Things That Did Not End Up Working Out [/ˈθɪŋz ðət ˈdɪd ˈnɑt ˈɛnd ˈəp ˈwərkɪŋ ˈaʊt/] n - Not every good idea survives contact with the floor. Some fail because the physics is harder than the slide deck, some because the operational cost runs too high, and some because they solve a problem that does not in fact exist. I record them here for two reasons: to keep the same proposal from resurfacing every year, and to show which constraints are real and which were only assumed.

This page stands counterpart to Field Notes, which gathers the fixes that worked, and to Problems We Could Only Work Around, which gathers the constraints that outlasted every fix.

The entries

• On Moving Warm Air Instead of Cold Water [/ˈmovɪŋ ˈwarm ˈair ˈinstead ˈof ˈkold ˈwater/] n - an experiment in reusing exhaust heat for cooling that made the controls harder, not easier.
• 800V DC stayed in the lab [/ˈ800v ˈdk ˈstayɪd ˈin ˈthe ˈlab/] n - a distribution voltage that looked efficient on paper until the safety and equipment story caught up with it.

These connect to the larger reckonings in AI factory power and cooling, 800-volt DC distribution, and The New Buildout.