Each number comes from the NEAR BigQuery public dataset. Check it with no account: node thesis/verify.mjs, which confirms sampled receipts against NEAR's public archival RPC.
Pre-registered test: before querying November, we committed a prediction that the share would stay at or above 2.0% (PREDICTION.md, commit 9d19680). Nov 1–13 came in at 4.44% (11,556 function-call keys, 248,736 full-access). Query: sql/keys_nov_to_date.sql.
Through October the share stayed under 3%; up about 9× from August to October. Query: sql/keys_by_month_mar_oct.sql.
| Month | Full-access | Function-call | Share | Targets |
|---|---|---|---|---|
| 2026-03 | 10,661,054 | 855 | 0.01% | 132 |
| 2026-04 | 2,487,775 | 823 | 0.03% | 67 |
| 2026-05 | 643,668 | 902 | 0.14% | 144 |
| 2026-06 | 517,723 | 597 | 0.12% | 96 |
| 2026-07 | 482,472 | 423 | 0.09% | 74 |
| 2026-08 | 451,570 | 816 | 0.18% | 165 |
| 2026-09 | 330,893 | 5,963 | 1.77% | 1,348 |
| 2026-10 | 266,033 | 7,363 | 2.69% | 833 |
From August to October, transactions rose 7% while function-call keys rose 9×. Query: sql/activity_by_month.sql.
| Month | Transactions | Active signers | Function-call keys per 1M tx |
|---|---|---|---|
| 2026-03 | 107,857,335 | 381,465 | 7.9 |
| 2026-04 | 28,398,312 | 335,794 | 29.0 |
| 2026-05 | 24,893,046 | 359,151 | 36.2 |
| 2026-06 | 24,415,046 | 302,563 | 24.5 |
| 2026-07 | 22,874,925 | 252,587 | 18.5 |
| 2026-08 | 21,131,379 | 215,609 | 38.6 |
| 2026-09 | 24,004,533 | 302,272 | 248.4 |
| 2026-10 | 22,627,935 | 826,700 | 325.4 |
Function-call keys are a few percent of new keys, but they sign 13.1% of transactions. 134 of those 200 use a key for intents.near, from 1 signer; in all, 29 accounts sign this way against 384 with full-access keys. Source: node thesis/txshare.mjs, 300 random blocks from 2026-10-01 to 2026-11-13 (seed 1001), 1,521 transactions, each signing key read on chain.
| Measure (share of sampled transactions) | Value |
|---|---|
| Signed with a function-call key | 13.1% (95% interval 11.2%–15.6%) |
| Share of NEAR burnt by those transactions | 11.2% (95% interval 7.4%–16.7%) |
| Signed with a function-call key for Intents or Chain Signatures | 10.2% (95% interval 8.6%–12.2%) |
Intervals resample whole blocks (2,000 draws, fixed seed), because one relayer can fill a block. 0 of 1,521 transactions had no readable key and are left out.
| Function-call key for | Transactions | Signers | Example |
|---|---|---|---|
| intents.near | 134 | 1 | tx |
| v2.ref-finance.near | 26 | 4 | tx |
| v1.signer | 21 | 11 | tx |
| nearlytrade.near | 4 | 1 | tx |
| wallet.kaiching | 3 | 1 | tx |
| v2_1.omni.hot.tg | 2 | 1 | tx |
| aurora | 2 | 2 | tx |
| 0x86b14e5375d01286f8ff85fec5c3f114e46eed19 | 1 | 1 | tx |
| 0x6b0bf85de9a4838387179288f87f475eb3154f07 | 1 | 1 | tx |
| contract.saifu-network.near | 1 | 1 | tx |
Fees burnt offset about 1% of new issuance, so supply still grows about 2.47% a year. Burn would have to grow about 92 times to cancel issuance, so agent-safe keys are not yet a supply effect. This is the status of a thesis, not investment advice.
| Measure | Value | Source |
|---|---|---|
| Total supply, 2026-10-01 → 2026-11-12 | 1,307,742,039 → 1,311,523,824 NEAR | block headers |
| Net supply growth, annualized | 2.47% | block headers |
| Issuance cap (max_inflation_rate) | 2.5% | data/protocol_config.json |
| NEAR burnt, annualized | 354,044 NEAR (0.027% of supply) | chunk headers of sampled blocks |
| NEAR burnt, annualized, only blocks after the 30% contract gas rebate ended (block 223,012,337, 2026-11-04 11:46 UTC; 61 sampled blocks) | 315,587 NEAR, about 1.0% of issuance | data/protocol_switch.json |
| Share of that burn from function-call-key transactions | 11.2% | transaction outcomes |
STRENGTHENING, NOT CONFIRMED. The rules come from THESIS.md ("What would change our mind") and are applied by verdict() in lib.mjs. This is the status of the thesis, not a buy or sell call.
