Add another layer to the analysis.

That same data is fed into LLMs. Look at Twitter. Very little of that is "quality" yet it was used to train Grok. The same is true for the Meta family.

I agree our first focus is to fill our databases with stuff that is useful. Sadly, Web3 databases are trash too.