--- name: screen-log-5 description: Continuation of [[screen-log-4]] from 13 Sep 2026, after that file hit the 48KB cap. Same rules and columns as [[screen-log]] - one row per screen, written BEFORE the verdict, screen-time facts only, outcomes scored in the screen-log-outcomes files. Read and append on every screen. sources: [cowork] aliases: [screen-ledger-5] --- # Screen log, part 5 (from 13 Sep 2026) Rules and columns are in [[screen-log]]. Rows under 2KB. The RAILX screen and its first two passes are at the end of [[screen-log-4]]. ## Log ### 2026-09-13 ~08:35 UTC. RAILX THIRD PASS: THE DEMAND TEST. Ben corrected the "potential" answer: measuring adoption three hours after launch measures nothing, the question is product-market fit, "is this something people WILL start using". He is right and it is now a standing rule. Token 0xfEAd46F7baBd84b7f450aaEEcE081deF6e973c28. Artifact https://claude.ai/code/artifact/84d13f4f-f924-435f-aec5-9ab8817fb9de **THE MECHANISM, derived rather than assumed: adoption(t) = latent demand x awareness(t) x trust(t) / friction. At t=0 awareness is near zero and trust is zero, because the first thing anyone thinks about a new project is "will this still be here next month, or is he going to exit". So day-one adoption is near zero regardless of how badly the world wants the product, and a download count on day one is not evidence in either direction.** My earlier use of "733 SDK downloads" as the potential answer is retracted. **IS THE JOB ALREADY BEING DONE? YES, HEAVILY.** 6-hour window (blocks 61,601,803 to 61,815,368) across the nine largest RH stock tokens: NVDA 41,989 transfers · SPY 27,470 · GOOGL 23,078 · AMZN 16,026 · AAPL 14,598 · TSLA 9,480 · MSFT 6,788 · META 3,223 · COIN 1,965 = **141,617 transfer events, 50,386 unique transactions.** A 900-tx random sample resolved to **721 unique senders and 191 distinct destination contracts.** Top destinations: 0x65050a9b7e5075a2ba5ced7b1b64ee66262c40dc (Pons curve, 16.4%) · 0xccc88a9d1b4ed6b0eaba998850414b24f1c315be (7.8%) · 0x8876789976decbfcbbbe364623c63652db8c0904 (7.1%) · 0x9689992f5b5c09447f15906d8d11214944488341 (6.7%) · 0x4337084d9e255ff0702461cf8895ce9e3b5ff108 (6.6%). Uniswap Universal Router selector 0x3593564c present; ERC-4337 EntryPoint v0.7 0x0000000071727de22e5e9d8baf0edac6f37da032 present; Multicall3 0xca11bde05977b3631167028862be2a173976ca11 present. **The pain is real and at scale.** **THE CUSTOMER LIST, ENUMERATED BY NAME from DefiLlama 30d revenue rather than given as a TAM, and this is where it collapses.** GMGN $22,672,406 (Telegram bot, 10 chains) · Maestro $767,616 (9 chains) · KyberSwap Aggregator $324,289 (20 chains) · Virtuals Protocol $364,715 (AI agents, 4 chains) · Definitive $300,812 (9 chains) · Rainbow Wallet $140,264 (34 chains). **BUILD-VERSUS-BUY: every one is multichain, so each already solved routing on 9 to 34 other chains before Robinhood Chain existed; two of them ARE the routing layer and compete rather than buy; and GMGN earns more on this chain than every native token combined and will never rent its core execution path. EVERY NAMED CUSTOMER ALREADY OWNS WHAT RAILX SELLS.** **BUILDER SIGNAL vs TRADER NOISE, counted separately: builder signal = 1** (@fire_solana "will try this tech when i get home"). Everything else in `to:railxdev` is trader noise ("burn ser" x6, "chad dev", "my balls are tingling") plus two collab DMs from @iZeusCrypto (225,562 followers, marketing). **Zero technical questions, zero feature requests with a use case, zero third-party repos importing @railx/sdk. COUNTERFACTUAL: if RailX vanished tonight, nobody complains by name.** One holder spotted the real gap unprompted: @Spoiler862307 "If could make pricing for API using RAILX coin this will bring value to the coin." **THE WEDGE, and it is genuine: AI AGENTS.** The only customer category where "the incumbent already built it" is false, because nobody has built agent rails for tokenized stocks on any chain. The MCP server is the right instrument. **Virtuals Protocol is live on RH Chain at $364,715/30d and is the single named entity worth chasing; every future update should be judged on whether it moved that one relationship.** Adjacent/competing on the same idea: @DeskOnHood ("The Agentic Data Layer for Stocks & Crypto on Robinhood") and @fletch_now (writes and deploys RH Chain dApps to spec). **THE SURVIVAL GATE, which is what actually throttles adoption right now, not features.** Client is open source (github.com/railxdev/sdk) so the code survives the founder, **but the API is closed, hosted, four days old and run by one anonymous person. Bus factor 1. No status page, no incident history, no company.** A professional team refuses that dependency regardless of quality. **It is also the most fixable item on the whole screen: doxxing, a status page, or a second maintainer moves it.** - **PMF VERDICT: DEMAND EXISTS, ALREADY CAPTURED, with one open wedge.** Not "no demand" and not "unserved". The commonest and most flattering trap: the market is real, at scale, and the addressable slice is empty. - **Written into `benchmarking` as question 2 (THE DEMAND TEST, tests 8a to 8h, four-label verdict) plus blocking gates 7, 8 and 9; plugin ben-research-stack v3.3.0.** /preferences.md was rewritten and condensed in the same turn because it hit its 16KB injection cap. - OUTCOME: pending, score in [[screen-log-outcomes-5]] #### 2026-09-13 ~08:50 UTC. RAILX FOURTH PASS: THE TRUST ARCHITECTURE. Ben's question: "couldn't he deploy his tool in a way that is independent of him, verifiable on chain, so people trust it because of the way it's deployed, and he revokes access?" **Answer: yes for the money, no for the service, and those are two different questions with different answers. Only one of them gates real customers.** **QUESTION 1, "will he rug the token": FULLY SOLVABLE ON CHAIN, TODAY, CHEAPLY.