Comments on: WordPress XHTML Validator Plugin http://www.movingtofreedom.org/2008/02/10/wordpress-xhtml-validator-plugin/ free software, free culture, free association Fri, 29 Aug 2008 23:43:58 +0000 http://wordpress.org/?v=2.0.11 by: Somebody http://www.movingtofreedom.org/2008/02/10/wordpress-xhtml-validator-plugin/#comment-3457 Mon, 18 Feb 2008 23:00:22 +0000 http://www.movingtofreedom.org/2008/02/10/wordpress-xhtml-validator-plugin/#comment-3457 I did use a different IP range to post. The math problem was "What is the sum of 5 + 7?" on the failed post. The answer I gave being 12, of course. It follows that it was not rejected because the math was incorrect, although that is the reason claimed in the error message. It also wasn't a transient glitch; I resubmitted it a couple of times with identical results. In fact, the originating IP and the added P.S. were the only things to differ between the failed and the successful post. Which means that the only thing that the failed post contained that the successful one didn't was a Bell Canada IP address. Which means that that was the criterion for rejection. Which means that somewhere, some software is configured to reject posts based on originating IP with an error message that incorrectly claims that they got the math problem wrong. Configured, in other words, to lie. It seems unlikely to be akismet, which apparently silently rejects posts rather than produce any error message at all. I'd suggest you examine closely the source code for the math test plugin. It may be that it has some extra "features" besides simply testing for the answer to be correct. Likely it accessed a remote database to decide what IPs to block, since it seems you didn't yourself tell it to start blocking Bell Canada IPs, yet its starting to do so must have been triggered by something. The extra "features" are probably well-intentioned spam-blocking measures, but obviously have been implemented naively. Most likely, they used a blacklist intended for blocking e-mail spam that preemptively includes the ranges of large ISPs' DHCP servers, since email originating directly from an ISP user's computer instead of going through their ISP's MX box is likely spam. It should be obvious why blocking such addresses is smart for email despamming but stupid for policing blog comments. :) As for why my ISP's DHCP ranges would have been added only recently to whatever blacklist was used, that will likely remain a mystery. :P I did use a different IP range to post.

The math problem was “What is the sum of 5 + 7?” on the failed post. The answer I gave being 12, of course. It follows that it was not rejected because the math was incorrect, although that is the reason claimed in the error message. It also wasn’t a transient glitch; I resubmitted it a couple of times with identical results.

In fact, the originating IP and the added P.S. were the only things to differ between the failed and the successful post. Which means that the only thing that the failed post contained that the successful one didn’t was a Bell Canada IP address. Which means that that was the criterion for rejection.

Which means that somewhere, some software is configured to reject posts based on originating IP with an error message that incorrectly claims that they got the math problem wrong. Configured, in other words, to lie.

It seems unlikely to be akismet, which apparently silently rejects posts rather than produce any error message at all.

I’d suggest you examine closely the source code for the math test plugin. It may be that it has some extra “features” besides simply testing for the answer to be correct. Likely it accessed a remote database to decide what IPs to block, since it seems you didn’t yourself tell it to start blocking Bell Canada IPs, yet its starting to do so must have been triggered by something.

The extra “features” are probably well-intentioned spam-blocking measures, but obviously have been implemented naively. Most likely, they used a blacklist intended for blocking e-mail spam that preemptively includes the ranges of large ISPs’ DHCP servers, since email originating directly from an ISP user’s computer instead of going through their ISP’s MX box is likely spam. It should be obvious why blocking such addresses is smart for email despamming but stupid for policing blog comments. :) As for why my ISP’s DHCP ranges would have been added only recently to whatever blacklist was used, that will likely remain a mystery. :P

