pull down to refresh
Before Feb 2023, average block size was ~1.2 MB. After ordinals and inscriptions began exploiting the SegWit witness discount, average block size rose to ~2.29 MB at its peak in March 2024.
This could use some historical context. When segwit was proposed, it was clearly communicated that segwit would increase the blocksize up to a maximum of just under 4 MB. Bitcoin Core’s 2015 Capacity Increases FAQ estimated that blocks composed of standard single-signature txs would have about 1.6 MB, and that the blocksize would likely increase to slightly larger than 2 MB with more complex transactions.
That's roughly a 75-90% increase in bytes per block at the same 10-minute interval. The blockchain grew ~90 GB in a single year during 2023-2024.
Indeed 2.2 is about 1.83×1.2, and the 7-day average blocksize peaked to above 2.2 MB twice in February and March 2024.
However, I’d like to point out that the yearly blockchain growth did not increase by 75–90%—which a reader might infer from how that paragraph was phrased—it increased by less than 50% from 2022 to 2023.
So, in the historical context, we got only in 2023 into the estimated range of what was accepted when segwit was adopted, when the blockchain grew by 53,999 blocks and 92.0 GB—or 1.70 MB per block. Since that peak, block size has been receding: in 2025 we had 53,082 blocks and 83.5 GB blockchain growth for 1.57 MB per block in average, which is around the lower end of the estimated blocksize for segwit adoption.
In other words, blocks have been smaller than expected.
Appreciate the constructive comment, thanks for the historical context and data.
The 75-90% figure was specifically about per-block size, not yearly blockchain growth. Your own calculation confirms this: You note 2.2 is about 1.83x of 1.2, which is an 83% increase in bytes per block. The 2.29 MB peak I cited from March 2024 (from mempool.space 7-day average data) puts it at ~91%. So the per-block claim holds within the stated range.
You're right that a reader could infer I meant yearly growth increased by 75-90%. That's a fair criticism of my phrasing, and I should have been more precise. Your figure of less than 50% for total yearly growth is accurate.
Where I'd push back is on "blocks have been smaller than expected." The 2015 Capacity Increases FAQ projected 1.6 MB for blocks full of standard single-signature P2PKH transactions. Before February 2023, blocks averaged ~1.1-1.2 MB, well below that projection. What brought blocks up into the 1.6-2.2 MB range was primarily non-monetary data exploiting the witness discount, not a surge in monetary transaction demand. When blocks sit at 1.57 MB in 2025, a meaningful portion of that capacity is still non-monetary. Saying blocks are "smaller than expected" is technically accurate against the 2.0 MB multisig projection, but it frames the situation as though demand for monetary transactions filled those blocks to expectations. The composition tells a different story.
The decline from 2.29 MB (2024 peak) to 1.57 MB (2025) reflects cyclical inscription demand, not the problem resolving itself. The witness discount that makes data embedding economically attractive is unchanged.
Full blocks is the natural state of the Bitcoin system, because demand for cheap highly-replicated perpetual storage is unbounded. When monetary transactions don’t fill the blocks, the underdemanded blockspace will be filled with more frivolous things.
I could see how your observations might support an argument for a temporary blocksize reduction, or to reduce the witness discount, but I can’t figure out how you get from your observations to supporting the introduction of seven consensus rules that disrupt development of more sophisticated monetary rails and only impact ~3% of the data insertion transactions observed on the network at this time.
Fair point on the SegWit block-size context.
My main question is simpler:
Was the extra SegWit capacity meant to support monetary throughput, or to subsidise permanent arbitrary data markets?
And in your view, how have SegWit, Taproot, inscriptions, Ordinals, Runes and the related changes improved Bitcoin as money?
More specifically, how have they made the protocol more decentralised, easier for ordinary people to validate, and less vulnerable to attack?
I’m asking genuinely, since I’m not a developer and I’d like to understand the argument more clearly.
Here's a predicted next block on mempool.space:
Most of the highlighted transactions are runestones or something. Correct me if I'm wrong, but all of those are BIP110 compliant and will continue to be included in blocks even after BIP 110 activates.
