Decrease the block size limit

[JeremyRand]

And now, for everyone's Favorite Topic (TM): the block size limit.

The AAA hardfork is planned to remove the BDB lock limit, which effectively raises the block size limit from some nondeterministic amount (somewhere between 500 kB and 1 MB) to 1 MB. I've commented in the past that this is probably not a good thing. Here are some more details on this.

• In the past month, the average block size was less than 3.4 kB.
• In the past 3 months, the highest average block size for a single day was under 33 kB (June 24).
• Luke Dashjr pointed out to me on #namecoin-dev that one of the major safety issues with larger block sizes is that they cause the IBD time to increase over time.
• This means that SegVal has a large (and asymptotic) improvement on what block size is "safe".
• Luke suggested that 256 kB is probably safe (in terms of IBD time), without SegVal.
• If we softforked to 256 kB as part of AAA, we would still be 8x lower peak usage than the limit, which would then improve further once we do SegVal.
• If we somehow end up hitting the limit, the main effect would be that transaction fees increase. This isn't bad for us, the main result of that would be that squatting would be more expensive.
• I therefore suggest that we seriously consider rolling a 256 kB block size limit decrease softfork into AAA.
• I also suggest that the softfork sunset circa 1 year after taking effect, so that we're not stuck with it forever.

--- Want to back this issue? **[Post a bounty on it!](https://www.bountysource.com/issues/46703449-decrease-the-block-size-limit?utm_campaign=plugin&utm_content=tracker%2F6562918&utm_medium=issues&utm_source=github)** We accept bounties via [Bountysource](https://www.bountysource.com/?utm_campaign=plugin&utm_content=tracker%2F6562918&utm_medium=issues&utm_source=github).

[JeremyRand]

@domob1812 Curious what your take is on this.

[domob1812]

I'm not strongly in favour of reducing the block size explicitly, but as you say, it won't have any practical impact in the near term anyway - and it can indeed compensate for removing the lock limit. So I'm fine with putting in a temporary (with a predefined timeline to raise again) block-size limit. 256 KiB sounds also good to me.

[JeremyRand]

Updated stats from BitInfoCharts:

• Average block size in last month: 6.994 kB. Around double what I reported 13 months ago.
• Highest average block size for a single day in the past 3 months: 8.073 kB (July 8). Significantly lower than what I reported 13 months ago.

As far as I can tell, all of the arguments I made 13 months ago for decreasing the size limit still apply.

[domob1812]

I don't have a strong opinion here. In general, I'm fully on the "Bitcoin Core small block" side - so as long as noone objects, I think we can decrease the block size together with AAA. Personally I'd probably just leave it as in Bitcoin, because I think that size (even if used fully) is fine to run full nodes on something like my old laptop - but if you (and/or others) feel like we should actually decrease, then we can do it.

What about going down to 100 KB (=400k block weight)?