Portal:Weather – Overturn. Seems like we have a pretty solid consensus that the main maintainer not being notified of the deletion discussion and the "unmaintained" argument being (potentially; there are several weighty counterarguments that a lot of outdated content was in the portal) wrong are problems so severe to warrant overturning the deletion and beginning a new one. Regarding the merits of the deletion itself it's a bit harder to assess as portal deletion discussions tend to be very open ended (we don't have a set of guidelines that describe when a portal is allowed and when not) and there is plenty of disagreement on whether the keep and delete arguments in the MFD were adequate. Jo-Jo Eumerus (talk) 09:57, 26 December 2019 (UTC)[reply]
The following is an archived debate of the deletion review of the page above. Please do not modify it.
I am the primary (only) maintainer of this portal. I have put a lot of effort over the years into making this a low-maintenance, sustainable portal that can be incrementally updated. This is precisely *why* there are so many subpages, since I've put a lot of care into ensuring that there will always be relevant and randomized, fresh content based on the current date, and there will never be broken links. I've made hundreds of edits this year alone, not just in the "on this day" section but also in the "did you know" section and adding a new featured picture. This is all the result of hundreds or maybe even thousands of hours of effort over more than a decade, searching for relevant weather events from a given date, and adding them to the "on this day" section so that they appear around the appropriate anniversary.
Many of the "delete" comments seem to have lazily just looked at the history of the main portal page, seen very few edits, and thrown up their hands and said "Well, no one's working on it, get rid of it!". If they had actually done some digging and seen the history of transcluded pages, they would have seen that yes, this portal has been heavily maintained over the years. Why is this even a proper rationale for deletion anyway?? Even if it were true, no one is even addressing the merits of the content as it exists (or rather, existed I guess).
Most annoying of all, I was not notified about this deletion discussion. When I went in to make some updates today (as I often do), I saw that the entirety of my work had been deleted, without even being able to offer a defense.
The closing admin's reasoning is completely incorrect: "There is no argument that this portal is unmaintained and serving inaccurate information to readers." Well, let me give this argument then: I have made hundreds of edits to this portal's subpages in this year alone. If anything, I've updated this Portal more in the past year than in any year since it was first overhauled over a decade ago. The nominator pointed to a single example of inaccurate information that would have been easily corrected, and used that as representative of the entire breadth of pages under the Portal. Why were the first comments pointing to WP:FIXIT ignored??
Do people want more selected articles? I can do that! Do people want a place to report inaccurate information? I can do that! Do people want instructions on how to add their own weather event to the list of 1000+ "On this day" links? I can do that! No one ever asked, so I never thought these were priorities that could lead to the entire portal being deleted out from under me.
Please restore all 2100+ pages that I have worked hard on for more than a decade, and maybe next time use a little more discretion and transparency when deleting a huge body of work like this. RunningOnBrains(talk) 19:53, 18 December 2019 (UTC)[reply]
For those without admin privileges, this tool shows the number of pages created (not number of edits, which is higher) by me in the Portal:Weather space. You can see I created more than 200 now-deleted pages in the space in the past year alone.-RunningOnBrains(talk) 20:06, 18 December 2019 (UTC)[reply]
Overturn there were two major arguments for deletion in that discussion: that the portal wasn't maintained and that it contained errors. The first argument is clearly not true, the OP was making an average of more than one edit a day to the portal and has done so for a long time. Admittedly nobody pointed this out but that just suggests the discussion was defective. The fact the OP as the portal's main maintainer wasn't aware of the discussion also weakens it. The argument that the portal contains errors contradicts WP:SOFIXIT and WP:ATD, as was pointed out in the discussion. Those are respectively a guideline and a policy and people citing them shouldn't be downweighted on strength of argument, as the closer seems to have done. Hut 8.5 22:32, 18 December 2019 (UTC)[reply]
How would the nominator have known whom to contact? Or which of the thousands of subpages were edited recently? Was an MfD tag put on the portal page, and if so, why wasn't that sufficient to notify the maintainer? – Levivich 22:35, 18 December 2019 (UTC)[reply]
