(Note: This question might be off-topic for MO but the only other plausible alternative that comes to mind is Academia Stack Exchange, and there are some features of this question that are, to some extent, peculiar to research mathematics.)
I recently learned that it is more difficult than I expected to document, in a permanent manner, a correction to an error in a published paper (let me call this an "erratum" although some people call it a "corrigendum").
The most obvious approach is to publish an erratum in the same journal that published the original paper. However, many journals have a policy that the erratum has to go through the full refereeing process all over again. While I don't personally know of any case where the erratum was rejected, the nuisance of having to go through the refereeing process is enough to discourage some authors from bothering to submit a formal erratum.
Another approach is to try to post the erratum on the arXiv. However, it seems that the arXiv has certain policies on errata that may also pose obstacles. For example, suppose that the original article is an old one that is not already on the arXiv. Apparently the arXiv will not accept the erratum unless the old paper is first posted to the arXiv. But in some cases there may be no way of doing this without violating copyright.
A third approach is to post the erratum on your own website. But most people's websites eventually disappear after they die, so this is not a solution to the problem of permanently archiving the erratum.
Is there any other good way of permanently archiving an erratum to a published paper?
It seems that many mathematicians take pride in, and rely on, the high accuracy and reliability of the published mathematical literature. However, if there are barriers in place that disincentivize the publication of errata, the reliability of the literature suffers.
ADDENDUM (July 2019): I thought I'd give an example to illustrate some of the difficulties. I recently submitted a corrigendum to a 2006 Notices of the AMS article of mine, "You Could Have Invented Spectral Sequences."
The good news:
- The Notices accepted the corrigendum without making me go through a full refereeing process, and published it on their website.
- If you go to the main Notices website, you can find the link to corrigenda/errata without too much difficulty (although it appears that currently, my corrigendum is the only item there).
- When I Google the article title along with "site:ams.org/notices" then both the article itself and the corrigendum show up.
The not-so-good news:
- If you didn't know about the corrigendum and just navigated to the relevant back issue of the Notices on their website, you would find no indication that there is a corrigendum.
- It's not clear that the aforementioned Google search works for everyone (e.g., it doesn't currently work if I instead try Google Scholar or DuckDuckGo).
- The Notices told me that the corrigendum will not appear in the print version of the Notices.
- I wrote to Mathematical Reviews and Zentralblatt MATH a few weeks ago to ask them if they would make a note of this corrigendum, and have not heard back from either one.
ADDENDUM (November 2019):
The MathSciNet review of my article now contains a pointer to the corrigendum on the Notices website.