I once spoke with the founder of Gilded (a crypto accounting firm), and he said that dealing with data from different blockchains was a big pain in the ass. I can only imagine that the "100+ L2s" future won't make it easier. BTW the author's comments about challenges coming up with scale reminded me of the article that we once had on Kiwi about the early days of Google: https://www.newyorker.com/magazine/2018/12/10/the-friendship-that-made-google-huge
Not to be a shill, this is actually exactly the problem that we're trying to solve at SQD. Sure, we're just about 12% there, but it's quite incredible how much of the data backend is badly functioning/ patchy pipelines from more than one indexing service just because none cover all the L1s, L2s, 3s. Tbf, all the L2s are definitely a pain, and modularizing further does little to aid the data access and retrieval issues - but that's why it's fun to try and build something that does fix it.