Levivich having had a chance to cool my head, I can understand the lack of notification. The opener of discussion did notify the creator of the portal, and the appropriate Wikiproject. My arguments against the conclusions reached still stand, however.-RunningOnBrains(talk) 04:52, 19 December 2019 (UTC)[reply]
An MFD was placed on Portal:Weather on November 17, 2019, at 20:28 by Mark Schierbecker. Best, Barkeep49 (talk) 23:04, 18 December 2019 (UTC)[reply]
A glance at Special:RelatedChanges would have easily shown that Runningonbrains was maintaining the portal. The argument for deletion here was that the portal wasn't being maintained, it's hardly unreasonable to expect the nominator to check whether anyone is maintaining it. Hut 8.5 23:15, 18 December 2019 (UTC)[reply]
Editors need to think past the letter of the law. When nominating a single article for deletion, standard-level effort at notification is appropriate. When dealing with a collection of 2000 pages, you really need to put in extra effort to make sure everybody who is interested knows about it. In this case, the portal maintainer was off-wiki from 17 November to 18 December; the entire MFD took place during a period when they were not around. What was so important about deleting these 2000 pages that it couldn't wait for them to return? Or, maybe try a little harder to notify them. -- RoySmith(talk) 15:39, 19 December 2019 (UTC)[reply]
You're presupposing that anybody in that MfD was even aware that there was an editor who was updating some (like 10%? less?) of the 2100 subpages. It's a highly-unlikely circumstance that you have a maintainer who is regularly maintaining the portal, but only a certain part of it and not the main part, and on top of that, the editor is a regular contributor but doesn't log in for the almost-two-weeks that the portal is tagged for MfD. Now, there's absolutely nothing wrong with not logging in for two weeks, nor with maintaining some but not all of a portal. In the past year that I've been rather active in portal deletions, I've never seen this very unique confluence of circumstances: 2100 pages, some being maintained, maintainer not logging in for MfD duration. I bet nobody can name another example like this. That's a one-in-a-million coincidence. Have we ever seen before a portal with that many subpages? I don't think anybody–not the nominator, not !voters, not the closer, and not the maintainer–did anything wrong here. It's just a "perfect storm" of circumstances. – Levivich 16:48, 19 December 2019 (UTC)[reply]
RoySmith not that it matters but as a point of order, I wasn't fully off-wiki during the time of the discussion, I was definitely browsing during this time, so if I received a message I would have seen it.-RunningOnBrains(talk) 01:48, 21 December 2019 (UTC)[reply]
Overturn. Insufficient notification. While portals are now recognized as a dubious concept, that is not justification for the deletion of such a large number of old portal pages so hastily. --SmokeyJoe (talk) 23:24, 18 December 2019 (UTC)[reply]
Overturn. The closer starts by saying that There is no argument that this portal is unmaintained and serving inaccurate information to readers yet neither of those assertions is correct. The portal is maintained, as mentioned in the nom above, and a simple investigation would have revealed the status. And nobody in the discussion mentioned inaccuracy at all. I fully sympathise with the OP's annoyance at this hasty and ill-thought-through deletion, given that the community explicitly decided not to deprecate maintained portals, and this is one. Thanks — Amakuru (talk) 00:02, 19 December 2019 (UTC)[reply]
Overturn deletion. Insufficient notification, and false claims from people who clearly don't have a clue how old-style multisubpage portals work. Two errors (one relatively minor) in 2100 pages is a low error rate, though I'd suggest the maintainer either checks included bios of living people for their death on a regular timetable or sticks to bios of the deceased. Espresso Addict (talk) 02:51, 19 December 2019 (UTC)[reply]
Overturn and relist. I'm not sure I can blame the closer, given the arguments. Numerically, it's about a wash, but the keep arguments are mostly rather vague. I can't really blame the closer for down-weighting many of them. But, taking a step back, it's clear we ended up in the wrong place. Perhaps no rules were broken by not notifying the portal maintainer, but the lack of such notification clearly led to a sub-standard discussion. The closing statement says, There is no argument that this portal is unmaintained and serving inaccurate information to readers, and the deletion was based largely on that. But, the person best equipped to counter that argument didn't even know the discussion was going on. A defective notification and a close-call decision that depends on discounting some arguments don't seem like the process we want to use to delete over 2000 pages. If after another week of discussion, it still ends up as delete, that's fine, but at least Runningonbrains should have a chance to argue their case. -- RoySmith(talk) 03:18, 19 December 2019 (UTC)[reply]
