As Luke DashJr has previously suggested, reduce the blocksize to 300k thereby increasing scarcity for blockspace which would consequently drive L1 fees higher, and reduce/extend low cost operation of bitcoin nodes.
First principles thinking would have you consider both extremes. What about a 1GB block size? What about a 1KB blocksize? Probably neither of those would be acceptable, so how do we arrive at the appropriate limit? To me it seems the limit should be variable and based upon a reasonable hardware capacity such that nodes can still be run worldwide, but where transactions remain affordable for all. The current limit is too low imo and unnecessarily punishes economic activity on-chain to ensure nodes can be run on hardware that practically nobody uses.