** A buyback run by hand from an EOA is a promise; the same buyback routed into an ownerless contract anyone can trigger is code. A poll on the 27.4M bag followed by a manual transfer is still a promise; sending it to the burn address, a timelocked vault with a public unlock, or a merkle distributor settles the moment the tx lands. **AND HE WOULD NOT HAVE TO BUILD ANY OF IT. Two permissionless fee-automation layers are already live on Robinhood Chain and this is the reusable finding:** - **PonsVault V2, ponsvault.com, VAULT 0xFdae23CE76018Da62507bB5ef20e6ef5450e8312.** Verified live 13 Sep 2026. Its literal product copy: "Creator fees land in your pairing asset. The contract spends them on buybacks, staking, or a stock dividend, **every time, without an operator**... Anyone can press run, or our keeper does it for you." Three templates shipping: **Staking** (holders stake, earn creator fees in the pairing asset pro rata), **Stake & Burn** (stake to earn, half the fees buy the token back and burn it), **RWA Dividend** (fees buy a tokenized stock, holders claim by balance, no staking). Lottery "soon". Pairs against ETH, USDG and 20+ stock tokens. **Caveat: it attaches AT LAUNCH ("we deploy the token through pons v2, create the vault, and route fees to it"), so retrofitting an already-launched token is the open question.** - **Sinjoh, sinjoh.com, INJOH 0x2cC0FAC44B8252f6B10208B091aFf2c94B4da77D.** Five live protocols, the generic retrofit option: **Fee Router** (split creator fees, swap into tokens and RWAs, add to LP, then send, airdrop or burn), **Liquidity Manager**, **trustless Merkle Airdrop Distributor** (any account creates and funds its own airdrop, any ERC-20), **Raffle Rewards** (hourly draws settled by ECVRF, winners paid without claiming), **Treasury Vaults** (immutable, single governor, timelocked handoffs). Integrates Pons, Flap, Pools, LetsCash. - **PONS FEE PLUMBING, read on chain 13 Sep 2026:** the fee-claim escrow `0xd3AFEB2a57f70eF218Aa82451c51B2fb0416Ac9e` (codesize 1,932) exposes `claim()` 0x4e71d92d plus 09ad4dd9 / 1698755f / 32f289cf / 379607f5 / 70a08231 / d5d44d80 / f59e38b7, and has **NO owner(), NO transferOwnership, NO setFeeRecipient and NO setRecipient**. So the creator address is fixed at launch and the fee destination cannot be reassigned. The launcher `0xA84D64A2128f7B5eb24e619b2D25beC240F1EF3e` exposes none of those either; only the factory `0x7eD598BcEf8bd9Edd8C97A195C6d13f40801EC7e` has owner/transferOwnership. **Consequence: a Pons creator cannot hand the fee stream to a contract after the fact. He can only claim to his EOA and then forward, so the residual trust is the forwarding step, and that residual is publicly measurable every hour by comparing escrow accrual against what reaches the burner.** **QUESTION 2, "will the service still exist in six months": NOT SOLVABLE ON CHAIN, because an off-chain API is off-chain compute.** Only three things move it: **open-source the server so it can be self-hosted** (an open client with a closed server leaves the dependency exactly where it was, which is RailX's current position), **attested execution in a TEE** so integrators verify the running code rather than trust it, and **put the registry on chain** as an immutable or timelocked contract mapping each ticker to its real token and multiplier, so integrators read the chain and never depend on his server for the one thing that matters. - **THE TRAP, and it is the reusable lesson: fixing the wrong half.** Making the money trustless is cheap, visible and popular, and it matters least, because no integrator ever feared he would rug the buyback. They fear the endpoint 404s in March. **The on-chain registry plus an open-sourced server is what turns "anon with a closed API" into a dependency a real team can accept, and it is worth more to the price than any burn.** A real name and a company still beat all of it for an enterprise buyer, because their risk is contractual rather than technical. - **Written into `benchmarking` as 8i "SEPARATE THE TWO TRUST QUESTIONS", which requires every screen to name the specific deployment that would remove the founder from the trust equation and say which of the two questions it answers; plugin ben-research-stack v3.4.0.** - **PLUGIN VALIDATION TRAP, cost one rejected install: a SKILL.md frontmatter `description` is capped at 1024 characters.** The benchmarking description hit 1,100 after two feature additions and the plugin was refused with "field 'description' in SKILL.md must be at most 1024 characters". Now 973. **Check `len(description)` before packaging any skill edit.** ### 2026-09-13 ~11:20 UTC. BLAST / "experiment blast", Robinhood Chain. Token `0x724a163f0081fA1771590Dda9b55eDDe2c5B823a`, pool `0x4c25338827fb1227ee9d021320f8c282d613ecf8180cbfdf3e27e5a764c978ff`. Site experimentblast.com + launch.experimentblast.com. Screened at $2.72M mcap, $172K liq, $4.72M 24h vol, 5.9h old, 13% off a $2.97M peak, base $41K. Ben's buy line was $890K (missed). LENS: MEME. Four-pass live format (kill screen / settled / counting / judgement) at Ben's request. **VERDICT: buy.** Falsifiers: (1) blast size on /api/blasts falls under ~$100 for 30 min = engine dead; (2) X account stays suspended and no replacement builds an audience; (3) a copycat with the same mechanic takes the name. **MECHANISM, and it is the cleanest value accrual seen on this chain.** Every 180s an engine claims 100% of Pons creator fees in ETH, market-buys BLAST in the pool, and burns what it bought. Three txs per cycle (claim/buy/burn), all published. **This PASSES the quote-asset-direction test in [[rh-value-accrual]] that killed 64 of 136 RH "burns": the ETH moves INTO the pool.** Not fee collection, not inventory transfer. - 118 blasts in 5.9h = one per 2.97 min. 7 of 118 failed (6%). - **49.83 ETH = $123,263 spent buying back. 223,258,472 tokens burned. totalSupply 776,741,528 of 1,000,000,000 = 22.33% gone in under 6 hours.