Musings

A few weeks ago I fell down a YouTube rabbit hole of old TV commercials from the 70s, 80s, and 90s…the kind I grew up with. In those days, the formula was different - Head & Shoulders warned you about the social consequences of dandruff. Slim-Fast promising you’d finally fit into those jeans. Listerine reminding you, with some urgency, that your breath was a problem. It was a little cringeworthy. But it also made me notice something. Those ads shared a common assumption: that

There's a tension at the heart of qualitative research that doesn't get talked about enough: discussion guides designed to maximize learning often do the opposite. Here's how it usually plays out. Stakeholders want answers. The pressure to get every possible insight out of time spent with respondents is real. So the discussion guide gets packed. Every angle gets covered. What started as a guide morphs into a script to make sure nothing slips through the cracks. Then you get i

An interesting experiment from the journal Patterns details an experiment scientists conducted with self-referential loops similar to those used to power AI (in this experiment: image-text-image-text...). The findings reveal (to no one's surprise) that without outside intervention "autonomous AI feedback loops naturally drift toward common attractors — very generic-looking images, which we call 'visual elevator music.'" Logically, this makes sense. When you remove the poets,