Endorse but relist based on new evidence – Notification to creator, WikiProject, and MfD tag on the portal page, was sufficient notification. The close was correct: literally zero editors argued that the portal was maintained or that it served accurate information; the !keeps were "sofixit" or otherwise stated the portal needed work. I strongly disagree with comments above that the deletion was hasty (12 days isn't hasty) or ill-thought out, or that delete !voters based their !votes on the portal history (those are experienced editors who know to look at subpages). There weren't two arguments for deletion, there were three. The third was that the portal had 2100 subpages, which is unmaintainable–more pages than we have the editors to maintain. This argument was discussed explicitly in the MfD. It seems to me that the nom and other voters were focused on Selected Articles and Selected Bios (that's the nom's examples of "never-updated" and "errors"), whereas the DRV filer is saying they maintained the DYK, OTD, and FP sections–not SA or SB. I can understand how !voters looking at SA and SB may not notice, among 2100 pages, that other pages (DYK, OTD) had recently been edited. I can also understand how a maintainer can miss an MfD tag on the portal for 12 days if they happened to not log in during those 12 days–and the contribs show that Runningonbrains didn't make any edits between Nov 18 and Nov 30. So I don't think there was anything wrong with how this MfD went down at all. Nevertheless, since there is an editor who has been building and maintaining this portal for a long time, and didn't get a chance to participate in the MfD, I can't see any reason not to relist it. I believe there's precedent for portals being kept when there's evidence of maintenance, and nobody brought forward evidence of maintenance in the MfD. Rather than seeing this as a "defective close", I see this as a "relist to consider new evidence" situation. – Levivich 04:58, 19 December 2019 (UTC)[reply]
Overturn and Relist - I see some nonsense arguments to religitate the deletion discussion. I also see that the maintainer was not notified of the nomination for deletion. I don't see any plausible substantive arguments on appeal. I do see a valid procedural reason for this appeal, and that is the lack of notice. I don't think that I am required to declare that I !voted to Delete; I will make that declaration. I am arguing differently than I did at AFD because the issues to be decided and burden of proof are different. I still think that the portal should be deleted; but the portal maintainer should be given a chance to make the case for the portal. Relist it for another seven days. Robert McClenon (talk) 17:15, 19 December 2019 (UTC)[reply]
Overturn, the main objection ("unmaintained") was only partially true. The maintainer's deleted edits show hundreds of edits to portal subpages in the last two years. I can understand why the nominator did not find out who was the maintainer from just looking at the main page history. Anyway, this is probably a good warning against having only one person checking a portal -- if they go away for longer than a week, they may miss an MFD completely. —Kusma (t·c) 18:59, 19 December 2019 (UTC)[reply]
Overturn Even for the sake of argument you discount everything the maintainer wrote above, I don't even think the close was correct given the discussion, probably a no consensus at best, but I would overturn to a keep based on the flawed deletion rationale. Portal MfDs are especially terrible since there's not a lot of rules in place, so they tend to be "won" by whichever group of users shows up in force. SportingFlyerT·C 21:23, 19 December 2019 (UTC)[reply]
Overturn the anti-portal campaign needs to be brought under control and a good start would be to an overturn here. There was no consensus for deletion. Lepricavark (talk) 00:15, 20 December 2019 (UTC)[reply]
Endorse There are three components to this deletion review nomination:
"I wasn't notified" -- Nowhere do the rules say it was necessary for you to be notified, given that the portal was tagged for twelve days
"The deleters were mistaken; the portal was in fact maintained" -- Mark Schierbecker presented evidence that it was unmaintained, that (assuming I read him or her correctly) it described someone who died five years ago as still living. Several users also !voted delete on the basis of the difficult-to-maintain structure, rather than any apparent lack of maintenance. Furthermore, this entire line of argument is claiming the !voters erred, rather than the closer, which is simply not within the scope of deletion review.
"Please don't destroy my hard work" -- This is simply not a valid argument.