** Verified by direct `totalSupply()` RPC call, not the site. - **Rate ~18% of cap PER DAY.** The best name on the 6 Sep burn board (PAIR) was 364% annualised = ~1%/day. **BLAST is ~18x the top of that board.** - HONEST COUNTER: the engine is a ~3% tax on turnover recycled, not new money. $4.72M vol x ~3% = $141K, which reconciles with the $123K spent. It cannot outrun its own volume. **DEPLOYER / ENGINE WALLET `0x6f6ef2631f5103793f8cb085032dd4df54ff7377`.** 4 buys, ZERO sells across 28,985 chart points. **Holds 0 BLAST and 0.00198 ETH.** The 0.18 ETH launch dev-buy went into the burn, not a bag. `totalOpsEth = 0` on the pad: he has not taken even his own 3% ops cut. UNRESOLVED: prior launches by this wallet, because Robinhood Blockscout was Cloudflare-walled from the container all session and Dune returned "would exceed your configured datapoint limit per billing cycle". **CONTRACT:** bytecode has burn(uint256) and burnFrom, and NO mint, NO owner (8da5cb5b reverts), NO pause, NO blacklist, NO setFee, NO setMaxWallet. GoPlus returns near-empty (`is_open_source: 0`) for this token, so the selector scan was the source. **SHIPPING: a second product in 83 minutes.** Token genesis 1789277072; BLASTPAD (launch.experimentblast.com) genesis 1789282085. Working launchpad with wallet connect, logo upload, 4-step flow. At screen time: **39 coins, 464 pad blasts, 12.41 ETH spent, and 0.963 ETH ($2,380) of other coins' fees had bought and burned 1,237,894 BLAST.** Split is 9000/700/300 bps (coin / BLAST / ops). **⚠ THE X ACCOUNT @experimentblast IS SUSPENDED.** twitterapi.io user/info returns `{"unavailable":true,"unavailableReason":"Suspended"}` and from:experimentblast returns 0. **It is at least the third suspension** (@formemeonly007 "When 3rd X account suspend?"; @icewizardcz, @0xloveavax both asking why it keeps happening). Question 7a (is the dev reactive in replies) was therefore UNGRADEABLE. A meme with no megaphone is the single biggest unpriced risk here. **LORE IS THE WEAK HALF.** No reference at all: the name IS the mechanic ("every 3 min a new BLAST into the chart"). Instantly legible, which is worth a lot, but nothing to own. **$BLAST already belongs to the Blast L2 token `0xb1a5700fA2358173Fe465e6eA4Ff52E36e88E2ad` ($41M)**, so every ticker search drags in the wrong coin. **IMPOSTOR CHECK: five copycat BLAST contracts already live on RH Chain within hours**, all nameless and socials-free: `0x454c3D1E686b3c2ab2DC3DDb0E853116deD7D2C4` ($50K/$46K liq), `0x260371e92215895128E3Ece4Af42839B07080cDd` ($51K), `0x78460e1bD4EEA1a50373aD8492E9684692696946` ($45K), `0xFAD6b1b6d2De2572FdB713c155d2F1Dc88a73a48` ($25K), `0xf0CBbD4aD041416d65718b4fE574A061Fc4E0a8E` ($23K). The real one is 50x the largest fake and is the only one carrying the website/socials on DexScreener. **COUNTS at screen time.** 6,467 unique wallets traded in 6h, arrivals per 30 min 584/292/938/1864/539/298/509/527/236/143/362/175 (peak past, but turned back UP in the last hour). Last 30 min 433 buys / 365 sells, net +3.71 ETH. Every 30-min bucket net positive except two. No distribution shape. **COVERAGE: ZERO tier-one.** blknoiz06, theunipcs, pingucharts, CrashiusClay69, 0xkioto, metaversejoji, dontbuytops all 0 posts. Callers present: @CookerFlips 141k (CA only, "no idea where this goes"), @chinapumprocket 111k, @RH_Gems 53k, @meliboi_sama 12k, @SixteeN_E16 8.7k. Median shiller 310 followers, 11 of 20 under 500. **MOAT:** product moat weak, rebuild ~2 days for the engine + ~2 for the frontend. Token moat = the pad's 7% toll on other coins' fees, real but only $2,380 so far. Meme moat bad: the name is unowned and already cloned 5x. **The actual differentiator is presentation, not mechanism: a live countdown clock that makes the buyback a watchable event every 3 minutes.** **REUSABLE:** the project's OWN API is the best data source on it and needs no bridge — `experimentblast.com/api/{token,blasts,trades,chart}` and `launch.experimentblast.com/api/stats`. `/api/chart?limit=100000` returns every trade with wallet, side and ETH size, which gives unique-wallet arrival curves and buy/sell netting without Blockscout or Dune. Check any such project for its own endpoints before fighting Cloudflare. - OUTCOME: pending, score in [[screen-log-outcomes-5]] ### 2026-09-13 ~13:15 UTC. B / 「B案」, Robinhood Chain. Token `0x19C8FcEE15af82c583516a9c0E062e51F200B5b2`, pool `0x21cd61959007b1bb45e36cb556472a46fad111118ab29aa7df5817691fc7175e` (QQQ-paired, Uniswap v4). Screened 18 min after launch: base $39K, screened $243K -> $357K during the screen, liq $36K -> $48K, 24h vol $563K. Ben's buy line $75K (missed, 4.3x above it). LENS: MEME. Four-pass live format. **VERDICT: hold.** Not avoid. The single thing that flips it to buy: one real account (50k+ followers, not a caller bot) posting a B案 thread that names THIS contract, which resolves the contract split and gives the ticker an owner. **THE LORE IS THE STRONGEST SEEN ON ANY SCREEN SO FAR, AND IT IS JAPANESE, NOT CHINESE.** 「B案」= "Proposal B" / "Option B" (Japanese, "B-an"). On 8 Sep 2026 JR East announced the Suica Penguin (Japan's transit-card mascot since 2001) graduates 31 Mar 2027, and put three successor candidates to a public vote: **A案 = squirrel, B案 = rabbit, C案 = mole. Winner announced 18 Nov 2026** (Suica's anniversary). B案 the rabbit is the runaway fan favourite. - **JR East's own announcement post @JREast_News = 221,583,401 views, 134,775 likes.** @livedoornews summary 41.6M views. @ntv_tetsudobu 17.1M. - Organic fan output at 5 days old and still climbing: @pika870_wtam 6.2M views, @crazy_tanoshi 4.26M, @eiija64825 3.68M, @matsuikentaro1 3.61M, @open10opan 1.36M (hand-sewn plushie, second one), @SaltPotato207 1.03M. Fan art, plushies, "B案と添い寝したい". - **CHAIN FIT IS WEAK, unlike the 宋江/Robinhood worked example.