I’ve always treated Cites & Insights as “published”–that is, once an issue appears, it doesn’t change. I don’t correct typos or meaningful mistakes. (When the publication moved to the current domain, I revisited each PDF to change the domain name in the masthead, but made no other corrections.)
I’ll stick with that standard for actual errors–cases where I’ve left out a word or said something incorrect. Naturally, I try to do followups when needed, but it’s good to keep the published record intact. And I don’t plan to go back and fix dumb typos in past issues…
But I just replaced the PDF and two of the HTML essays for the current Cites & Insights, and it’s likely that if a similar situation arises in the first week after a new issue’s published I might do the same.
What changed? Three cases where the string “egan” within a word appeared as “Elgan” instead–one “bElgan” instead of “began,” one “elElgant” instead of “elegant,” and one other (I’ve forgotten the string). In no case could an incorrect meaning have been assumed; it just looked stupid. A reader in Australia alerted me to the problem this morning.
Clever people can probably guess what happened…and I really should know better. Here’s the whole silly story:
In an attempt to minimize typos and other errors introduced in the copyfitting process, and to give the material one last read, I now consistently print out an issue after I’ve gotten it to the desired length, let it sit for at least a day, then read the hardcopy as carefully as possible, marking any changes.
In this case, one section of the Kindle & ebooks essay included notes on a Mike Elgan column–and somehow I’d managed to alternate “Elgan” and “Egan” roughly equally throughout the notes. I wasn’t sure which it was, and did the search to verify that it’s Elgan.
Then (ahem) I did a “replace all”–and, duh, forgot to check the “Match case” box.
Actually, I think I had the section of the text highlighted–but I’d also forgotten that one step backward in Word 2007 (from Word 2000, and this may have changed earlier) is that “replace all” no longer limits itself to a highlighted region, asking before going any further. (I can’t find any way to restore that limitation. Anyone out there know of one?)
So there it is: My extra step to minimize errors worked great…except for introducing a few new ones.
It’s good to be perfct. It’s also unsusual.