Pppery I disagree with your characterization of my argument. I am saying that the !voters said things that were completely false, and the closer took these incorrect things at faith, which 'is' completely within the scope of deletion review. A single error that went overlooked does not point to something being unmaintained. If that's the case I guess we should delete half of Wikipedia. Furthermore, even if this portal were truly "difficult to maintain", that is not a valid criterion for deletion that I am aware of. If standards have changed, and "hub pages" (as the nominator refers to them) are required, that is something that could easily be created. As for not wanting someone to "destroy my hard work" (your quotes, not mine), my argument is that drastic actions (the deletion of a large body of ongoing work) should be held to a higher standard for complete deletion. Maybe that's still not a valid argument, but the point here isn't my feelings, it's the process.-RunningOnBrains(talk) 00:54, 21 December 2019 (UTC)[reply]
User:Runningonbrains, Can you clarify what was said in the MFD that you think is "completely false"? (Note: "maintenance" means checking and if necessary fixing existing content, not just adding content). Re the "single error": the filer of the MFD did enough to demonstrate that the portal was not being kept up to date. DexDor(talk) 17:36, 24 December 2019 (UTC)[reply]
Overturn The discussion showed no consensus (not even a rough one) so should have been closed as "no consensus". I don't especially see the MfD's keep arguments as being "vague" (as suggested above) and in good measure the delete arguments were not based on fact. So, by counting or weighting I do not see any consensus. Thincat (talk) 10:35, 20 December 2019 (UTC)[reply]
@Thincat: I think you were referring to me regarding the vague keep comments. More specifically, arguments which consist entirely of, Keep WP:SOFIXIT (x2) or Keep per WP:ATD don't give any specific reason why this portal is worth keeping. That's what I was referring to as vague. -- RoySmith(talk) 19:05, 20 December 2019 (UTC)[reply]
Yes, I took these remarks to be in response to the three claims in the MFD nomination that the portal was not being updated. The responses therefore are saying update the portal. This would be regarded as an entirely satisfactory response to an AFD asking for deletion because the article was not being updated, wouldn't it? (And, by the way, it seems the claims were, at least partially, wrong). Thincat (talk) 19:19, 20 December 2019 (UTC)[reply]
Overturn As this seems to have been a spectacular blunder and competence is required, we should reflect on the process. WP:DGFA explains how closes should be made:
Use common sense and respect the judgment and feelings of Wikipedia participants.
As a general rule, don't close discussions or delete pages whose discussions you've participated in. Let someone else do it.
When in doubt, don't delete.
In this case, there was no consensus, common sense was not used, the feelings of the Wikipedians were not respected and benefit of the doubt was not given. I'm going to ensure this is logged as evidence in the Portal case. Andrew🐉(talk) 19:37, 20 December 2019 (UTC)[reply]
Comment I was the editor who notified the WikiProject. Editors trying to delete a portal rarely alert anyone other than the creator of its main page (often inactive for years), so I check daily for new MfDs and notify any appropriate active projects. I also make a quick check for a maintainer, but unfortunately on this occasion I overlooked Runningonbrains. Certes (talk) 22:48, 20 December 2019 (UTC)[reply]
Overturn The portal attackers should cool it. This was a maintained portal. Lightburst (talk) 00:16, 21 December 2019 (UTC)[reply]
Overturn and Relist. I have every confidence that the closer used a reasonable competence to verify claims made on both sides. However, given the somewhat decentralized structure of some sorts of portal construction, it clearly required more effort than was performed. Virtually every delete !vote made an "unmaintained" assertion; even the closing statement asserted lack of maintenance as a reason for deleting. OP's nomination here blows all that up. This case demonstrates that a guideline on portal construction is necessary; further, it demonstrates that such a guideline requires input from parties who believe in and maintain portals themselves. BusterD (talk) 02:09, 21 December 2019 (UTC)[reply]
Overturn - 'spectacular blunder', not IMO by the closer, but in the premise (that the portal was not maintained). This casts doubt on other deletions of portals. Oculi (talk) 11:25, 21 December 2019 (UTC)[reply]
Endorse per Levivich, Pppery, and others. Solid rationales provided by the above. I agree with all of it. It was a solid close. Consensus can change and there's nothing stopping nom from restoring the portal, but I see no need to restore the subpage. Just begin anew. Portals can be easily re-created. Doug MehusT·C 01:39, 23 December 2019 (UTC)[reply]
@Dmehus: Do you really think that 2100 subpages of selected article excerpts, images, "on this day", etc. could be "easily re-created"? If so, how would that be better than restoring the deleted versions with attribution history? Certes (talk) 01:56, 23 December 2019 (UTC)[reply]
@Certes: I don't it's necessary to re-create those subpages. They're just automatically generated lists of pages we featured, essentially, for this portal. Also, I don't see whyWP:ATT makes sense in the case of portals. Little original editing actually takes place in the portal namespace. Some willing editors could easily just re-create Portal:Weather relatively quickly and feature new articles going forward. No need to say what articles a previous iteration were featured. --Doug MehusT·C 02:02, 23 December 2019 (UTC)[reply]