** A Japanese train-card rabbit has no relationship to Robinhood Chain. It is a hitchhiker. - **LEGIBILITY IS HALF-BROKEN.** Instant in Japan, zero in Western crypto. Crypto X is already misreading it as "Plan B" (the Bitcoin guy) — @onrecoverymode posting "People who say it is not Plan B... EVERYONE is referring this as Plan B". A ticker that needs a correction thread is a ticker that travels badly. - No prior B案 coin on any chain before today. First time, but everyone launched at once. **THE STRUCTURAL KILLER: NOBODY OWNS THE REFERENCE, AND THE SPLIT IS ALREADY VISIBLE ON HIS OWN CHAIN.** 12+ B案 contracts across 3 chains inside 20 minutes. At screen time: | chain | contract | mcap | liq | 24h vol | |---|---|---|---|---| | robinhood (rival) | `0xb955218d260F9d90ae00eF62A4f674f9c253F57c` | $292,961 | **$145,506** | $198,970 | | robinhood (Ben's) | `0x19C8FcEE15af82c583516a9c0E062e51F200B5b2` | $292,615 | $43,550 | **$563,206** | | solana | `7Ggb8sFqmdadFhjvqYFVL2kJgCqsBZnunhDTq7B9D6sK` | $42,416 | $37,932 | **$712,607** | Also BSC `0x72B73736Cd236d62714bE40156E9e3719De87777`, `0xa7380802C4BA23e1D4d5ad7abBEa627096aC7777`, `0x0f9eA801b3Dff094b9389834c2ceCEe616668888` and a third RH one `0x969FE3D9E5717a0f72a084db3622B6810822A5b5`. - **Two coins, same chain, same name, same $293M-cap-scale, neither official. That IS the moat failing in real time.** - **The Solana copy already does more daily volume than the Robinhood one.** If Western crypto adopts this meme it adopts it on Solana. A Japan meme on RH Chain is betting on the wrong venue. **IMPOSTOR CHECK RESULT, and it inverts the RAILX pattern.** Ben's CA was FIRST (pair 12:59:55 UTC vs rival 13:02:17, 2m22s apart) and owns 8 of 8 X contract mentions — the rival has **ZERO** X mentions. But the rival carries 3.4x the liquidity, **so a DexScreener ticker search sorted by liquidity still returns the wrong one first.** Same failure mode as RAILX, opposite cause: there the fake was richer, here the second-mover is richer. **Standing lesson: on a swarm-launch meme, liquidity rank and attention rank routinely disagree, and attention rank is the one that matters.** **DEPLOYER: clean, thin, anonymous.** Pons V2 `launchAndBuy` at 12:57:21 UTC, launcher contract `0xe33E9E479dF8802cb0866d5d05258bEc4cF62948` (PonsV2LaunchAndBuy). Human EOA **`0x4cd6FC583fb63E251D58c4188AF32f108bA786C7`**, no prior launches on this chain, holds one other token (777 CRUSH `0xb04337e84DE579522E708670aF2aF8407352b4ce`). Holds **7,682,307 B = 0.77% of supply**, has not sold. Not bundled, not a self-snipe farm. Note the Blockscout `creator_address_hash` `0x3711ceA4feaDE896C913C68F01Eda97Cb06D1A42` is the CREATE2 factory with 0 txs, not the human — **read the creation tx's `from`, not the creator field, on Pons launches.** **HOLDERS: steep and clean.** 14 (launch) -> 195 (+4m) -> 856 -> 915 -> 971 -> 968, i.e. +115 in ~90s. Top two slots are contracts (PonsV2LaunchLocker `0x267444D099b10fB5Ed7c3Cc7B7c767AdcA574952` 8.16%, PoolManager `0x8366a39CC670B4001A1121B8F6A443A643e40951` 8.04%). Largest human 3.16%, then 2.52 / 2.30 / 2.23. **No uniform slices, no coordinated farm.** 5m 544 buys / 516 sells (bid alive); 1h 3,081 / 3,645 with price +561%, which is flippers cashing into a rising bid, not distribution. **NO OPERATOR AT ALL.** DexScreener `info` block is empty: no website, no X, no telegram. No project account exists anywhere. Q3/Q4/Q6/Q7/Q8 are all structurally ungradeable — nothing shipped, nothing claimed, no buybacks, no burns, no fee share, and nobody to ask. GoPlus returns `{}` for chain 4663 on this token. **COVERAGE IS 100% BOT.** 31 posts / 20 unique authors sampled and **not one real trader with a real audience.** All caller bots and scanner feeds: @bitecong (32.5k, "AI Signal RBH"), @robhousesignal, @xbtscout, @Dexradaralerts, @VantageAlpha_1, @shardsca1ls, @RHSol_scanner, @Solfighters01. The 221M-view Japanese story has not crossed into crypto X yet — that is simultaneously the bull case (unexploited) and the reason there is no bid to trust. **TWO DATED RISKS NO OTHER SCREEN HAS HAD:** - **JR East owns the character.** It was drawn by a commissioned young creator for a live corporate campaign. Real IP-takedown exposure if this gets large enough for Japanese media to notice. - **18 Nov 2026 is a hard binary expiry.** B案 wins the vote = catalyst. B案 loses = the meme's reason to exist dies on a known date. **First screen with a scheduled narrative kill-date; price this asymmetry explicitly on any future vote/contest meme.** **NET BUY TO RUN:** L = $45,800, so ~$50K of net buying takes it 10x. Mechanically cheap; cost was never the question, contract selection was. - OUTCOME: pending, score in [[screen-log-outcomes-5]] ### 2026-09-13 ~14:00 UTC. KAI / Kairence, Base + Robinhood Chain, SAME address `0xca18A528Ea897040f715edC92e6e4572780c5ca1`. LENS: TECH. Artifact https://claude.ai/code/artifact/0708abb5-45c6-417d-b19c-6956488651ab **VERDICT: buy, small.** Single flip: one outside operator putting >$50K/day through the pad. **WHAT IT IS.** Sovereign-agent launchpad. Launch a token and it gets a brain, a computer and a bank account only it controls. The novel bit: 50% of supply is paired against **kDIEM**, a 1:1 wrapper on Venice's DIEM, and the DIEM behind the pool is **staked at Venice (1 staked DIEM = $1/day of inference forever)**, so buying the token buys the agent more thinking. Agent then rents an Akash box (~$3/mo) with its own key. Base = brain, RH Chain = markets/stock pairs. **SCREEN-TIME NUMBERS.** Price $0.000847, mcap $800K, liquidity $489,956 across 5 pools (Base KAI/kDIEM $245K, KAI/USDC $63K, KAI/NVDAc $40K; RH KAI/SPCX $83K, KAI/USDG $58K), 24h vol $265K = 33% turnover. Holders 941 (Base 867 + RH 74). Age 31d (Base 13 Aug), RH side 9d (4 Sep). ATH $0.002407 on 9 Sep = **$2.27M, now -65%**, but base was $229K on 2 Sep so this is a HIGHER LOW, and it printed +76% on the screen day. Net buy for 10x = $530K; for 50x = $1.49M. **CROSS-CHAIN SUPPLY TRAP CHECKED AND IT IS NOT THERE — record this, it nearly went the other way.