Had a good time going through the entries of Reddit Scaling Bake off Challenge. Brought several projects building L2 on ETH to the forth. List below (it contains some L1 compatible with EVM also)

Though I don't have the technical prowess to compare and contrast each participant based on their GitHub, some interesting observations from end user perspective.
1/ L2 will definitely scale $ETH and reduce gas fee in a huge way. All participants insist this.

2/ Projects with token have more support (comments in reddit) and funds rather than the projects without one. Probably bag holders of the tokens.

3/ @StarkWareLtd proposal was one of the most complete imo. Cost Analysis, Explainer Video and most importantly popular wallet support are the strong points. Only thing missing was an actual demo site.

4/ Though not a layer 2, @NEARProtocol proposal also look great. Native support to EVM and the demo site looks cool (With just a single shard). User can actually test all the actions requested by Reddit.

5/ @ConnextNetwork have a unique proposal too. In fact they are fundamentally different from other scaling solutions and also are compatible with several other L2 (Matic, Optimism, xDai). Tried spacefold a little too.

6/ @solana (again a EVM compatible layer 1) had a proposal too but there was no demo but explainer of their existing break Solana (which I tried already).

7/ There were several others from Matic, OMG, Fuel labs, Aztec which and I'm still reading and will continue in part 2.

8/ Problem 1: L1-L2 transport is going to be an obstacle from UX perspective. As an end-user if I'm using an non-custodial solution, I value decentralisation and would want to do things on-chain. L2 resonates like a custodial solution unless the some kind of STP could be there.

9/Problem 2: L2 and EVM compatible L1 are no different. EVM compatible L1 chains are better funded and probably might be able to get things done faster and also give a non-custodial feeling.

10/Problem 3: Applications cannot reconfigure themselves for L2 and hence they will dominate the future decisions on L2. if L2 (& native L1) projects should figure out a way to be compatible with the existing applications, then we might see the breakthrough.