At TidBITS, we tend to talk for a long, long time about site changes before we do the work. This can be agonizing. Why can't we just do X already, whatever X is? But there's a big upside: we often have the whole conceptual framework in place, and it's just a matter of a little - okay, sometime a lot of - script programming and database manipulation to bring our ideas to fruition.
That's how we created the TidBITS Commenting System, which enables anyone to append comments to one of our articles. Adam Engst wrote about how we designed the system and how it works in "" (3 July 2009).
Ten months in, we're rather pleased about how well the TidBITS Commenting System has worked. I chose this point to take a look at comments because of a debate that's been broiling about anonymous comments and their place on news sites.
Most notably, we require that each commenter verify himself or herself for each browser used by following a URL in an email that's sent. That email links to a page that sets a browser cookie. (In the future, we'll have full-fledged accounts. We're testing the Take Control account management system now, which will eventually be extended to TidBITS readers; see "," 7 April 2010.)
This is in line with a general direction by media sites to require some sort of external verification, instead of allowing commenters to post without any outside-the-system check. The New York Times, as the idea of purely anonymous commenters might be on its way out. Facebook Connect and Twitter verification have been part of that trend, too, enabling people to use a single identity to comment across multiple sites. And  that instituted a stricter commenting system caused a quick drop in comments but ended up generating more and better comments over the long run.
Some sites, of course, simply don't allow readers to post comments at all. Our dear colleague, John Gruber, has never allowed comments on, prompting a short-lived joke site called "Daring Fireball with Comments" that revealed the true horror of what could have been. (I won't link to the site, because it's now just a mock-up of an iPhone with a note to John thanking him for taking the joke well - and a thousand junk comments promoting exciting drugs.)
For our part, we wondered if asking people to enter and verify an email address would be too high a bar, as low as that might seem. We don't benefit financially from additional page views of posts with comments (perhaps a few dollars per month at best), so we neither wanted to pump up comments to inflate page views artificially, nor to discourage readers.
The results have exceeded our expectations. We're nearing 4,000 comments made across 440 articles by nearly 1,900 people - about half the articles and links we've published in that period. And while we certainly have our regulars, we also see that many people are willing to register in order to leave a single comment.
Some of our more popular articles have had dozens of comments, and have provoked terrific discussion that extended the articles in directions we couldn't have imagined. For instance, Adam's "" (18 August 2009), generated 91 comments, many extremely thoughtful. But we're happy if an article attracts only a comment or two with useful information.
We are also totally cool with being corrected when appropriate. If we misspell a word, use clunky language, cite a fact incorrectly, or rely on tortured logic, we hear about it in the comments, and we often update the article to address the problem. (Thanks for keeping the tone polite and helpful in such criticisms!)
The response to the 1,000-character limit (about 150 words) hasn't been bad. Most people stay within the limit, and only occasionally does someone need to post a second comment to finish their thought. We're not Twitter, but neither do we wish to encourage epics.
As administrators, we gave ourselves a few simple commands to edit or delete comments, and to ban users. A banned user could, of course, register again with a new email address, but we've had to ban only 30 people, or about 1.5 percent of our commenters. Most were for egregious spam, and just a couple for over-the-top inappropriate behavior. We delete comments only occasionally, and mostly when they're no longer relevant after an update to an article (no one needs to read about a spelling mistake that was fixed), and we use an extremely light hand when editing, fixing typos if we can, or removing excess quoted material.
I've been running mailing lists and forums since 1990, when I had a desktop-publishing list with a staggering 1,000 worldwide subscribers (mostly in academia, but also through Internet gateways such as The WELL). In that time, I've never designed nor participated in any discussion or commenting system that required so little handholding.
Sure, we know that TidBITS readers are the best - you really are - but we also expected to get griefers and trolls who like to find systems to exploit. The email verification loop isn't a big deal for people who want to make other people unhappy, but apparently that, plus some AJAX elements we built for posting, have prevented any real abuse. So far. (We also have virtual tripwires set to alert us to attacks, so hopefully we'll be able to shut down any that do occur.)
We're happy with what we built, although we're still thinking about what else we can add, beyond the near-future addition of identifying messages from staff members. Once we have our account management system in place, a variety of possibilities present themselves, and we're pondering what to do first.
So what would you like to see? As you know, your comments are welcome.