#corrected: Fixing your errors on Twitter

I screwed up on Twitter yesterday morning. In the grip of nerd rage over a story about an Apple patent application–and without sufficient caffeine in my body–I tweeted that the Cupertino, Calif., company had received a patent on a feature that had debuted in a third-party app some three years before its 2012 filing.

Delete tweetThe problem was, Apple had only applied for a patent on a text-while-you-walk system that would overlay message conversations on your phone camera’s view of your surroundings. Oops.

So I tweeted something, um, transparently wrong. Now what? I’ve attended more than one panel discussion on this, and the answers usually get stuck on one of two conflicting imperatives: Don’t let the error go unfixed, but don’t look like you’re hiding the mistake either.

(See my earlier post about documenting changes to your story, if necessary in comments you leave yourself.)

Since you can’t edit the incorrect tweet or even flag it as wrong in the way you could amend a flawed story or blog post, letting it stand risks perpetuating the mistake. But if you delete it, then the evidence of your error vanishes.

What I decided to do was to delete the tweet, follow up by saying what I’d gotten wrong, and then redo the original tweet with a reasonably obvious hashtag, #corrected, to indicate that it was a “CX” for an earlier version:

Does that routine work for you all? Or am I once again seriously overthinking something that people with real jobs don’t worry about at all?

In other news, earlier this afternoon I was glad to see that the Ask Patents clearinghouse for prior art will include this Apple filing in an upcoming call for submissions:


Corrections and changes can’t be clandestine

In the bad old days of paper-only journalism, you couldn’t change the text in an already-printed story, but at least newspapers almost always ran the correction in the same spot (usually, a box on A2 quietly dreaded by all in the newsroom). We’ve now flipped around the problem: It’s trivially easy to fix a story that’s already online, but you can no longer count on getting notice that it was corrected.

WordPress update buttonAnd while I’d much rather see stories get updated early and often to fix mistakes and incorporate breaking news, to do so without telling the reader you changed them is… kind of a lie. It suggests that you never made any mistakes in the piece when you really did. And since somebody will always notice the change, if not take a screengrab of the original copy, you risk trust rot setting in among readers.

Ideally, the content-management systems in use at news sites would automatically time-stamp each update and let readers browse older versions, as you can with the “View history” button on any good wiki. But some three years after online-journalism pioneer Scott Rosenberg urged just that and heralded the arrival of a WordPress plug-in to automate public revision tracking, I see few sites following that practice. More often, the bad copy goes down the memory hole.

If you run your own site, the lack of built-in version-browsing can’t stop you from telling readers you changed the copy–just strikethrough the offending text if it’s a minor fix or add a date- or time-stamped note to the end of the piece calling out the correction. (Since WordPress.com doesn’t provide a way for readers to compare revisions like what blog admins get in the editing interface, that’s what I do here.) That’s also how I handle things at the few freelance clients that allow me to sign into their CMS.

What do you do if you lack that access and a “CX” might otherwise go unremarked? Here’s my fix: Once your editor updates your post, leave a comment on it, linked back to a page or social-media account publicly recognized as you, that notes the error and the correction. Readers may not see that comment, especially if some relevance algorithm hides it by default, but at least you’ve documented the change in the closest possible spot to the original mistake.

Trajectory of an error

A week or two ago, as I was reading the corrections box on page A2 of the Post, I thought to myself that it had been a while since I’d had to run a “cx” on my own work. I credited having a saner workload… and then wondered if I was due for an error anyway.

Turns out I was. I left a simple but stupid mistake in the feature I wrote for Ars Technica about the unlikely success of indie ISP Sonic.net–listing the price of this provider’s 1 Gbps fiber-optic service as $79.99 instead of $69.95. My editor at that site unknowingly put that price in the headline and therefore, as you can see from the link above, memorialized it in the story’s address too.

Sonic’s CEO Dane Jasper spotted the mistake within a few hours of the story’s appearance and notified me in a Twitter message. I e-mailed my editor, who had it fixed minutes later… and then I could get on with my “how could you?!” follow-up. (Figuring out how an obvious error wormed its way into a story is more constructive than walking around and cursing at yourself.)

The Versions feature of Apple’s OS X Lion, as seen at right, revealed that I didn’t add the price of the service to a draft of the story until Feb. 13–weeks after I’d started my reporting. Then I typed in the wrong number and kept using it from then on.

The Evernote file with my notes from interviewing Jasper and some of his customers had never included that price. My e-mail showed that I did mention the right number, rounded up to $70, in my pitches to Discovery News (which should soon post my take on what a connection that fast feels like) and then Ars–but had subsequently written “$79.99” to one source on Jan. 25 and to another on Feb. 11.

It appears that this number lodged itself firmly in my brain and never got out.

I can’t blame any of the usual excuses, like not having time to verify things. I spent weeks on this piece and checked just about everything else–the copy I filed had a link back to the City of Sebastopol’s demographics page confirming its population as 7,397. But for the number that wound up in the headline, the most important one in the piece, I never thought to link back to the relevant press release on Sonic’s site. (Doing so would have also avoided the confusion expressed by one Ars reader: Why isn’t this service listed among Sonic’s services?)

So that’s this week’s lessons re-learned: Put the important numbers in your notes at the start instead of leaving them in your head, and link to your sources, so readers don’t have to take your word for things. Or just don’t be a flake.

(I’ve yet to see any readers call me out on this. But I’m irked anyway, which is why I just devoted almost 500 words to unpacking my mistake.)