The failures you don't publish are the ones the next lab pays for. Treating negative results as first-class, citable data is how methods stop being fossils.
Negative results belong in your methods because reproducibility depends on knowing what doesn't work — and because the person who learned it deserves credit. jnar stores them as structured, attributed objects attached to the protocol, not as folklore.
A negative result isn't just an experiment that failed. It's a precise, reusable piece of knowledge: this reagent at this concentration suppressed the signal; this step only works below 8 °C; this antibody lot gave nonspecific bands. Each one narrows the search space for everyone who comes after.
Published papers almost never contain these. The methods section describes the path that worked, stripped of every wrong turn — which is exactly the information a replicator needs most.
The incentives are against them. There's no journal slot, no citation, and no credit for "here's what didn't work." So the knowledge lives in lab notebooks and Slack threads, and it leaves when the postdoc does.
Labs routinely spend months rediscovering a dead end that a lab down the hall already mapped — and neither of them ever finds out.
The fix is structural. On jnar, a negative result is a typed object you attach to a protocol — with the conditions, the observation, and your ORCID. It's searchable, it's versioned alongside the method, and when it changes how the protocol is done, it shows up in the fix history.
That turns a frozen method into a living one: the protocol you cite next year is measurably more reproducible than the one published today, because the failures have been folded in.
You don't have to re-type anything. Bring an existing methods document into jnar and the structure — steps, reagents, figures — is extracted for you; then you add the failures you already know about. The first conversion is free.
