Vitalik: Why I'm not worried about EIP-1559's variable block capacity?
TechFlow Selected TechFlow Selected
Vitalik: Why I'm not worried about EIP-1559's variable block capacity?
EIP-1559 does not pose higher risks to clients than the fixed gas limit mechanism.
Navigating Web3 tides with focused insightsOne criticism of EIP-1559 is that block size becomes variable, fluctuating in the range [0, 25M] instead of the fixed 12.5M gas limit, forcing clients to handle double the load. This argument is further extended: if we believe clients can handle such high loads, then they should be able to sustain that level at all times—so why not abandon EIP-1559 and simply do something more useful, like doubling the block size limit?
The core idea behind this reasoning is that the primary risk of large blocks comes from the largest blocks processed by clients, not average block size. I believe this view is incorrect (and thus EIP-1559 does not impose greater risks on clients than a fixed gas limit mechanism). Here’s why.
Recap: Why don’t we immediately raise the gas limit to 100M?
Three reasons:
1. Block processing time under normal conditions increases
This would increase from the current ~400 ms to about 3.2 seconds, leading to many negative consequences:
-
Very high uncle rate, promoting centralization
-
Only the most powerful nodes could remain synchronized; all others would struggle
-
Even the most powerful nodes would require significantly more resources
-
Longer delays when resynchronizing after brief outages (e.g., moving your laptop node from home to a café)
2. Worst-case block processing time increases due to DoS attacks, rising from today's 20–80 seconds to potentially 160–640 seconds.
3. Storage growth rate increases
This would rise from ~50 GB/month today to ~400 GB/month, resulting in:
-
Much slower synchronization
-
Much higher storage requirements
-
Slower disk access, as large databases are inherently slower to query than small ones
Note: All points under reasons 1 and 3 apply only to long-term normal usage, not peak periods. Therefore, when considering peak impacts, only reason 2 needs to be examined.
Argument 1: EIP-2929 already offsets EIP-1559’s shortcomings
EIP-2929 increased the gas cost of storage access operations, tripling the gas required for worst-case DoS attacks. This means that EIP-2929 combined with EIP-1559 actually reduces net worst-case block processing costs by a factor of 1.5 compared to the current state.
A natural follow-up question arises: "If EIP-2929 is so effective, why not just raise the gas limit to 25M or 37.5M?" The answer is simple: reason 2 is not the only barrier to increasing gas limits. Even if DoS concerns were fully resolved, the issues under reasons 1 and 3 would persist for the foreseeable future. Therefore, the additional headroom provided by EIP-2929 cannot justify a major increase in block capacity.
Argument 2: Short-lived peaks caused by brief attacks are far less harmful than prolonged attacks
Suppose an attacker floods the chain with junk data at maximum block capacity (twice the target), causing gas prices to rise by a factor of 1.125 per block. This increase is exponential: five consecutive full blocks (~65 seconds) raise gas prices 1.8-fold; after 5 minutes, 15-fold (225-fold after 10 minutes). To sustain the attack, the attacker must pay transaction fees at these skyrocketing rates. Thus, a realistic attack might last around 5 minutes.
What happens if clients receive blocks generated during this 5-minute window (each taking 20–60 seconds to process)? Clearly, chain processing slows dramatically. Many short forks emerge. In fact, forking implies that even after the attack ends, an attacker with relatively low hash power (e.g., ~20%) could still revert transactions. This is bad—but it’s far better than an attack lasting hours or even days.
Most transactions and services today already wait longer than 5 minutes for confirmation; only extremely fragile services would be disrupted by a 5-minute processing delay. In contrast, rollbacks or denial-of-service lasting hours or days—as seen in the 2016 Shanghai attack—cause severe damage.
Therefore, a 5-minute peak of 25 million gas is far less risky than a permanent 25 million gas limit.
Argument 3: Short-term peaks already occur
The inherent Poisson process of proof-of-work mining means block production has randomness in timing. In fact, randomness alone causes a doubling of chain capacity roughly once per week, with peaks lasting five minutes.
Note: This results from many same-sized blocks arriving close together, not from individual large blocks. But to my knowledge, there is no evidence or reason to believe that gas consumption per block grows super-linearly with block size.
Thus, to some extent, dealing with peaks is a known quantity—and the ecosystem has so far tolerated their impact.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
@VitalikButerin
