I recently shared a Tweet about file names.

The inspiration for this was adding a new podcast to my podcatcher. For reasons that are mostly nerdy, I use bashpodder. I run it a couple of times an hour during my waking hours and stream or copy the files to whatever device I happen to be at. It’s a setup that works pretty well for me in general.

The downside is that all of the files get dumped into a directory by date. Some podcasts (e.g. Marketplace) do a good job of naming files: I know what show it is and when it’s from just by looking at the file name. Others use the network (e.g. “GLT” for Gimlet Media) and a string of numbers without any obvious meeting. The worst offender is Art19.com, from where I get “The Greatest Generation” and Akimbo. Those shows have UUIDs as filenames.

I can understand why, on the backend, that is beneficial. The files themselves are just one part of (I assume) a database of shows. No human ever has to touch it, so you might as well name it in a way that minimizes the risk of a naming collision. But it’s extremely hostile to the user.

I suspect that most podcast listeners these days use an app and don’t directly download the files. But for those that do, sane file names are important. A friend asked about using just the date as the file name, as he apparently does for recordings from his church. That’s even worse, because it assumes that the listener saves them in a unique location.

When it comes to media that you intend for others to download, it’s vitally important to not make any assumptions how they will store it. Maybe they save everything to their Downloads folder and never move it. If two separate items were produced on the same day, one of them will potentially get overwritten. That’s probably not what you want to happen.