** Base totalSupply 943,899,310 and RH totalSupply 948,134,179 look like 1.89B summed. They are NOT additive: the bridge is **LayerZero lock-and-release**, so Base custody `0x0dddbb3c012b1418c777bd327061d653b22cf0f9` holds 157.2M and RH custody `0x213943730d7B2E9F88331623B12C710eC44fEb21` holds 790.9M, and 157.2 + 790.9 = 948.1 = one supply. **DexScreener's $800K is correct.** Standing rule refinement to the RIPE trap in [[data-access]]: sum totalSupply across chains, THEN subtract bridge-custody balances on every chain before concluding a double-count. Both proxies share the same implementation `0xD068090363f8652Cc1D3Ec59bEaF063E2D2b4e06`. **CONTRACT: cleanest read of any screen so far.** Verified `AgentToken` on BOTH explorers, solc 0.8.28. Single `_mint` in the constructor, the only mint in the contract's life, so supply is monotone non-increasing (their own invariant AT-INV-2). `burnFrom` **reverts by design** (BurnFromDisabled). No pause, no blacklist, no owner-mint, no fee-on-transfer. Only extra surface is `Humanship` (human / transferHumanship / acceptHumanship). Deployed by an **AgentFactory contract** `0xF5C5d11cBd93Cf777DD5310425C7a713F95d0EC7`, not a hot wallet. Base creator EOA `0x0147b7e34a26c4d7f444e548a35f97b437c157ff` holds **1.095%** (10.3M ≈ $8.7K) and has not sold. Largest EOA holder 2.91%. **THE WEAK LEG, AND IT IS THE WHOLE BEAR CASE. PMF: DEMAND EXISTS, ALREADY CAPTURED.** In 31 days the pad has produced **six agents besides KAI**, combined 24h volume **$1,698**: BASETARD `0xdAB6f10B4F8d21E13bF9EA69eEDe45E66C4Ffca1` $31K/$427 · RAWR `0x9716Fd34Dc5eD2e74F830a8fC07ac49D5D6beca1` $31K/$455 · WORM `0xa79fF60Ace2792de9a0782b2D8AE9874905e7ca1` $28K/$229 · BRAIN `0x8E7517bd2476fC74D78026A8Bae5269bf0779ca1` $22K/$242 · MEOW `0xaA026B1a53Fb4ef6094D13D4faE38BDF2E026ca1` $20K/$248 · WOOF `0xd193604529Ac73f252C8c9e1b8BbC45db260Dca1` $48K/$97. **Pad revenue to KAI = 25% of a 1% fee on $1,698 ≈ $4/day.** Virtuals ($405M) already owns "launch an agent token". Nobody complains by name if it vanishes. Genuine unserved wedge: **LP-as-inference-budget**, which nobody else has built. **REUSABLE METHOD — HOW TO COUNT A LAUNCHPAD'S REAL CUSTOMERS IN ONE CALL.** Find the pad's mandatory quote asset (here kDIEM `0xf8B22f75b7Ee248fF723650f43C98B253e7dfb60`), then `api.dexscreener.com/latest/dex/search?q=` returns **every token ever launched on that pad with its cap and volume**. Beats scraping the pad's own JS-rendered registry (kairence.ai/tokens is a Next.js shell; /api/* all return the SPA, not JSON). Use this on every launchpad screen from now on. **OPERATOR: @kex_pet, anon, 976 followers, X account open since 2009, 4 months on the project. GRADE B+ against audience size.** Reply latency under 2 min in public to hard questions ("who is behind this project?" 13:18 -> "I'm the dev" 13:19). **Ten dated releases in 18 days**: three-market pools 26 Aug, backing/burn 27 Aug, Akash self-lease 31 Aug, auto-staking 1 Sep, memes.kairence.ai 2 Sep, RH chain + bridge 3 Sep, buyback rework 3 Sep, sovereignty spec 8 Sep, on-chain encrypted agent chat 9 Sep, Arc announced 13 Sep. All four product URLs open and work (/launch is a 42KB app, not a landing page). No price targets, no paid callers, no selling. **Two real negatives: no GitHub at all (searched the API, zero repos and zero users), and his own 11 Sep post "humans write specs, AI writes the code" — on a protocol holding bridge custody.** **PUMP ECONOMICS: understands it, thin in dollars.** Burn KAI -> mint soulbound xKAI -> 50% of the agent's own-token fee stream forever. **56M KAI (5.6% of supply) burned, verified by totalSupply delta from 1B.** But he **removed the buyback burn on 3 Sep** — buybacks now sit in a reserve. Treasury he cited 3 Sep: ~$9,000. **THREE FRAMES.** F1 fundamental anchor **$90K–$150K** (0.1–0.2x): $4/day pad revenue at a generous 50x AI-infra multiple plus ~$9–15K treasury; a floor reference on a 31-day-old pad, not fair value. F2 live-product floor **$700K–$1.2M** (0.9–1.5x): verified contracts, working product, 941 holders, shipping dev — **spot sits inside this band, so you pay for the product and nothing for the pad.** F3 narrative ceiling **$5M–$15M** (6–19x): second-tier agent pads trade $10–40M (Clanker $11.7M, Griffain $13.6M) against Virtuals $405M; the 9 Sep $2.27M print on $1.24M volume shows the tape moves. **COMPS.** VIRTUAL $612M/$3.89M liq · PONS $376M/$6.54M · DIEM (the upstream) $64.3M/$8.12M · CLANKER $11.7M/$1.39M · PAIR $5.36M/$275K · **KAI $800K/$490K/$265K vol** · NEST $125K/$33K. KAI has the highest volume-to-cap ratio on the board (33%) and liquidity worth 61% of its own cap. **CHAIN CONTEXT.** RH TVL $934M (+81% 30d, +437% 60d), DEX vol 24h $1.50B / 7d $13.1B (+15.7% w/w) / 30d $33.9B, fees 30d $276M, 87 protocols. Base DEX vol 24h $489M / 7d $6.76B / 30d $27.5B, 175 protocols. KAI fee base as share of chain volume: 0.0023% RH, 0.047% Base. **SLOT: attention rank unranked** (976-follower dev, one 472K mention from @AltCryptoGems 24 Aug); **structural rank #1 of 1** on welding LP depth to a staked inference budget. **TOLL BOOTH FAILS FOR NOW** — take rate is real (5% of every quote leg on every agent) but <$1/day; chain 10x with share 3x = $30/day. Toll booth built, no cars. No first-mover premium priced. **MOAT.** Product moat thin and rests on Venice's DIEM continuing to exist. Token moat weak (<$1/day of capture). **Rebuild 14–21 days** for one competent solo dev: v4 single-sided pools, a 1:1 wrapper, an Akash lease script, a LayerZero token. **IMPOSTOR CHECK: clean, with the now-familiar liquidity-rank trap.