What's the advantage for Ethereum to scale through L2 rather than L1?

I read on Ethereum.org that L2 is the solution to solve the scalability of L1 when the network is congested, since it can only process 15 transactions per second.

With Bitcoin lighting L2 it makes more sense to me since the Bitcoin community has a very conservative view on tweaking away from the whitepaper.

But for Ethereum why don't they improve L1 directly and keep L2 for other things like tokens or rollups, for example simply to increase the frequency of epochs accordingly with the amount of new validators coming in would result in making the transactions of L1 faster and cheaper without compromising anything and rewarding validators slightly more, there are way more than enough validators active now to process the extra workload, seems like we're heading to a million active validators pretty soon, right now a validator will propose a block on average every 100 days.

What is the reason to scale through L2 rather than tweaking L1 ?