Pavel @spavel.bsky.social · 28/03/2025The friction lets the tax-preparer slow down long enough to build a mental model of the task. Not trying to make the complex simple (lying), just understandable. Even the grouping of fields creates beneficial friction through focus. These fields are on another page...because I don't need them *now* 1520
Pavel @spavel.bsky.social · 28/03/2025Analog designs get this right more often than digital designs. @theohonohan.bsky.social has a great old post about this: "The Smithsonian did some research [on] anticipated and satisfying experiences in museum visits. The most satisfying [wasn't] the most anticipated." x.com/theohonohan/...x.com 2500
Caravelle @caravellin.bsky.social · 28/03/2025Good thread! Not really visible without a a Twitter account. I can't do a screenshot ATM but here are papers @theohonohan.bsky.social linked in it: repository.si.edu/bitstream/ha... (I'm guessing for this one, couldn't see the whole link but seems to fit) reganforrest.com.au/2014/04/ipop...repository.si.edu 160
Pavel @spavel.bsky.social · 19/05/2025A cleaned-up and expanded version of this thesis, in case you need to convince some AI-pilled manager that you can't just take these things at face value.Pavel@spavel.bsky.socialIf you wanted to refer to this rant in a professional context, a cleaned up version is now up! The design of LLM chatbots doesn't only make hallucination inevitable - it makes the user less likely to spot errors and react to them. The path of least resistance is drawn thru thoughtless copy-pasting.nngroup.comAI Chatbots Discourage Error CheckingAI hallucinations threaten the usefulness of LLM-generated text in professional environments, but today’s LLMs encourage users to take outputs at face value. 1478