I feel like we have gotten off on the wrong exit. OP_RETURN is a testing code that should have never been made standard in 2014. The sole problem of OP_RETURN and Annex are data storage burden. I admit with Moore's Law, it is likely not a storage issue. My contention is that processing, bandwidth and RAM do not adhere to Moore's. Both OP_RETURN and Annex increase the bandwidth necessary for expedient download. Especially with future low end hardware. To be clear, I find BIP 110 wanting. - My approach would be to modularize the reference code into about 5-6 manageable chunks instead of the monolithic codebase it is today. With node policy, consensus, the Bitcoin spec, and peer logic, so on as separate modules independent but adhering to the same spec. - Remove OP_RETURN standardness - Remove the witness discount - Review the BIP 341 OP_codes With that, spam would simply pay for their data usage and ACTUALLY compete on level ground with simple transactions. (Instead of at a 1/4 weight discount)

You need to hash the annex to calculate the wtxid, which goes into the witness merkle root in the coinbase, which you check when validating a block though. You also hash it as part of taproot witness signature valuation, which is different from other witness data (eg inscriptions).