| Rule | Fires? |
|---|---|
| Function-call share of new keys above 10% for 3 months running | no |
| Function-call share of new keys back under 0.5% after the hackathon | no |
| Pre-registered: November share at or above 2.0% | yes |
Three predictions were fixed in PREREG-2.md before any block of the window existed. Its sha256 b1eb3999362fb691a47706974dc841d027f2df091ed452ae1d86f1f68b4197a0 is timestamped in Bitcoin with OpenTimestamps (proof PREREG-2.md.ots; check it with node thesis/check-anchor.mjs). Window 2026-11-14T06:00:00Z to 2026-11-14T18:00:00Z: 200 random blocks (seed 1114), 1,278 transactions, 268 signed by function-call keys. Data: data/txshare_prereg2.json. Scored by node thesis/score-prereg2.mjs.
| Prediction | Bar | Result | |
|---|---|---|---|
| 1. Share of transactions signed by function-call keys | at least 11% | 21.0% (95% interval 18.4%–24.2%) | holds |
| 2. Share signed by function-call keys for Intents or Chain Signatures | at least 8% | 18.1% (95% interval 15.9%–21.6%) | holds |
| 3. Most common function-call key target | intents.near, strictly first | intents.near 187 of 268; next v1.signer 44 | holds |
Result, from the table PREREG-2.md fixed in advance: "All three hold". "Rare to add, common to use" repeats on new blocks: function-call keys are load-bearing infrastructure, led by Intents.
Stated in advance: the window was a Saturday and the hackathon's deadline day, and this tests whether the finding repeats, not whether it grows. The frozen command ran from 2026-11-14 23:09 UTC for 66 minutes (3,292 retried RPC requests). THESIS.md explains why that was after the scheduled time.
About 1 key in 9 points at an account with no contract deployed, while agent infrastructure holds under 2%. Source: node thesis/classify.mjs → data/targets.json (on-chain state, 2026-11-11).
| Target kind (from on-chain facts) | Keys | Share |
|---|---|---|
| No contract deployed (calls fail) | 1,048 | 11.3% |
| Agent infrastructure (Intents, Chain Signatures) | 164 | 1.8% |
| Token, NFT or app contracts | 8,079 | 87.0% |
Mostly a game, staking apps, meme sites and social logins, not agents. Query: sql/fc_receivers_sep_oct.sql.
| Contract | Keys | Accounts | Check it yourself |
|---|---|---|---|
| aa-harvest-moon.near | 3,433 | 3,392 | r1 r2 r3 |
| meta-pool.near | 816 | 538 | r1 r2 r3 |
| nearlios.swiftarena6534.near | 758 | 584 | r1 r2 r3 |
| linear-protocol.near | 319 | 197 | r1 r2 r3 |
| world.pumpopoly.near | 312 | 91 | r1 r2 r3 |
| nearlytrade.near | 277 | 166 | r1 r2 r3 |
| meme-cooking.near | 266 | 225 | r1 r2 r3 |
| cookie.nekotoken.near | 252 | 129 | r1 r2 r3 |
| ftv2.nekotoken.near | 243 | 143 | r1 r2 r3 |
| social.near | 198 | 102 | r1 r2 r3 |
| 0fb5bdc4d3ee4230.trade.unrlzd.near | 197 | 1 | r1 r2 r3 |
| contract.saifu-network.near | 174 | 5 | r1 r2 r3 |
| social | 153 | 121 | r1 r2 r3 |
| de8a9ccc3e8cbb1b.trade.unrlzd.near | 113 | 1 | r1 r2 r3 |
| meteor-account-verifier.near | 106 | 89 | r1 r2 r3 |
| chipfi.near | 99 | 70 | r1 r2 r3 |
| intents.near | 96 | 32 | r1 r2 r3 |
| 0f600f4b9723b12d.trade.unrlzd.near | 86 | 1 | r1 r2 r3 |
| protocol.newgoat6636.near | 85 | 71 | r1 r2 r3 |
| 177d0dba946d26e4.trade.unrlzd.near | 85 | 1 | r1 r2 r3 |
This button asks NEAR's public RPC, from your browser with no account, for each of the 120 sampled receipts. It confirms that each one added the function-call key we counted, then checks whether that key still exists today. It takes one to two minutes: the free public RPC allows only a few requests a second.
Built by node thesis/rebuild.mjs. See THESIS.md for the evidence against and what would change our mind.