Making long talks fit a commute

邓超 — Mon, 14 Sep 2026 04:38:54 +0000

Making long talks fit a commute

A forty-minute talk is a strange artifact. It is too long to watch at a desk and too short to justify a flight, so most of them end up in a saved list that nobody opens twice.

My fix is unglamorous: I only keep audio, and I only keep the ones I would replay. Talks I download, talks I convert, talks I finally listen to while walking. The barrier was never interest — it was that a video tab demands your eyes and a commute does not offer them.

Two things changed the math for me.

First, treating audio as the default. A conversation, a lecture, a recorded interview — if it survives being audio-only, it was worth keeping in the first place. I pull the audio, name it properly, and drop it on my phone. The workflow I settled on, including the naming pattern that keeps a queue usable, is written up at YouTube MP3 Hub.

Second, cutting the length before the first listen. A long talk often has twenty good minutes in it. My rule is one pass at slightly faster speed, and anything that does not hold up gets deleted without guilt. A queue of twelve things you actually finish beats a library of three hundred you never start.

The bigger shift was realising this is a home-system problem, not a media problem. The same pattern shows up with backups, documents, and the shelf of cables I refuse to throw away: the tool is rarely the bottleneck, the decision about what to keep is. I wrote more about that framing, and about the cheap versions of these habits, at AI Home Digest.

None of it is impressive. It is just the difference between a saved list and a walk that was actually better for it.

A two-rule photo archive, four years later

邓超 — Mon, 14 Sep 2026 04:38:52 +0000

A two-rule photo archive, four years later

Most photo advice starts with software. Mine started with a dead laptop.

I had about nine years of pictures on a machine that stopped booting on a Tuesday, and the only copy that mattered was the one I had been meaning to make. Since then the rule has been simple: everything lives in two places before the week ends, and the folder names are boring enough that I never have to think about them.

Rule one: one folder per year, one subfolder per month. No themed folders, no best-of collections. The moment I start curating while filing, the filing stops happening. Month folders are also the only structure that survives a restore without a spreadsheet.

Rule two: names are for searching, not for sorting. A file called 2026-03-14-kitchen-table.jpg is findable in a decade. IMG_4821.jpg is a mystery. I rename on import and never look back.

Everything else is optional. Duplicate detection, ratings, faces — all nice, and all things I have started and abandoned at least twice. The two rules have survived four years precisely because they take ten minutes a month and require no decisions.

The part I underrated was the resizing step. Phone cameras now produce files big enough that a year of pictures is annoying to move around, so I keep a stripped-down copy of each month at a reasonable size and treat the originals as an archive I rarely open. I wrote up the exact settings I use, along with the folder layout, over at LapPhotos — and a short version of the import checklist that I keep pinned.

The other habit that stuck came from the opposite direction: using AI tools for the tedious parts instead of the creative ones. Batch renaming, generating alt text, drafting captions — all things I write about at AI Image Prompt when a workflow turns out to be worth keeping rather than just clever.

Four years in, the archive is not beautiful. It is complete, which is the only thing I actually needed.

DEV Community: 邓超

Making long talks fit a commute

Making long talks fit a commute

A two-rule photo archive, four years later

A two-rule photo archive, four years later