S15 E42: Super Testnet & Shinobi on Bitcoin Projects That Suck

If you asked me the day before whether I would sit Shinobi next to a live mic on Bitcoin Takeover, I would have said no. The man builds a reputation on calling people idiots on Twitter, and I assumed the polite outcome was a walk-off. Instead, on Pizza Day inside Paralelni Polis in Prague (the "parallel society" hackerspace, not a police station, which is what my mouth kept trying to call it), Super Testnet showed up with a list and Shinobi showed up with a grin. The format we agreed on was simple and cruel: Bitcoin projects that are overhyped, underbuilt, or just kind of suck.

Super Testnet is the builder who keeps shipping weird Lightning toys, BitVM experiments, and privacy side quests (he already sat on my couch once for S15 E6 with John Light and Alexei Zamyatin). Shinobi is Bitcoin Magazine's technical editor, the old Block Digest co-host, and the house critic who will roast your favorite stack without asking permission. Together they treated CoinJoins, Lightning implementations, Nostr, Drivechains, BitVM, SeedSigner, and half the scaling theater as a roast menu. Fun. Educational. Brutally honest. Exactly what the title promised.

Taproot: Foundation Sold as a Privacy Spell

We opened with Bitcoin itself as the master of suck, then zeroed in on Taproot. Super Testnet's developer complaint was concrete: tweaked keys without an easy y-value (the old "charity bit"/parity pain) make construction miserable. "Give us back the freaking charity bit," he said, "because otherwise it sucks to make anything that uses tweaked keys." Number one on his list: Taproot and Bitcoin in general. Hard to argue when you have tried to build on it.

I played the tourist who upgraded for privacy theater. Multisig that looks like ordinary spends. Chain analysis obsolete. Taproot Wizards vibes. "It sounds good," Super Testnet shrugged. "Doesn't work." Tools for the full promise are still thin.

Shinobi was sharper about the marketing campaign: "The whole campaign to activate Taproot wildly oversold the privacy benefits and I shrieked at people the entire time." Taproot lays a foundation for protocols that can improve privacy if Lightning and other multisig stacks actually use it widely.

Influencers, in his telling, boiled that into "Taproot makes things private," which is not the same claim. Schnorr signatures do what they say. They are still harder to work with than the brochure implied, especially once you start aggregating and wrestling parity bits. We were not here to celebrate what works. We were here for the suck list.

LND's Microsoft Play & the Rent-Seeker Problem

Shinobi wanted LND first. Super Testnet did not think LND sucks. Perfect disagreement. Shinobi's charge is cultural and structural: Lightning Labs ships features without coordinating specs with other implementations, then lets market share turn the result into a de facto standard. "It's the Microsoft play of everybody uses our shit so whatever we do is just going to de facto be a standard," he said. Loop versus Blockstream/CLN's peer-swap. Liquidity ads versus Lightning Labs' own product attempts. The pattern, in his view, is ignoring protocol-level fixes so a company can monetize the shortcoming.

"Their business model is literally monetizing shortcomings of the protocol itself," Shinobi argued. After Tornado Cash and Samourai-shaped legal weather, leaving holes for middlemen feels like a broken incentive for an open protocol. Super Testnet pushed back hard. BOLTs 1 through 11 exist because the early teams did cooperate. Building products on top of an open protocol was the business model from day one. "They're just not doing it themselves and they're under no obligation to," he said of protocol-level liquidity work. Enough real bad behavior exists in the world without criticizing people for declining to build your preferred open fix. The roast stayed alive because neither side blinked.

Wasabi: Half a Protocol & a Tor Circus

I had the bad idea of reminding the room that Wasabi had long sponsored the show right as Shinobi warmed up. There were three CoinJoin stacks ordinary people actually used: Wasabi, Samourai, JoinMarket. Samourai's company layer was already in legal freefall. Wasabi's coordinator was shutting down. Why did they suck while they lived?

Shinobi's technical complaint was that Wasabi never fully implemented WabiSabi as designed. A huge piece was supposed to be multi-transactional coin selection that looks at the graph over time, not just the current round in a vacuum. "They never actually got around to doing that," he said, "which is one of the most critical pieces of making these massive multi-denomination CoinJoins work safely." He also claimed Tor circuit management on the coordinator path was a mess: nesting HTTP inside Tor circuits, redundant connection logic Tor already handles, even Cloudflare-shaped dependencies in places that should be "open a Tor circuit to your coordinator's hidden service, done."

I was in the room for earlier Wasabi drama and pushed the softer version of history: roadmaps have stages, NothingMuch designed ambitious crypto, Adam (nopara73) wanted something third parties could audit, and unfinished does not always mean abandoned. Shinobi was not buying "stage three is coming." Cutting corners, in his view, was the process. I even reminded him I once told NothingMuch to fork Wasabi and prove the better design in public. The architect blocked me for that, which is its own genre of open source tragicomedy. For the older maker-taker CoinJoin lineage that still tries to stay coordinator-light, my episode with Lord Kristaps on JoinMarket remains the companion piece.

