How Kusama's Last Validator-Set Reduction Affected Decentralization and Coretime Use, According to New 2026 Data

I like proposals that come with their own homework. When the Polkadot Forum began debating shrinking Kusama's active validator set, the response that caught my eye was not another opinion thread - it was a pair of data reports with methodology attached, posted on October 2026 under the title "Decentralization and coretime metrics for set reductions" (topic 19223). They answer the two questions every set-reduction debate actually turns on: how do you define decentralization, and how do you count the cores Kusama really needs? The short version: the last reduction did not hurt decentralization, and Kusama is using far fewer cores than it used to assign.

Key Metrics at a Glance

Why "decentralized enough" needs a number, not a feeling

The report's framing is worth stealing for any chain: decentralization is "the number of operators I have to whack with a wrench or bribe to break BFT" (the Nakamoto coefficient) combined with how power is distributed from smallest to largest operators (Gini, plus size-class bands). The author makes size bands scale-free by default - under 1%, 1-5%, and over 5% of seats - because absolute bands (1 / 2-5 / 6-25 / 26+ nodes) flatter the picture once the set shrinks: the large classes simply disappear at low set sizes.

The harder methodological problem is identifying which validators are really one operator. Most stashes have no on-chain identity, so the report layers heuristics: identity, shared payee accounts, proxy keys, and closed nominator sets backing the same validator group. You pick how aggressive to be - L0, L1, or L2 - and the on-chain coverage is high enough (90.4% at L1, 98.1% at L2) that the choice changes confidence, not direction.

What the data actually showed after the last reduction

The headline finding is quietly good news for reduction advocates: the relative composition of operator sizes was not negatively impacted by the last set reduction, and further reductions are not expected to change that. But the honest caveat matters as much as the headline: the Gini coefficient improving should be read with a pinch of salt. Seats are integers, so compressing the range of operator sizes mechanically shrinks the gap between smallest and largest, and the Gini falls with it. The largest players roughly retain their relative share of seats. Equality improved; concentration at the very top did not move.

Coretime: the unused-cores surplus has collapsed

On coretime, the report counts a producing core as one that included a block within the 24-hour window - an upper bound on real demand. From roughly November 2024 to October 2025, Kusama had a massive surplus of assigned-but-not-producing cores, about two to three times the producing count, driven by the Spammening, placeholder parachains, and coretime negotiation purchases. Since then the surplus has tightened to roughly 10-20%. Peak daily producing cores on Kusama were 47 in March 2026 with a 43.2 monthly mean; Polkadot peaked at 57 in March 2026. On-demand cores are still underused - 98% of that demand came from a single March/April 2026 flurry.

How to read these reports

The methodology is the story here: definitions chosen up front, heuristics you can toggle, coverage numbers disclosed, and an explicit admission that mistakes are the author's own. That is what makes a set-reduction debate evidence-based rather than vibes-based.

TL;DR

• What: New 2026 reports quantify Kusama's validator-set reduction: decentralization metrics and coretime usage (forum topic 19223).
• Why: The last reduction did not harm operator-size composition, and the assigned-but-not-producing coretime surplus has collapsed from 2-3x to about 10-20%.
• Caveat: Gini improvement is partly an integer-seat artifact; largest operators keep their relative share, so watch the Nakamoto coefficient and top size class, not just Gini.
• Watch: Further reductions (600 to 500 to 400 to 320) and whether peak producing cores (47 on Kusama, 57 on Polkadot in March 2026) stay within the shrunken set's capacity.

Sources

• Polkadot Forum: Decentralization and coretime metrics for set reductions, topic 19223 (2026)
• Decentralization report
• Coretime report

Gemma Nguyen is TotesTek's Content Lead & Journalist, covering cryptocurrency, Web3, DeFi, blockchain technology, and emerging tech trends.