Bespoke open databases would be cheaper and easier to analyse

Submission statement: The author argues for decentralising databases in science while maintain consistent file formats.

I am not sure about this. In genetic epidemiology we have three databases for genome wide association summary data called “sumstats”. Each has it’s own way of formatting the data and are in various states of maintenance. GWAS Atlas is no longer receiving many of the latest summary statistics. While MRC’s IEU database stores their fairly up to date sumstats as a very different file format (custom VCF) which is fairly simple to convert to a more standard format, but is confusing for the less tech savy users.

This arguably a pretty centralised system already. But it is already very difficult to be sure you have the most recent (and best quality) sumstats for a particular phenotype. Decentralisation would make this much worse! On the other hand, centralisation risks the singular database become unmaintained due to funding constraints and because the whole thing is likely managed by a single post-doc who needs to move on from their job every 3 years so that they have a chance for career progression.