Because all of these replications aren't contributing much; they're bare bones efforts with little to no scientific insight. Not one of these papers is conclusive in terms of showing evidence for it being a superconductor. None of them even contribute meaningfully.

This happens in machine learning all the time. Low quality papers rush in after every major release and announcement in order to be first. But in the long term they're meaningless because it takes time to do a good job.

Good labs don't want to announce half done maybe results. They want to announce conclusive comprehensive high quality results they can stand behind. That's what moves science forward.

Plenty of labs are working on lk-99, but they won't publish this sort of half assed analysis.