate then you should not be unilaterally replacing it without prior discussion. – J. Johnson (JJ) (talk) 22:38, 4 October 2011 (UTC)

Thanks for contacting me. I welcome feedback. Just on the technical side of things, can you clarify what you mean by:

• "conversions need to be done only once, not every time an article is generated"

? Lightmouse (talk) 22:43, 4 October 2011 (UTC)

I agree, and I don't think there is support for rolling out {{convert}} on articles that have already been manually converted, and I don't think Lightbot has approval to do this, and this is a deceptive edit summary. Hesperian 23:36, 4 October 2011 (UTC)

It's generally known that template conversions are more reliable than manual conversions. But many editors, possibly including yourselves are unaware of just how frequent the errors are. For example, at least two of the four articles reverted by user:Johnson contained errors that were fixed by the template. The general error rate isn't as high as that but they're significant. Humans just aren't as good at arithmetic as computers. Arithmetic errors are almost undetectable by human editors. I've seen some that have been there for years. There are also many format defects and inconsistencies.

I note the comment: "conversions need to be done only once, not every time an article is generated". Is that the source of your unhappiness with templates? Lightmouse (talk) 00:03, 5 October 2011 (UTC)

{{convert}} is syntactic sugar. It makes things easier for experienced editors, but erects a barrier to entry for newbs by replacing simple, intuitively understandable wikicode with an opaque template call. Hesperian 00:12, 5 October 2011 (UTC)

My watchlist is full of these diffs, as far as I can tell you're simply doing a mass rollout of the {{convert}} template under a deceptive edit summary. I'm blocking. Hesperian 00:06, 5 October 2011 (UTC)

Hey, that's unreasonable. There's no evidence that I'm trying to deceive you or anyone else. You haven't provided a reason why the edits are harmful and yet I've explained why they're helpful. You've made multiple assertions and instead of a two-way debate, you reach for your gun. A thirty minute cycle from asking a question to pulling the trigger. Shame on you! Lightmouse (talk) 00:16, 5 October 2011 (UTC)

I blocked a malfunctioning bot account. Bot accounts have no claim on my civility. Nor has blocking a bot account any impact on my ability to engage in discussion. Hesperian 00:23, 5 October 2011 (UTC)

I apologise for my use of the word deceptive though. The following diffs, being all of the Lightbot edits to appear on my watchlist today, all share the trait of having an edit summary that does not reflect the true nature of the edit: . In every case, the edit is a straightforward rollout of {{convert}} over the top of manually converted units. I assert that this is harmful; that you are not approved to do it; and that the edit summary is misleading. I withdraw the unintended assertion that you are being deliberately misleading. Hesperian 00:29, 5 October 2011 (UTC)

You need to be civil with humans. I'm a human. Blocking without discussion wasn't reasonable or necessary. Shoot first is a bad policy. It may make you feel good but it's not good for the project. Put your gun away and behave respectfully with me as I've been behaving with you. Your assertion of harm needs to be backed up with a statement of what the harm is. If you don't like the edit summary, you could tell me what edit summary you prefer. Hardly a blocking crime. Sheesh, please cooperate with the Misplaced Pages project.

Now, just on the technical side of things, please can you tell me:

• the block isn't visible to me on the Lightbot page. Why is that?
• I understand that there is a formal process to request unblocks but that isn't shown on the Lightbot page. What is it and is that also supposed to be visible to me?

Lightmouse (talk) 00:40, 5 October 2011 (UTC)

I guess it is up to you how you interpret me blocking your malfunctioning bot. You can consider it uncivil if you wish. I've been considered uncivil in the past for all sorts of odd reasons, such as my correcting an editor's article prose. You're also at liberty to employ rhetoric like "crime", "put your gun away" and "please cooperate with the Misplaced Pages project". And feel free to diminish my concern at your misleading edit summary by mischaracterising it as "I don't like it". It's all good. Whether my block was "reasonable or necessary" is of course a matter of opinion, not fact. I think it was entirely reasonable, and I think the requirement that it be necessary is setting the bar a bit too high. The bot was doing unapproved things that I don't think it should have been doing, under a misleading edit summary. In my opinion, that is sufficient justification to block it pending resolution. Hesperian 01:31, 5 October 2011 (UTC)

