> Scaling L1 is a tough-to-impossible proposition. There's just no way to store everyone's transactions on everyone's computers.
This is the comment that made me sign up for an account: Thanks for doing that, danuker ;)
In the most strictest sense: Yes, it is impossible to store all transactional data on just one computer, just like it would inevitably be impossible for one server/database to host an entire company's data.
The conventional way of doing this would be to shard the database so that it could be spread out amongst multiple servers, simple enough. This can't be done directly in a trustless system, however: Additional checks, data structures, & procedures need to be created so that it can be allowed to do so.
At a high level overview, data availability sampling [0][1] is required to make the attempt possible. When combined with Verkle trees [2][3] (Merkle trees + zk proofs) however, further improvements towards the number of possible shards could be made, increasing the storage capacity of the entire network even further.
It's not about storing everyones transactions on everyones computers. It's about distributing it to enough computers in order for transactions to be safe and fast. Algorand seems to handle this on L1 just fine.
Are you specifically answering this about Algorand or are you thinking of other platforms? Algorand seems to be sufficiently decentralized, and I wouldn't describe it as "an inefficient database", but I'm happy to be proven wrong if you have evidence of otherwise.