** A DexScreener KAI ticker search sorted by liquidity returns **KAIKO `0x1E12d405031ce390A0f3bAd7CA59422C67B98524` first, $1,270,324 liq and $0 volume** — a different project with a similar name. A dead copy "KAI" `0x7a208A4BD166F01Bf2E556dBcf724ECD5793bebC` ($12,740 liq, $26 vol, 2 Sep) also exists on RH. The real one is identifiable because it uses the **same address on both chains**, which is itself the cleanest tell. **KILL RISKS.** Bridge custody (157M Base + 791M RH) with no audit; AI-written protocol code holding money; total dependency on Venice DIEM paying $1/day; pad stays empty. - OUTCOME: pending, score in [[screen-log-outcomes-5]] ### 2026-09-13 ~14:00 UTC. 中国制造 / "Made in China", Robinhood Chain. Token `0xE339A891D14FafD37AF44A3C9F227856268A1620`, pool `0x8999bac12892b3ab6734645b2d043d8325a8f32c3f9a6b8696df4f0d4fc83ef1` (USDG, Uniswap v4). LENS: MEME. Four-pass live format. Screened 3.9h after launch: base $40K, peak $433K (10.9x, 11:07 UTC), screened $51,382 = -88% from peak, liq $18,421, 24h vol $1,063,070, 812 holders. Ben's buy line $130K. **VERDICT: avoid.** Single flip: @Trader4apes reappears (X community / any reply) AND holders turn from 812 back up through 900. **LORE IS EXCELLENT, CHAIN FIT IS NOT.** 中国制造 = "Made in China", correct standard phrase, barcode logo, on-chain name field literally `Made in China`. **Best legibility of any screen so far: needs zero explanation in either language.** @TI_Hugh 13:02 "Is there anyone in the world who doesn't understand 'Made in China'?" **But no Robinhood connection at all** — contrast 宋江 (Chinese Robin Hood on Robinhood Chain), the exact-fit benchmark. Dev's own link attempt is a joke: "Pons is made in china and we launched on pons". **Chain-agnostic meme hitchhiking a rotation.** **DEPLOYER `0x87ae15bcbf080df92bda15bf6884c9db9c023808`** (Pons V2, `PonsV2LaunchDeployer` factory `0x3711ceA4feaDE896C913C68F01Eda97Cb06D1A42`, deploy 10:05:10 UTC, nonce 17). **Self-sniped 57,292,641 = 5.73% inside the deploy tx `0x0ae1f0611174f2117a1f903233ddd281c75369a48fdce40cdc42acc1313f46e0` for 0.0935 ETH (~$232), then dumped 100% of it 216 blocks = ~22 SECONDS later** (tx `0x53590ffbaeaab7523e219ea1253607a57c4a6f1a091b828c5f4d47bf33b4ab79`) for 0.1704 ETH (~$422). Profit ~$190. **Holds 0 tokens, 0 USDG, 0.000198 ETH (~$0.49).** Two-sided: no belief, but also zero dev overhang. **⚠ CORRECTION MADE MID-SCREEN, AND IT IS THE REUSABLE TRAP. `0x267444d099b10fB5Ed7c3Cc7B7c767AdcA574952` is the `PonsV2LaunchLocker`, NOT a whale.** I first reported it as "a migration sniper holding 8.16% who has never sold" because it is a contract, nonce 1, bought at the exact graduation second and never sold. **On Robinhood Chain ordinary user wallets are ALSO contracts with nonce 1, so "is it a contract" is worthless as a sniper test here.** The same 8.16% locker slot appears in the B案 row above. **Standing rule: on any Pons launch, hard-exclude `0x267444d099b10fB5Ed7c3Cc7B7c767AdcA574952` (locker) and `0x8366a39CC670B4001A1121B8F6A443A643e40951` (v4 PoolManager) before computing concentration.** Corrected: top10 27.06% -> **20.34%**, top20 38.53% -> **31.29%**, largest single holder 8.16% -> **2.82%** (`0x9d5eabfb22df4b2d155f32c6d38b01d30c558657`). Holder book is clean, no uniform slices. **HOLDER VELOCITY IS THE KILL, and the slope is monotone.** 559 (10:10 migration) -> 963 (10:40) -> 1,209 (11:00) -> **peak 1,232 (11:10)** -> 1,075 (11:30) -> 1,096 (12:00) -> 925 (12:30) -> 883 (13:00) -> 832 (13:30) -> **812 (13:50). -34% from peak, every bucket since 12:00 lower than the one before.** **BUY/SELL TRAP FIRED EXACTLY AS SPECIFIED.** Headline 24h 7,079 buys / 6,127 sells looks bullish. Last 30 min at the pool: **322 buy txs moved 148.1M tokens OUT, 156 sell txs moved 169.7M tokens IN, net -21.6M into the pool, with price 1h -47.4%.** Twice as many buyers, sellers still winning = small buyers absorbing few large sellers. **15-min buy volume 905M tokens (10:00) -> 276M (11:15) -> 90M (13:30) -> 44M (13:45), a 95% collapse.** **STRUCTURE: no base.** Every 15-min candle since 11:15 makes a lower high and a lower close, $444K -> $44.8K. Only bullish item is one higher low at 13:45 (41,164 -> 44,807, closed green), 15 min old. **Heaviest volume of the day, $516,122, traded between $139K and $444K mcap (10:45-11:30 buckets), so Ben's $130K buy line sits directly under the whole wall of trapped bags.** **THE DIFFERENTIATION ANSWER IS THE REAL VERDICT: the Chinese meta on RH Chain is running hot and this coin already lost the slot.** 奶蛙 (Milk Frog) `0xBC18D0B1d039107fC74ea97C333dA6323b87723D` **$679,684 mcap / $67,098 liq / $4,363,743 vol / +1,649% / 7.2h** is the leader. 宋江 `0x7ffCBA5424dF045e7a4334D80b011FE975A35E2D` $146,361 / $35,264 / $1,907,190 / +183% is second and has the exact chain fit. 豹富 `0xdAd4D9552E3761FD42CbBEC21752911fa48d0c3F` dead at -93.7%. **中国制造 is third place in a race that already has a winner. The money did not leave the meta, it rotated to 奶蛙.** **IMPOSTOR CHECK: Ben has the real one, comfortably.** 5 tokens share the ticker on RH Chain: the copies are `0xcce7651eb9eC62ceD80D0B62720d6aBfB5F563A0` ($6,260 liq), `0x15CCF41cF2f432570a5D0A24044e04f1dd85FA4e` ($6,230), `0xf6EaF5bfeF2aA9E2E61C1E0672A7554E43D1d0F1` ($4,460), `0xf46d1860c07C3B770308C9b8c0C2444B5E0EeDfE` ($4,165). 8+ more on Solana, all under $63 liq. **@wirebotRH refused a buy for @kellyonwire at 12:51 UTC specifically because 5 tokens share the symbol — the bot got it right.** **MOAT: none, and this is what makes it unfixable.** Token moat zero (nobody owns "Made in China", a sixth copy costs $200 and 90 seconds on Pons). Rebuild 4 minutes. **On a meme the moat is the reference and the reference needs a landlord; the landlord here is a 77-follower account that has said nothing for 2h45m.** **OPERATOR @Trader4apes ("Bezerk"), 77 followers, account opened 30 Jul 2026, 101 posts, blue check. GRADE D.** Dated shipping list: 09:58 the meme post (115 likes, **11,174 views on 77 followers** — strong organic reach, the one genuinely good operator datapoint), 10:16 "dex paid" (the only money he spent), 10:46 + 11:10 the pons jokes, 10:54 one reply to another dev. **Silent since 11:10 through a -88% drawdown.