Why the first idea is a trap

For you only

The mechanism behind those results has a name: design fixation. David Jansson and Steven Smith showed it in 1991: give designers an example solution and they reproduce its features, including its flaws, and explore less of the space, mostly without noticing. On a build day, the example is the pair's first idea. Agree on a direction at 9:35 and it quietly becomes the wall every later thought bounces off. Two solo starts are the cheapest fixation-breaker there is.

And the effect is not stuck in 1991, or in banner ads. Engineers at Georgia Tech reran the comparison on physical engineering design in 2022: the parallel group again produced better results, explored more of the solution space, and reported more confidence and less anxiety. My favourite detail comes from the same research group a year earlier: people preferred the iterative approach even while the parallel approach was beating it. Expect your room to resist the split. The data says do it anyway.

One more study for your sceptic, because it predates the AI era by three decades. In 1996, Jakob Nielsen and Jan Maurits Faber had four designers independently design the same telephone interface, then measured usability. Iterating on the single best of the four scored well. Merging the strongest ideas from all four scored better, 70% above the starting designs and clearly ahead of plain iteration. Keep that in your pocket for the noon decision: a good merge can beat a good pick. The next page says what a good merge means.

You have seen this move before, by the way. Jake Knapp's design sprint makes everyone sketch alone before the group decides anything; he calls the principle working alone together. Stanford's ME310 engineering course has teams build a deliberate "dark horse" prototype alongside their main direction, precisely to fight fixation. The parallel build is the same old wisdom, repriced by vibe coding: sketches cost an hour, working prototypes used to cost a week, and now they cost a morning.

Sources: David Jansson and Steven Smith, "Design Fixation", Design Studies, 1991 · Murphy, Floresca, Fu and Linsey, Research in Engineering Design, 2022 · Jakob Nielsen and Jan Maurits Faber, "Improving System Usability Through Parallel Design", IEEE Computer, 1996 · Jake Knapp, John Zeratsky, Braden Kowitz, Sprint, 2016.