Why wouldn't it be preferable to simply reinstate the deleted portal, which (contrary to claims made at the MfD) was being maintained? What on earth would be the benefit to completely starting over? Lepricavark (talk) 03:25, 23 December 2019 (UTC)[reply]
WP:G4? My endorse was unequivocal, not "endorse but relist" or "endorse but allow recreation". There is no indication that consensus has changed here. * Pppery *it has begun... 02:46, 23 December 2019 (UTC)[reply]
There's no indication that there was ever a consensus in favor of deletion. Lepricavark (talk) 03:25, 23 December 2019 (UTC)[reply]
In the (unlikely but possible) hypothetical scenario in which this discussion was closed as endorse, then this discussion will be that indication. * Pppery *it has begun... 04:11, 23 December 2019 (UTC)[reply]
Overturn. The closer saw no argument that this portal is unmaintained and serving inaccurate information to readers. We now have new evidence of hundreds of edits this year and that this is a low-maintenance, sustainable portal which does not require daily tinkering to justify its survival. The inaccurate information consists of one easily fixed omission of a meteorologist's death. I suggest transcluding excerpts to prevent such errors in future, but it's not a valid excuse for deletion. Certes (talk) 11:17, 23 December 2019 (UTC)[reply]
Endorse the close which didn't say there shouldn't be a weather portal, but said it could be "recreated in a more maintainable form at such time that there are editors who are committed to maintaining it regularly.", but restore the portal's main page to userspace to help the OP start creating a transclusion-based portal (should they choose to do so).
There are several irrelevant and misleading statements in the comments above - for example, "why did no one think to remove it from ..." as (1) there's no need to remove portal links (in most/all cases links to non-existant portals are not shown) and (2) doing so would (presumably) mean re-adding the links if the portal was re-created (as allowed by the close).
The XFD included statements such as that the 41 selected articles/biographies were created in 2008-2010 and never updated since (even where the subject of a biography had died). Assuming that statement is correct it indicates a lack of maintenance. The OP may have been adding more pages to the portal, but that's expansion not maintenance. A fundamental problem with portals appears to be that editors like to create/expand them, but (unlike with articles) editors don't like to maintain them (especially maintaining parts of a portal created by another editor).
That an editor put hundreds of hours into a page(s) (that few if any readers ever looked at) isn't itself a reason to keep. In fact, it's a reason for encouraging editors to do something more useful instead.
A portal consisting of thousands of copied subpages that probably had just one person (the page's creator) watchlisting each one and doesn't have strong support from other editors interested in the topic (e.g. who would copy across changes from the articles) was not maintainable in the long term. DexDor(talk) 14:18, 24 December 2019 (UTC)[reply]
The above is an archive of the deletion review of the page listed in the heading. Please do not modify it.
The decision to delete this page is the result of an extremely narrow interpretation of WP:NCORP and WP:ORGIND. As editors, our mandate is not to re-interpret editorial decisions made by legitimate media outlets. Doing so puts the entire premise of WP as an encyclopedia at stake. The analysis presented by HighKing essentially challenges the editorial decisions of an independent newsroom to publish a story about this subject. HighKing is basing their analysis on the fact that the stories about this subject contain minimal sources outside of the subject. While it might be reasonable to challenge the newsroom on their reporting, it is not our role. It is not unusual to read articles with limited sources. The subject of this page has been profiled in a number of different, legitimate, independent, media outlets. We cannot take the rigid stance that if a piece of news is not reported to a degree that we would prefer, that it is therefore illegitimate for inclusion as a proper citation. Web pages, press releases, and other obvious self-promotional channels are clearly not legitimate sources for citations. However, it is not our role to be challenging the editorial decisions of major, independent media outlets in this way. While one could understand the need for additional citations for this subject, I firmly disagree with the decision to delete based on the analysis of one individual and urge further review of how WP:NCORP and WP:ORGIND policies are being interpreted in this instance. In the meantime, the decision to delete should be reversed. Coffee312 (talk) 16:48, 18 December 2019 (UTC)[reply]
Closer's comment I feel like this is an issue that Hijiri88 pointed out at another current DRV. In weighing the outcome of that AfD I see a consensus of editors who agree with HighKing's analysis and as such there is a consensus to delete, as I noted in my closing statement. Best, Barkeep49 (talk) 17:06, 18 December 2019 (UTC) FWIW Coffee made no attempt to contact me prior to launching this DRV, but did notify me of the DRV, which I appreciate.[reply]
Endorse.
