Segwit limited the block weight to 4,000,000 weight units (WU). Blockspace has been almost fully demanded since Inscriptions started taking off around February 15th, 2023. Bitcoin Core reserves 4000 WU for the coinbase and block header, so I’ll count any block above 3,960,000 as full.

There have been 119,418 blocks since February 15th 2023. Out of those, 341 blocks were empty (below 2000 WU), 1155 blocks were non-empty up to 90% full (i.e. between 2001 and 3,600,000 WU), 292 were between 90% and 99% full (3,600,001–3,960,000 WU), and 117,630 have been full (3,960,001 or more). In the past ~27 months, 98.5% of all blocks were using at least 99% of the block weight. I have no idea what you are talking about when you say that blocks are not full.

Reply to this note

Please Login to reply.

Discussion

Mea culpa, reading my own post, two things come to mind. Until Bitcoin Core 29.0, we were reserving 8000 WU for the block overhead, it only went down to 4000 WU in 29.0., I should have been counting blocks as full starting at 3,992,001 weight units or more, i.e. 99.8% full (less than 8000 WU left). Per this metric, there were 117,100 blocks full instead of 117,630, which also let’s us deduce that there were 530 blocks that were between 99% and 99.8% full.

Wouldn’t you expect 100% full blocks in BTC?

If there was no pilot in 1-2% of all planes you wouldn’t call them full either.

I am not sure I get your point. There are occasionally some blocks that are completely empty, and sometimes miners accidentally include less transactions than are available, e.g. because they build a block with a recently restarted node.

We also have recently cleared the mempool a few times, so at those times non-full blocks would have been expected.

Ah ok. I knew about completely empty blocks. Didn’t know the other reasons. Thank you for the explanation.