As for the technical issue, there's no template on the Lightbot page because I didn't template you, because I notified you in discussion, because I don't know of an appropriate template for this situation, and because I subscribe to at least some aspects of WP:DTTR. Formal processes around unblocking generally assume that the person objecting is constrained in their choice of actions by virtue of being blocked. You're not blocked, and are free to go off to AN/I or BRFA or wherever you want to seek block review on your bot. Hesperian 01:39, 5 October 2011 (UTC)

Sorry for butting in. A note to Lightmouse: templates do overload long articles due to the wikimedia parsing limits, and substituting them with plain text does speed up loading (experimentally verified). Materialscientist (talk) 00:58, 5 October 2011 (UTC)

You're welcome here. I've seen people debate it before and conclude that templates are fine. My technical knowledge doesn't permit me to say one way or the other. When the block is lifted, I'd be happy to try and track down an appropriate venue for such a discussion. Have you any idea how I get the block lifted (I thought there would be a standard template with instructions)? Lightmouse (talk) 01:20, 5 October 2011 (UTC)

To Hesperian, the bot IS approved to perform such conversions. See BRFA/Lightbot 5/6/7/8/9/12/13/14/17, amongst others. Headbomb {talk / contribs / physics / books} 01:36, 5 October 2011 (UTC)

Looking, can't find it. Where is the approval that covers this diff? The diff does not "add conversions"; the conversions were already there. It does not "delink common units of measurement"; the units were not linked. It does not "modify template parameters"; no templates were in use. All the diff does is replace already converted units with a template call. Hesperian 01:51, 5 October 2011 (UTC)

That falls under the general task of janitorial edits to units. Those have been trialled, had plenty of opportunity for comments, and have been going on for months if not years. In this case, the bot is fixing overly precise conversions, per MOS:CONVERSIONS. 1,800 km should be converted to 1,100 miles, not 1,120 miles. Headbomb {talk / contribs / physics / books} 01:57, 5 October 2011 (UTC)

Is there any approval request in which Lightmouse has stated their intention to replace every unit conversion, whether correct or not, with a call to {{convert}}, and this has been approved? I'm not really interested in whether Lightmouse (or you) can identify an approval that is worded in such a way that this can be filed under it after the fact. I'm interested in whether there is community consensus to roll this out. Hesperian 02:14, 5 October 2011 (UTC)

That edit also changed "(5 miles)" to "(5.0 miles)", which seems inappropriate. Dicklyon (talk) 02:22, 5 October 2011 (UTC)

It also corrects missing non-breaking spaces. Furthermore, the edits reverted by user:johnson (who started this thread), were fixes to incorrect arithmetic in two out of four cases. Reverting error correction isn't good. Lightmouse (talk) 02:05, 5 October 2011 (UTC)

• "I guess it is up to you how you interpret me blocking your malfunctioning bot."—No, Hesperian, read the WP:ADMIN policy: it's up to you to explain it. Tony (talk) 05:48, 5 October 2011 (UTC)
  • Um, the comment you quote is my response to the assertion that blocking Lightbot was an act of incivility against Lightmouse. The way you've pulled my words out of context is surprisingly clumsy for you, Tony. If you require further explanation as to why I blocked Lightbot, you might try asking me a direct question. Hesperian 05:58, 5 October 2011 (UTC)

The Admin policy clearly states what is expected of you when you use the tools. Among these expectations are concerns about:

• "Failure to communicate – this can be either to users (e.g., lack of suitable warnings...
  • Within 30 minutes of starting a dialog, you implemented a block without warning.

• "Administrators are expected to lead by example and to behave in a respectful, civil manner
  • You called me "clumsy".

