Many of my posts which have links to blog articles such as https://fharrell.com/post/po are having the URLs systematically changed to http://www.fharrell.com/p/blog-page.html which is not even a valid page. Help appreciated!
-
3$\begingroup$ I am trying to reproduce this behavior, but to no avail. I was able to fix the bad link at stats.stackexchange.com/a/571910/919. Could you provide a little more detail about how these links are being changed? What exactly is happening? $\endgroup$– whuber ModApr 17, 2022 at 19:01
-
$\begingroup$ Regular links like hbiostat.org/rms or fharrell.com/post... are being changed to fharrell.com/p/blog-page.html and I'm fairly certain there was nothing I did to make that happen. That's because my google analytics don't show hits on /p/blog-page.html until very recently. $\endgroup$– Frank HarrellApr 17, 2022 at 19:52
-
2$\begingroup$ I get that, but what do you actually observe? For instance, does the change happen when you post something, or does it occur quietly over time to existing posts? Does it happen right when you type or paste a link into a post? $\endgroup$– whuber ModApr 17, 2022 at 20:43
-
1$\begingroup$ IIRC any kind of automated link correction subsequent to posting ought to show up in the post history as an edit by the 'Community' user, & I haven't found any of those. $\endgroup$– Scortchi - Reinstate Monica ModApr 18, 2022 at 8:12
-
$\begingroup$ Sorry @whuber - the change is made weeks after the post was made in some cases without me editing the post. $\endgroup$– Frank HarrellApr 18, 2022 at 11:10
-
$\begingroup$ I notice all the ones that remain were posted in 2017. This makes me wonder whether something happened in that year alone that you have just discovered. Are you aware of any other flawed URLs besides those? $\endgroup$– whuber ModApr 18, 2022 at 13:44
-
1$\begingroup$ I don't know of any other flawed URLs. The first attempted access to blog-page.html was Jan. 8, 2018. If the number of replaced URLs doesn't grow I can go back and edit the 8 or so occurrences. $\endgroup$– Frank HarrellApr 18, 2022 at 14:33
1 Answer
Web archive
These links seem to have been in place at least since October 2020, at least one and a half year ago.
We can see this in the following archived webpage which contains the link http://www.fharrell.com/p/blog-page.html
Stackexchange dump
It is possible to dig further by using the database dump from stack exchange.
It seems very unlikely that the system can change the body text without leaving a trace in the post history. That history can already be viewed via the website and there is no trace in the history like user 'community' editing some link or style.
Anyway, if there would be such a change that did not show up in the post history then it should be still possible to see it in the database dumps. Because I have currently no access to my laptop, I can not perform that search.
Potential explanation
The web address ending with /p/blog-page.html
is not so strange. It might stem from the blogspot webpage: https://feharrell.blogspot.com/p/blog-page.html
It could be possible that you wrongly combined two url's. This might explain why, as Whuber mentioned in the comments, only posts from 2017 have the error.
Other occurances (and potential explanation 2)
The link also occurs in comments
Analysis of variance for binary data
Some intuitive and simple approaches are discussed in BBR Chapter 16 - see link at fharrell.com/p/blog-page.html
And on twitter
https://mobile.twitter.com/5_utr/status/1497980771432386563
Clinical trialists: please quit computing change from baseline. See BBR Chapter 14: http://fharrell.com/p/blog-page.html
On twitter it seems that also others have linked to the broken links:
I can't search this easily because I don't have a Twitter account and that makes Twitter evil towards me when I try to search it's pages. But anyway, those references might indicate that possibly you were using your website in a different mode with different links around 2017.
And yes indeed, the page https://www.fharrell.com/p/blog-page.html
used to exist back in 2017. Or at least, the internet archive seems to agree with me. This page can be found in the archives
On 27 December 2017 it looked like this: https://web.archive.org/web/20171227182433/http://www.fharrell.com/p/blog-page.html
On 18 January 2018 it was a 'page does not exist' webpage: https://web.archive.org/web/20180118062901/http://www.fharrell.com/p/blog-page.html
-
1$\begingroup$ Thanks for the exceptional investigation. I can only guess that when I copy and paste from the browser addressbar that some kind of extraneous URL pieces were brought in by my system. I'll hand edit all the occurences and I added a reply to that tweet with the correct address. $\endgroup$ Apr 19, 2022 at 16:57
-
1$\begingroup$ @FrankHarrell there are also 12 comments (aside from the answers). You can see them in this image: i.stack.imgur.com/jT4rM.png You can obtain them with a query on this page data.stackexchange.com/stats/revision/1581447/87442/… (I can't copy paste the links from the CSV file because my phone does not read CSV files). $\endgroup$ Apr 19, 2022 at 17:12
-
1$\begingroup$ On twitter it is more than just that single example. See my recent update of the answer. $\endgroup$ Apr 19, 2022 at 17:18
-
2$\begingroup$ If you have the capabilities on your webserver or if your host allows it, you might be able the redirect the traffic to a different webpage instead of the standard 'page not found'. This would require access to the rewrite engine. A simpler solution would be to manually create a page on the physical address and have the redirecting done by using JavaScript or by making a custom-made 'page not found' page. $\endgroup$ Apr 19, 2022 at 17:27
-
2$\begingroup$ +1 for "I don't have a twitter account". I'm not on twitter & I'm perfectly content with that state of affairs. I often read about other people tearing their hair out about some nonsense on twitter. $\endgroup$ Apr 20, 2022 at 11:13
-
1$\begingroup$ Thanks for the super help @SextusEmpiricus - I made all the corrections manually. To bad you can't edit comments; I just added additional comments. $\endgroup$ Apr 20, 2022 at 16:27