Lately, people have been talking about “modularization.” Frankly, for end users, the biggest changes might boil down to just two things: the same operation—whether it runs on which execution layer or which DApp, and where the final settlement happens—is starting to become a matter of choice; and user-experience metrics like fees, latency, and failure rate are easier to break down and optimize separately, instead of trying to solve everything in one go. The upside is that you can use a cheaper, faster layer. The downside is also very real: more bridges, longer paths, and more risk points. And when something goes wrong, it’s harder to tell which layer is really to blame.

In the past couple of days, the whole hype around testnet incentives and points expectations has flared back up again, and everyone is speculating whether the mainnet will issue tokens. I was pretty interested at first, since the data can be used to estimate expectations. But after scrolling for a while, I realized there was too much noise in the information, so I ended up unfollowing for a bit to keep my mindset from being carried away. For now, I’m only keeping track of three things: the actual cost, security assumptions that can be verified, and whether I can replay and reconstruct this transaction on-chain to understand exactly what happened. As for the rest… for now, that’s it.