An editorial decision by an independent newsroom to publish a story does not make the non-independent story independent.
User:Cunard has a habit of reference bombing discussions with a large number of weak sources, weak in terms of demonstrating notability.
WP:Reference bombing has become a pretty standard technique for bamboozling reviewers. The answer is WP:THREE. If the best three are not good enough, no number of additional weaker sources will help. Name the best three. --SmokeyJoe (talk) 00:02, 19 December 2019 (UTC)[reply]
Endorse per SmokeyJoe above. The three best references supplied (whichever they are) don't add up to a demonstration of notability, and that was successfully argued in the discussion. — Amakuru (talk) 00:11, 19 December 2019 (UTC)[reply]
Endorse- the deletion discussion contained a bombardment of marginal or useless sources, a detailed rebuttal of them, and after two relists the consensus was clearly to delete. The WP:NCORP and WP:ORGIND might have been interpreted more strictly than some would like, but that was by the discussion participants. Not by the closing administrator, who only judged consensus. This was clearly a correct close. ReykYO! 09:02, 19 December 2019 (UTC)[reply]
Endorse I feel guilty being pinged and then failing to show up, but I didn't feel I had anything to add to this review, since it is obviously going to end in an endorse result. Perhaps we should review the deletion review process, whereby any frivolous deletion reviews that are clearly not worth the community's time are speedy-closed after one day, with the filer being warned for the first two breaches and then issued with escalating blocks for the third and later? Hijiri 88 (聖やや) 01:52, 20 December 2019 (UTC)[reply]
Oh, also the same warnings and escalating blocks for editors who repeatedly source-bomb AFDs with a bunch of news articles, books and journal articles they themselves clearly haven't read. I have seen more than my fair share of AFDs end in "keep" or "no consensus -- default to keep" results because of this kind of disruptive behaviour. (Perhaps the most disruptive example in my experience was here, although in that case there was no "bombing", since only two or three sources the keep !voters hadn't read were actually presented. This is another example, since while the problem was more !vote-stacking than people posting links to sources they hadn't read/understood, the latter still definitely occurred -- from a syspop, who even defended an IP sock of a site-banned user and never retracted said defense!) Hijiri 88 (聖やや) 02:05, 20 December 2019 (UTC)[reply]
Endorse Delete was a possible close to this discussion and the close and delete !votes are grounded in policy, I don't see any reversible error here. The mere existence of sources does not imply notability. SportingFlyerT·C 03:04, 20 December 2019 (UTC)[reply]
Endorse Cunard has a bad habit of ref-bombing AfDs and 90% of the references he posts are crap. Coffee312's belief that NCORP was/is being extremely narrowly interpreted is entirely incorrect. Reading the above, it appears that Coffee312 is suggesting that if a major, independent media outlet publishes any kind of article mentioning an organization, then we are bound to accept the article as a sign that the company meets the requirements for notability. This is wrong. It is so wrong that the NCORP guidelines were significantly tightened up about a year ago in order to remove precisely the types of references that are a mix of churnalism and interviews masquerading as independent content which are prevalent in Cunard's list of references. In order to meet the criteria for notability, Independent Content must include original and independent opinion, analysis, investigation, and fact checking that are clearly attributable to a source unaffiliated to the subject. That is not a "narrow interpretation" and its aim is to exclude all obvious promotional content generated by the company. HighKing++ 11:23, 20 December 2019 (UTC)[reply]
Endorse, but with slight reservations. This does look a bit like a supervote, but when you dig into it you find a valid reading based on misunderstanding by some editors of the impact of churnalism on "referenciness". Guy (help!) 00:30, 21 December 2019 (UTC)[reply]
The above is an archive of the deletion review of the page listed in the heading. Please do not modify it.