You do good admin work, but this is not your best day as an admin. Lightmouse is one of our most experienced and skilled gnomes; while he has made mistakes at certain times, you owed him more than a sudden blocking; that kind of thing is for IP obscenity vandals. Tony (talk) 08:05, 5 October 2011 (UTC)

"You called me 'clumsy'." Again, Tony, I am surprised. I would expect an experienced writer like your good self to be capable of accurately parsing the assertion "The way you've pulled my words out of context is surprisingly clumsy for you, Tony.". This should be interpreted as labelling your action as clumsy, and you as not characteristically so. But <shrug> as I said to Lightmouse above, you're entitled to label me uncivil if you so wish. Hesperian 11:07, 5 October 2011 (UTC)

Can I just point out that Lightmouse hasn't been blocked - the bot has, in accordance with the standard header at User:Lightbot which says "Administrators: if this bot is malfunctioning or causing harm, please block it." For what it's worth I agree that the bot appeared to be causing harm, for the reasons Hesperian has set out above, and those that PamD has in the next section. —SMALLJIM  10:40, 5 October 2011 (UTC)

• @Lightmouse. Whilst I wouldn't consider the above examples 'errors', it is clear that Hesperian does. This, as xhe says, is a matter of opinion that can be discussed. In the meantime, I suggest that you put a 'stop' button like this one on Lightbot's user page. This will obviate the necessity of admin blocks, and would also help attenuate the usual angst that goes with such blocks and allow for a less recriminatory action–response. --Ohconfucius 16:16, 8 October 2011 (UTC)

Dynamic conversion

As you all seem to have gotten wrapped up about Lightmouse's behavior I am splitting off two sub-issues.

First: Lightmouse, the dynamic conversion is where the conversion is calculated everytime the article text is rendered for someone. As opposed to an editor -- or even a bot -- calculating the conversion once and inserting it into the article text. Dynamic calculations are useful where something is frequently changing, but although it is suspected that the kilogram bar is gaining weight it is not at a rate we need to be concerned about. _ J. Johnson (JJ) (talk) 21:29, 7 October 2011 (UTC)

It's an important question that deserves an answer. I know benefits but I don't know the technical cost of templates. It's an issue that goes beyond Lightbot so I've passed the question on to Template_talk:Convert#Dynamic_conversion. Lightmouse (talk) 15:55, 8 October 2011 (UTC)

When the issue is raised, a frequent answer is that performance doesn't matter. --Ohconfucius 16:18, 8 October 2011 (UTC)

True, as being a frequent answer, and even true in that we shouldn't sweat the details. But this kind of constant recalculation is totally unnecessary. _ J. Johnson (JJ) (talk) 18:50, 8 October 2011 (UTC)

I've passed this generic question about templates on to Template_talk:Convert#Dynamic_conversion. Let's continue the discussion there. Lightmouse (talk) 19:43, 8 October 2011 (UTC)

Arithmetic error?

You have (at least twice) said that your bot corrected "errors". Oh? If so, perhaps you should elaborate. Are you talking about out and out definite errors (on the order of "1 mile = 100 km", or "2 m = 1 ft")? Or are you taking about precision? It is often appropriate to round off numbers (esp. when they were rounded-off or approximated in the original), and I don't think your bot has the smarts to do that in any generallly acceptable way. If you were concerned about errors, then a more acceptable response would be to tag them for attention, not unilaterally blow them away. _ J. Johnson (JJ) (talk) 21:29, 7 October 2011 (UTC)

Your four reverts were:

• Straight Creek Fault 90 kilometers (56 mi) -> 90 kilometers (54 miles)
• Brothers Fault Zone 750 km (470 mi) -> 750 km (450 miles)
• Seattle Fault 70 km (43 mi) -> 70 km (43 miles)
• Puget Sound faults 125 km (78 mi) -> 125 km (75 miles)

I was trying to tell you that two (possibly three) reverts re-added errors of arithmetic that had been fixed. Despite me telling people twice that they were there, nobody said "ah yes, I see them now". Arithmetic errors go undetected for *years* because who's checking? Not you or me.

Three reverts removed non-breaking spaces that had been added.

Precision is part art and part science. The source unit is there for any analysis. By definition, the conversion in parentheses is secondary. It can be recalculated if a later reader or expert analyst challenges the precision. It's very common to see phrases like "the wood is about a mile (1,609 m) from the town)" which is weird. If precision matches within +/- 1 sigfig, I'm fine. I don't always agree with the template defaults and frequently use non-default precision.

That's three issues addressed by the template, there are more (e.g. fixing superscript issues, fixing abbreviation issues).

