Hacker News Reader

Comment by perihelions | Hacker News Reader

Another example of this was Millikan's determination of the fundamental electric charge (of one electron). The first, famous experiment had a major error; and subsequently everyone who independently replicated the experiment, for two decades, came up with the same error, in the same direction. The people whose experiments got the correct value probably self-selected themselves out of publishing it.

https://en.wikipedia.org/wiki/Oil_drop_experiment?useskin=ve... ("Millikan's experiment as an example of psychological effects in scientific methodology")

View on HN

Room-temperature superconductors have so many urgent commercial uses that if this fake, I doubt the error will persist for that long. There's too much at stake.

There's also probably a hundred labs skipping over replication completely to try and beat someone else out on a random permutation of these materials and methods and hoping to get lucky.

Yes, "replication" in this experiments usualy means foloeing the oficial recipe and many small variants.

Publishing an exact reproduction is difficult, but a new record a few degrees hotter is easy. (And even another variant with a lower temperature and more tolerance for current or magnetic field can be published with the right text.)

> but a new record a few degrees hotter is easy

Assuming one exists. Assuming it is real.

Yep. Easier to publish, assuming it exist and you can find it.

I can't see why they'd skip replication given the uncertain status of the original process. They'd try to replicate as closely as possible, confirm they have a superconducting material, then they'd branch out from their established process. Starting from a variation means they don't know whether it doesn't work because of the variation or because it doesn't work at all.

I hear some very clever labs can do more than one thing at a time.

That’s not unreasonable, but the high-risk, high-reward approach would be to let researchers elsewhere do basic verification, while you try to leap a step ahead of them by going directly to derivative variations.

Abandon the work if nobody can replicate LK-99 within a few weeks and you haven’t really lost out badly.

Boiling-water-temperature superconductors. Pinch me, it feels like a dream.

It is interesting to note that while the error was in the same direction, the magnitude of the error decreased in subsequent studies. No one expects to get exactly the same result, so there was room for variation towards the true result.

Reply on news.ycombinator.com