We need to actively discuss a permanent solution for maximum block sizes, including "default" settings, and put it into the major BCH implementations as soon as possible

Bitcoin Cash has been left vulnerable to the same attack that stopped Bitcoin Core from scaling.

We saw during the recent stress test that the "default" setting for ABC software of 2MB caused the majority of miners to leave the "default" in place.

Now we have the leadership of the ABC client talking about a cap that stays around 2x demand, without defining "demand", a critical part of any such proposal.

The definition of "demand" will determine if BCH can rapidly adjust to holiday shopping cycles, or a new killer app, or if it will have on-chain congestion every time some new use case or event causes demand to exceed supply.

Discussion about this "default" seems to come in the form of one or two-sentence group messages, which I can tell you from experience will lead to disaster, more divisions in the community, and possibly a failure of BCH to continue to get traction with merchants and users.

I believe we need a very simple, basic, scaling cap on the default blocksize that scales at fixed block heights based on predicted improvements in software efficiency, computing, storage and broadband costs.

Adaptive sizes can be proposed too, but are far more complex and more likely to break down in the future (see the DAA, and revised DAA).

Scaling ceiling limits are meant to predict what's safe for the network, while leaving miners to set soft limits on what they consider to be profitable. Adaptive blocksizes take some of the profitability decision making out of the hands of miners, and put this decision making into the protocol. Now, we've seen that miners often don't change default settings at all, so there's some merit to an adaptive blocksize to automate this decisioning.

Why don't we look at improvements in benchmarked performance of BTC > BCH since 2009 over cost reduction in computing, storage, and broadband, and implement a BIP-101 like limit for the default setting?

We can begin discussions around an adaptive blocksize now too, but I feel like that process needs more time to reach an optimal design.