Your raising this issue came as a surprise but your question seems reasonable. I've done a *lot* of work on units over many years. I've probably added more manual conversions to WP than any other editor. I remain a big fan of manual conversions in addition to the template. I hope that this can be resolved without anymore negative action. Lightmouse (talk) 16:41, 8 October 2011 (UTC)

"Arithmetic errors go undetected for *years* because who's checking? Not you or me." So write a bot that checks for arithmetic errors in conversions, and fixes the ones that are wrong. That would be easy to write — I say this as a long-time user user of the pywikipedia framework. The hardest bit would be nailing the regular expression that captures the manual conversions, and you've already done that bit. Then, get approval to run it — you'll get no opposition from me, as long as you're fixing targetted errors, not indiscriminately templatizing. Then, run it under an accurate edit summary. Hesperian 00:12, 9 October 2011 (UTC)

Such "errors" as you point out are entirely matters of precision (rounding?) rather than out-and-out mistake. (And "70 km -> 43 mi" as a correction to "70 km -> 43 mi" is either more subtle than meets the eye, or ridiculous.) These are not "errors", but only variance from your personal preference for greater precision; I could just as well claim that your results are erroneous because I want greater precision. And while more precise numbers might have been acceptable, the issue here (in part) is your heavy-handed insertion of {{convert}}.

This is the key point that you don't seem to understand: the issue is not a few supposed errors of arithmetic, but your heavy-handed response. You could have tagged the alleged problems, you could have asked to change them, even a simple replacement of the converted figure might have been accepted. But unilateral introduction of {{convert}} is not good.

And "there is more"?? I'm not certain I want to know what kind of "subscript issues" and such you have also been "fixing". That you have done "a *lot*" is a bit scary, considering what we have seen in this case. I would urge you consult with other editors before launching any major effort. At the very least, please don't run this particular bot any more. _ J. Johnson (JJ) (talk) 19:46, 8 October 2011 (UTC)

I don't understand. Arithmetically, 90 km is 56 miles, not 54 miles. Lightmouse (talk) 19:49, 8 October 2011 (UTC)

Strictly speaking, no. Because it is actually some small fraction more or less. What ever figure you come up with I can always trump with a figure with more significant digits. It is not a matter of "is", but how close, for varying amounts of "close". But please note, as I said above: the issue here is not arithmetical, but how you went about "fixing" things. _ J. Johnson (JJ) (talk) 20:30, 8 October 2011 (UTC)

You created this section to discuss arithmetic. The correct conversion is 56 miles but you replaced it with 54 miles. I see that as an arithmetic error. I accept that your key issue is not arithmetical, will you accept there is such a thing as arithmetic error? Lightmouse (talk) 20:57, 8 October 2011 (UTC)

Bulk replacement of existing manual conversions

Lightmouse, you say "I remain a big fan of manual conversions in addition to the template", yet starting with these edits on 1 October your bot has been doing bulk replacement of existing manual conversions with ones that use the template. That's what I found unhelpful. Why did you set the bot to make these changes?  —SMALLJIM  20:09, 8 October 2011 (UTC)

Which is right?

When an article has an existing conversion, how does the bot decide which value to believe? See Ledston. " 4km (3 miles)" (introduced in 2006) was altered by the bot to use the convert template and read "4 km (2.5 mi)", and has now been manually changed by another editor (still using {{convert}}) so that it reads "3 miles (5 km)". I don't know which is correct (though miles-first is certainly correct as it's a UK village). The bot can't tell which of the two figures is from the editor's source, as various factors will have affected their decision on which units to place first and which in brackets (existing style of article / convention for geog area / personal choice).

In hindsight it might have been wiser for the bot to concentrate on unconverted measures and delinking, and perhaps to flag up conversions which appear to be inaccurate for human attention. Some will be real mistakes, some will be approximations in situations where exact accuracy is not expected.

In some cases, I suspect that the bot has changed an article so that it no longer reflects the source from which the data is obtained - I looked at one of the examples cited above, , and I guess that in the first change "over 2,000 miles" was Forrest's own estimate of the journey ("by Forrest's reckoning" as it says), so that the bot has damaged the article by converting this to 2,200. PamD 08:12, 5 October 2011 (UTC), expanded 09:58, 5 October 2011 (UTC)

(which I hope your monitoring for comments re: its unblock) (talk→ BWilkins ←track) 14:00, 6 October 2011 (UTC)