** Holders publicly begging and ignored: @raycrypt0 "Make an X community?", @DenisWaldeaneb "X comm? Don't sleep bro", @VibesSushi 12:22 "Dont let this coin die theres alot of holders and the charts dying". @RH_Gems (53k) and @2i2jj (116k) both asked him to DM, no public response. **His timeline before this launch is genuinely good original meme-scouting (VULPES/Grok lore, pump.fun trademark filings, Simón the blessed dog) — he is a meme hunter, not a builder, and he traded his own launch like one.** **PUMP ECONOMICS: $0.00, priced.** Burns 0 tokens (balanceOf 0xdead = 0, balanceOf 0x0 = 0). Buybacks $0 (he has $0.49 of ETH). Supply unchanged at 1,000,000,000. Never engaged a single burn/fee-share question. **CONTRACT IS CLEAN, read from runtime bytecode because it is the only thing that mattered on the safety side:** burn(uint256) present; **NO mint, NO owner (8da5cb5b reverts), NO pause, NO blacklist, NO setFee, NO setMaxWallet, NO transferOwnership, NO renounceOwnership.** GoPlus confirms not honeypot, not mintable, no taxes, no proxy, open source. **Note GoPlus `creator_address` returned the Pons factory `0x3711ceA4...`, not the human — same trap as the B案 row, always read the creation tx's `from`.** Token also self-stores its metadata: selector `53cd512a` returns the socials struct (the x.com/Trader4apes status URL), `7165485d` returns the curve `0xbc42404a778e3730d56cfd902a6a6f606ba5fa7a`, `536dac9b` returns the Pons factory `0x7ed598bcef8bd9edd8c97a195c6d13f40801ec7e`. **NET BUY TO RUN:** L = $18,421, so **10x needs ~$19,916** and 50x ~$55,916. Mechanically trivial against $1.06M of daily volume. **Cost was never the question; the question is that nothing is bidding.** **COVERAGE: ZERO tier-one** (blknoiz06, theunipcs, pingucharts, CrashiusClay69, 0xkioto, metaversejoji, dontbuytops all 0). Shillers are caller bots claiming exits after the fact (@sharkapescall 38.8k "21.7x from our call", @Raghavak0nSol 136.9k "13x", @Shdow_Sol 3.2k "17.2X from my call", @KarolCordeiro_ posting a "WENT PARABOLIC" alert for $109K -> $106.5K). **Median shiller under 500 followers.** **REUSABLE METHOD: full holder ledger from RPC alone, no Blockscout, no Dune.** Robinhood Blockscout was Cloudflare-walled from the container all session again. Built the entire book with `eth_getLogs` on the Transfer topic in 10,000-block chunks over the token's life, then netted balances. **Chain runs ~9.87 blocks/sec, so a 4-hour-old token is ~142,000 blocks and 32,944 logs — about 15 calls.** Reconciled exactly to 1,000,000,000. This also gives the mint event (single mint, full supply to the curve), the dev's entry and exit with timestamps, holder count at any minute, and 15-min pool-flow netting. **Use this as the default on RH Chain instead of fighting the explorer.** - OUTCOME: pending, score in [[screen-log-outcomes-5]] ### 2026-09-13 ~14:15 UTC. YOINK / yoink.capital, Robinhood Chain. Token `0xF82baa01dA3f80C547F5826d47BcBb521A6848A5`, main pool `0x7aad4a20e73c6dce1dff13086e1f283572ef47af8bbddae6b491d9075641cacc` (ETH, Uniswap v4). LENS: TECH. Four-pass live format. Ben's buy line $175K (missed, 4x below spot). **VERDICT: hold.** Single flip: holders breaking back above ~2,600 and climbing. Second watch item: whether a real burn tx lands at 15:00 UTC. **WHAT IT IS.** A terminal notes app over an Obsidian-compatible markdown folder. A note can carry `claim` / `settles` / `reads` / `test` / `confidence` frontmatter, and on the settle date the tool reads the named source (file:, csv:, chain: over JSON-RPC, or manual:), writes the outcome back into the note, and scores the user with a Brier score plus a "said X%, right Y%" overconfidence gap. Also grades "dead notes" (nothing links in, never reopened). **THE PRODUCT IS REAL AND I VERIFIED IT BY RUNNING IT, which is the correct Q4 method for a CLI and should be the default from now on.** `git clone` + `pip install -e .` + `yoink demo` + `yoink score` in the container: it built a 35-note vault and printed the scoreboard (5 settled, said 75% / right 40%, +35 pts, Brier 0.388) with the five graded claims listed. **`python -m unittest discover -s tests -t .` → "Ran 201 tests in 0.456s, OK".** The README's "201 tests, zero dependencies" is exactly true. Repo: 13 source files ~124KB Python (4-pane TUI 21KB, cli 27KB, claim engine 11KB, hash-chained ledger, chain reader), 10 test files ~69KB, `.github/workflows/test.yml`, MIT. **⚠ TOOL TRAP THAT NEARLY PRODUCED A FALSE "EMPTY REPO" CALL — NEW AND IMPORTANT.** From this container, `api.github.com/repos/{o}/{r}` returns `{"message":"GitHub access to this repository is not enabled for this session..."}` and `github.com/*` HTML returns **403**, and `WebFetch` on the repo page returned **"This repository is empty, 0 stars"** — a completely fabricated-looking but actually sandbox-generated answer. I was one step from printing "the GitHub is empty" as a claim mismatch. **What actually works: (1) `raw.githubusercontent.com/{o}/{r}/main/` is NOT blocked and returns real files with real sizes, so use it to prove existence; (2) `git clone https://github.com/...` from the container WORKS; (3) the Chrome bridge loading `api.github.com/...` in a tab returns the real JSON (stars, forks, tree, contributors).** Standing rule: never report a GitHub repo as empty/missing on the strength of WebFetch or the GitHub API from this container. Clone it or raw-probe it. **GITHUB, real numbers via the bridge:** created 2026-09-11T10:54:31Z, pushed 2026-09-13T13:50:09Z (10 min after the token), **343 stars, 21 forks, 1 open issue, 1 contributor, 2 commits**, Python, MIT, 9.3MB. **The repo earned 343 stars two days BEFORE any token existed** — the strongest single fact on the screen. Weakness: one anon, two commits, no public build history. **DEPLOYER `0xf434aaa5dae43eb0761da12723e11c71b1349604`.