It seems to me that you may dislike the concept of a witness discount. If not, I don't understand how your proposal (or any proposal) will successfully limit the presence of transactions with arbitrary data in valid blocks.
the blocksize limit still seems like the best tool for controlling node-runner costs.
Appreciate the constructive comment.
You're partially right: Simple rune operations (basic mints, transfers with few edicts) encode compactly and would likely fit within BIP-110's 83-byte OP_RETURN cap. Complex operations (full etchings with name, symbol, premine, divisibility, spacers, cap, amount, height parameters, plus multi-output airdrops with many edicts) would exceed it.
But the main blockspace driver since February 2023 has not been runes. It has been inscriptions, which embed arbitrary data (images, files, executable code) directly into Taproot witness fields via OP_IF/OP_NOTIF script structures. That is where the bulk of the non-monetary blockspace consumption comes from. BIP-110 Rule 7 bans OP_IF and OP_NOTIF in Tapscript, which would effectively prevent inscription creation. Rule 2 (256-byte data push limit) constrains it further.
BIP-110's own text explicitly states: "This proposal does not aim to eliminate spam entirely" and notes that token protocols are "best countered in policy rather than consensus." It is a targeted first step at the consensus level, not a comprehensive solution to all non-monetary use. The fact that simple token transfers still fit in 83 bytes is by design, not an oversight.
On the witness discount: You're right that I have concerns about how it's being exploited. BIP-110 addresses the vectors (the script structures used to embed large data) rather than the discount itself. The BIP's rationale discusses this trade-off explicitly and opts for simplicity and speed of deployment, noting that more comprehensive changes could be pursued after the temporary period expires.
The blocksize limit and BIP-110 address different layers. The blocksize limit controls total growth. BIP-110 targets what fills that space.
I'm using bitcoin and interested in it because I want censorship resistant money (don't get hung up on this, I'm not accusing you or BIP 110 of censorship). We achieve this by having a set of block validation rules and not seeing it as our concern what someone does within that.
Now, let's accept for a moment your contention that the current block validation rules have a vulnerability in them and let's say we all run BIP 110 and correct it.
It is a targeted first step at the consensus level
Do you think it is possible that we will get to a point where there is disagreement about whether a certain kind of transaction is "monetary" or not and it won't be clear cut?
I highly doubt that all the humans of the world who want to use bitcoin will be able to agree on what counts as monetary and what does not.
It seems then that there will be some point where we must say: follow these rules and it's a valid transaction (regardless of whether a person is trying to embed data in it or not).
Personally, I think the current block validation rules are plenty good. I run a node. I'm a merchant who accepts bitcoin. I don't believe that adoption is primarily limited by the difficulty of running a node.
Because the problem isn't monetary transactions. It's non-monetary data displacing them while paying less per byte via the witness discount.
read what Scoresby said again. They're not displacing monetary transactions.
In fact, part of the reason Bitcoin is failing is that there are no monetary transactions. Everything's financialized and Saylor-ized now.
You're right that block frequency is fixed by the difficulty adjustment. But block frequency and blockchain growth rate in bytes are different things.
Before Feb 2023, average block size was ~1.2 MB. After ordinals and inscriptions began exploiting the SegWit witness discount, average block size rose to ~2.29 MB at its peak in March 2024. That's roughly a 75-90% increase in bytes per block at the same 10-minute interval. The blockchain grew ~90 GB in a single year during 2023-2024.
This is what makes it more difficult to run a node, because you need better hardware and more storage space. I've tried, it's near impossible to properly download and run a full node on an RPi4 or even a RPi5 since 2025.
It's because inscriptions stuff arbitrary data into the witness section at a 75% weight discount. That discount was designed to incentivize SegWit adoption for monetary transactions, not to subsidize data storage. Since Feb 2023, non-monetary transactions have occupied approximately 50% of available blockspace. Full documentation here: https://wtfhappenedinfeb2023.com/stats-about-spam
So the chain does grow faster in absolute terms when non-monetary data fills blocks to capacity at subsidized rates. That's the specific problem BIP-110 addresses.
As for "why not advocate a blocksize decrease:" Because the problem isn't monetary transactions. It's non-monetary data displacing them while paying less per byte via the witness discount. Restricting non-monetary data at the consensus level is a more targeted solution than shrinking blocks for everyone.