April 18, 2026 ~1 minute minutes read Marc

Unpublished, Unfunded, Unforgettable

Not every result makes it into a paper. The ones that don't are often the most useful. On failure as data, not verdict.

Some of my most useful “results” in molecular science never made it into a figure, let alone a paper. Looking back over 25+ years in comp chem, I realise a quiet truth: failure has been my most reliable co‑author — as long as I treated it as data, not as a verdict on my ability.

The patterns I keep seeing:

  • The MD run that “blew up” after 3 ns often exposed a bad assumption faster than any reviewer.

  • Rejected grants forced me to sharpen the real question and cut decorative aims.

  • Broken analysis scripts revealed where my workflow depended on undocumented heroics.

  • Awkward talks where someone in the back row asked “obvious” questions usually marked the point where my story wasn’t clear, even to myself.

What helps is a small shift in framing: instead of “Did this work?”, I now ask “What did this teach me about my model, my workflow, or my story?”

The win isn’t a spotless record; it’s a lab culture where people can say “this failed, here’s what we learned, here’s what we’ll change next”.

What’s one “failed” experiment, simulation, or application that quietly shaped your trajectory more than a success?