]]>
by: Scott Carpenter http://www.movingtofreedom.org/2008/02/10/wordpress-xhtml-validator-plugin/#comment-3453 Wed, 13 Feb 2008 19:30:36 +0000 http://www.movingtofreedom.org/2008/02/10/wordpress-xhtml-validator-plugin/#comment-3453 Hi, Somebody. First, this is likely a long-standing problem with <a href="http://akismet.com/" rel="nofollow">Akismet</a> that you've previously reported. It is unfortunate and I'm sorry that it happens, but I'm not planning to discontinue using Akismet at this time. (However, your comment did make it through without my having to approve it, so either you came from a different IP range, or something has changed with Akismet.) Second, "lie?" There may be some other technical issue, but is it necessary to imply duplicity is at work? I haven't heard of any problems with the math comment plugin, and don't know why it may have rejected a correct answer. Since your comment was allowed, it must have worked at least once. So, two things are at work. Like most WordPress blogs, I'm using Akismet for spam filtering. I'm also using the math comment spam plugin. (I previously used the Bad Behavior plugin but haven't for a while now.) Many comment posting problems would be caused by one or the other of these components. I have little control over what Akismet does, and from experience, it doesn't report anything when it swallows up a comment. The math comment plugin is so simple that I would be surprised if it suddenly started failing. Independent of Akismet, it will complain when it thinks you've entered a wrong answer. It's certainly possible that it has a bug of some kind, but I assure you there is no intentional misdirection going on. Back to the main part of your comment, I agree that bad spiders are a likely culprit, and the slashdot discussion goes over other causes. Hi, Somebody.

First, this is likely a long-standing problem with Akismet that you’ve previously reported. It is unfortunate and I’m sorry that it happens, but I’m not planning to discontinue using Akismet at this time. (However, your comment did make it through without my having to approve it, so either you came from a different IP range, or something has changed with Akismet.)

Second, “lie?” There may be some other technical issue, but is it necessary to imply duplicity is at work? I haven’t heard of any problems with the math comment plugin, and don’t know why it may have rejected a correct answer. Since your comment was allowed, it must have worked at least once.

So, two things are at work. Like most WordPress blogs, I’m using Akismet for spam filtering. I’m also using the math comment spam plugin. (I previously used the Bad Behavior plugin but haven’t for a while now.) Many comment posting problems would be caused by one or the other of these components. I have little control over what Akismet does, and from experience, it doesn’t report anything when it swallows up a comment. The math comment plugin is so simple that I would be surprised if it suddenly started failing. Independent of Akismet, it will complain when it thinks you’ve entered a wrong answer. It’s certainly possible that it has a bug of some kind, but I assure you there is no intentional misdirection going on.

Back to the main part of your comment, I agree that bad spiders are a likely culprit, and the slashdot discussion goes over other causes.

]]>
by: Somebody http://www.movingtofreedom.org/2008/02/10/wordpress-xhtml-validator-plugin/#comment-3451 Wed, 13 Feb 2008 05:05:38 +0000 http://www.movingtofreedom.org/2008/02/10/wordpress-xhtml-validator-plugin/#comment-3451 My guess is that the W3C servers get so many requests for these DTDs because of poorly-written (or just lazily-written) spiders that treat everything resembling a URL in page source as a link to follow. In particular, the one company whose computers generate 300,000 requests a day each is almost certainly running a spider out of its data center. The W3C could put a robots.txt on its site denying bot access to those URLs but a) a lot of the bots hammering it probably also don't respect robots.txt; lazy is lazy; b) there may be legitimate need for bots to access those pages, including validators, and c) if it <i>did</i> work, they'd just get 300,000 requests a day for robots.txt from each of those machines and any similar ones. P.S. Why are you blocking the ability to post from IPs belonging to Bell Canada, AND making it lie to these users about why they cannot post? (It incorrectly claims that they got the spam validation wrong, even when it obviously wasn't, e.g. was "12" in response to "What is the sum of 5 + 7?".) My guess is that the W3C servers get so many requests for these DTDs because of poorly-written (or just lazily-written) spiders that treat everything resembling a URL in page source as a link to follow. In particular, the one company whose computers generate 300,000 requests a day each is almost certainly running a spider out of its data center.

The W3C could put a robots.txt on its site denying bot access to those URLs but a) a lot of the bots hammering it probably also don’t respect robots.txt; lazy is lazy; b) there may be legitimate need for bots to access those pages, including validators, and c) if it did work, they’d just get 300,000 requests a day for robots.txt from each of those machines and any similar ones.

P.S. Why are you blocking the ability to post from IPs belonging to Bell Canada, AND making it lie to these users about why they cannot post? (It incorrectly claims that they got the spam validation wrong, even when it obviously wasn’t, e.g. was “12″ in response to “What is the sum of 5 + 7?”.)

]]>