** First ever tx 13:34 UTC today, funded 0.041 ETH from `0x2e16bbcd25...`. No prior launches. Received **5,770,084.33 YOINK = 0.577%** creator allocation inside the mint tx at 13:38, and sent 100% of it to `0x000000000000000000000000000000000000dEaD` at 13:46. Verified by direct `balanceOf` RPC on 0xdead (Blockscout truncates the destination to `0x0000000000` in tokentx output, which reads as the zero address — **balanceOf(0x0) = 0, balanceOf(0xdead) = 5,770,084.33**, so always confirm the burn destination by RPC, not by the truncated explorer column). Dev holds zero on this wallet. **🟡 COORDINATED LAUNCH CLUSTER, 8.3%.** Four wallets took 83.1M tokens (22.18M + 21.21M + 20.30M + 19.45M) inside 38 seconds of the mint, before the pool opened, all routed through `0x65050a9b7e5075a2ba5ced7b1b64ee66262c40dc` (Pons curve). `0x0a2fde1694e0e56df4fcc84d131068d442b522bf`, `0x35c1df83005a050a0006a37bf2e671534010ca4a`, `0x3cca253c62a326271deeaed5cec8410efaaab9b1`, `0x117f629dfa1f795802fb2dfd79689c0ad367f587`. Three checked: all first-seen 12–13 Sep, funded **0.0753 / 0.0758 / 0.0741 ETH from three DIFFERENT funders**, each spending **0.0683 / 0.0687 / 0.0665 ETH** in the same second. **Near-identical funding and spend sizes from distinct funders is the sybil tell; different funders is the disguise, not the disproof.** **COUNTS.** Migration 13:41:12 UTC, base $51K. Holders 1,917 (14:06) → 2,429 (14:12) → 2,491 (14:13:45) → **2,486 (14:15), i.e. the arrival curve flattened and ticked down at ~2,490 inside 35 minutes.** Mcap $885K (14:12) → $835K (14:13) → **$749K (14:15)**, so it made a new high well above Ben's watcher's stale "$450K, 23% off peak" reading and then rolled 15%. Main pool liq $80K; **27 pools, total liq $190,436, total 24h vol $1,483,699 = ~7.5x the pool in an hour (churn).** Main pool 5m 764 buys / 748 sells; all-pool 1h 7,896 / 7,110. **CLAIMS vs CHAIN: all true except one sloppy one.** 300+ stars ✓ (343). 201 tests ✓ (ran them). Dev supply destroyed ✓ (on chain). 🟡 **"Every 3 hours the supply gets smaller" — his posted clock reads `00:00 → 03:00 → 06:00 → 09:00 → 12:00 → 21:00 UTC`, which skips 15:00 and 18:00. Zero recurring burns had fired at screen time.** **PUMP ECONOMICS: $0 of real buyback.** The only burn is the pre-launch dev allocation, worth ~$1,000–$1,700 at the burn-time price and ~$4,800 at the $835K cap. He engages the topic instantly and unprompted but has spent nothing. **OPERATOR @defileo, 52,099 followers, account since May 2023, CA in bio. Reactive-fast, substance-shallow.** 16 of the last 20 posts are replies, **latency 1–3 minutes**, but the answers are "Moon", "Yoink it", "Let's gooooo fren", "Yesssssss"; a hack question got "No one hacked." and nothing more. **Shipping is the good half and it is a roadmap-on-time pattern, not a feedback pattern:** repo 11 Sep 10:54 → first post 11 Sep 13:54 → "51 stars in 3 hours" 11 Sep 23:23 → site/product live 13 Sep 13:08 → token 13:38 → dev burn 13:46 → repo push 13:50 → burn clock 13:55 → "Roadmap v2 in 3 hours" 14:11. **Product shipped two days before the token. Nothing new shipped post-launch.** **DEMAND TEST → DEMAND EXISTS UNSERVED, small pond.** (a) job done worse today by spreadsheets and Fatebook; (b) named customer segment is the CT caller crowd he explicitly targeted ("everyone on this timeline has a hit rate and nobody knows theirs") — real wedge, few thousand people; (c) substitution is a spreadsheet, low friction already; (d) **build-versus-buy is the weak leg: an Obsidian community plugin covers 80% in 3–5 days**; (e) builder signal ≈ 0 (1 open issue, zero technical questions on X, all mentions are 100x callers and whale bots; 21 forks is the only trace); (f) counterfactual: nobody complains if it vanishes; (g) wedge = public hit-rate for callers; (h) **SURVIVAL GATE PASSES OUTRIGHT, uniquely — MIT, zero dependencies, fully local, no hosted API, no key, no server, so the tool outlives the founder by construction.** Contrast RAILX, where the closed hosted API was the gate that failed. **MOAT.** Product moat thin, rebuild **2–3 weeks** properly or **3–5 days** as an Obsidian plugin. **Token moat is ZERO and the README says so in writing: "the program does not touch it... Nothing here works differently if you hold the token or if you never do."** Not a fee, not a gate, not a payment. A tip jar with a chart. **New reusable note: a project that documents its own token as functionally irrelevant has capped its Frame 1 at zero by choice, and no amount of product quality fixes that.** **IMPOSTOR CHECK: Ben has the real one.** Copies on RH Chain, all with no website and no socials on DexScreener: `0x6268808aa25b47f0aFFEee0E6139ab1CC9460d1D` ($130K/$36K liq), `0xa7f2a31bf6115c6515959E42bF09c692E5386dAF` ($167K/$29K), `0xda63410b61C80Dd29C151Dd7067fBc219Bba48A5` ($25K/$29K), `0x9A04405B2fa217305C7d93AC3c423Ee544BD2bd4` ($6.6K). Different older project `0xA2718f80f1FE0CdeC69c9023ee006807dD487a8c` (yoinkai.fun, x.com/yoinkfun). Also Solana and BSC fakes minted the same hour. The real one is the only one carrying yoink.capital + x.com/defileo. **LORE: weak, and that is fine because the lens is TECH.** "Yoink" = Simpsons-era slang for snatching something in one grab. Reference used correctly (one keypress and the page is yours). **No Robinhood Chain connection at all**, nothing to own, nobody can be its landlord. The name sells the product, not the coin. **COMPS at screen time.** PONS $378.9M/$6.58M liq/$2.63M vol · ROUTE $3.88M/$193K/$657K · SLVR $1.76M/$152K/$164K · **YOINK $749K/$190K/$1.48M** · PAIREX $697K/$69K/$1.14M · ACRE $4.8K/$6.7K/$0.52. **Real-product RH tokens settle $1.7M–$3.9M, so comps leave 2–5x of headroom; YOINK already out-volumes all of them, which is froth not depth.** - OUTCOME: pending, score in [[screen-log-outcomes-5]]