Nostr: Slow JSON & Identity Without Rotation

On Nostr, for once, they mostly agreed. Super Testnet opened with the boring, damning UX truth: events ride fat JSON wrappers, so clients drag key-value labels, braces, and quotation marks around every like. "It makes it really easy to develop on, which is nice, but it makes it really slow," he said. Scroll a client and you feel the overhead. Optimized social apps do not ship that way. Nostr optimized for developer joy first.

Shinobi went deeper: cryptography is not a magic sticker. "There is zero actual comprehensive thought for key management," he said. Compromise a key and how do you roll it back? How do you tell followers to stop listening? How do you rotate without torching your identity? The delegation NIP is clunky.

Continuous signing with one immortal key is the default lifestyle. He wants master keys that rarely sign, child keys that rotate, revocation that clients understand the same way. Super Testnet partially agreed and pointed at his Magic Web Store pattern: every store gets a fresh key authenticated by a colder parent key.

"Every service should give you a new key," he said. Shinobi's addendum was protocol-level, not a thousand rival NIPs: one scheme every client can follow, or the social graph fragments into incompatible identity islands. Super Testnet preferred Darwinism: let many schemes compete and users migrate to the winner. Classic Nostr fight, filmed live.

SeedSigner, ESP32 Toys & Software That Starts by Sucking

Shinobi could not defend SeedSigner and did not try. Putting key management on a general-purpose Raspberry Pi stack (firmware, OS, wallet app, threat surface stacked like pancakes) is, in his words, one of the dumbest designs in the space. "The complexity of what you have to deal with… is just a giant mess of ways that things can be compromised," he said. He noted Keith Mukai's later work pushing toward ESP32-class microcontrollers that are simpler to reason about. I like SeedSigner for third-world practicality and deniable game-console forks. Shinobi likes threat models. Both can be true.

We drifted into Ben Arc's Bowser wallet joke (ESP32 that looks like you are playing Mario while it is actually a hardware wallet) and LNBits. Super Testnet loves LNBits and still thinks it belongs on a suck list. "Things start out sucking," he said. "Rarely is anything really good when it first comes out." Power users who complain a lot are often proving the tool is useful enough to hurt. Fiatjaf and Ben Arc will survive the roast. If you want the longer Nostr and Lightning mood music from the season, start with Fiatjaf on Nostr, Drivechains & why Lightning sucks.

Drivechains: Counting to 13,000 vs Magical Miner Custodians

Then the religious war. Super Testnet is notoriously pro-Drivechain. Shinobi is notoriously anti. I am sponsored by LayerTwo Labs and still get to have an opinion: ideologically I like scaling paths where people can hold keys, I worry about mining centralization, and if covenants let someone mock Drivechains in the wild I want the free market to stress-test miner behavior. "We really just need something that can count to 13,000," Super Testnet grinned. Shinobi is opposed to counting to 13,000. That is the joke and the thesis.

Shinobi's definition matters. A Drivechain-shaped system is an open sequencer and open peg that anyone can theoretically progress, but miners de facto monopolize because they choose what lands in blocks. Formalizing that in BIP 300/301 is, to him, magical thinking that miners are somehow better custodians than other custodians. "Miners take custody of your money," he said. "That is not a solution to scaling in which users control their own keys." Super Testnet reframed activation like adding CHECKMULTISIG before Liquid existed: ship the opcode family so people who want the construction can try it. Support for activation is not the same as believing it is the best scaling story.

I floated Matt Corallo's "stop calling it MEV" write-up: randomize sidechain ordering from recent Bitcoin block hashes so miners cannot extract extra value without burning mainchain blocks. Shinobi called the direction conceptually right and practically mismatched to Drivechains that cleanly separate sequencing from peg enforcement. If the peg cannot enforce the ordering rule, miners ignore it and still control withdrawals. The Corallo-style fix fits other peg designs better (including some BitVM-shaped dreams) than Paul's classic Drivechain split. For the longer Drivechain philosophy episode, see S9 E4 or the season's own S15 E16 comedy special where Paul cannot say Drivechains. Paul's in-episode LayerTwo ad, Shinobi ruled, still cannot meme.

Rollups Without Exits, RGB Tweaks & Data Dumping

Core Lightning got a short Super Testnet jab: Zeus wants the bookkeeper plugin for transaction history, and basic history should not feel like a plugin lottery with APIs that move. Then Shinobi lit into Bitcoin rollups. The classical rollup idea is that you write enough summary data on L1 so users can exit unilaterally. "That sucks," he said of the architecture even before the Bitcoin twist. The coming BitVM-flavored crop that cannot offer unilateral exit is worse: users need an operator's permission to leave, yet the design still burns mainchain space "for" an exit path that does not exist. Naming Citrea, Alpen Labs, and friends, he asked what the data is even for if exits are gated.

Super Testnet separated engineering theaters from token theaters and asked how this differs from RGB, which can hash data into a pubkey tweak so you embed state in a transaction you were already making. Cool when it works. Still often becomes small Bitcoin txs made just to carry the payload. His deeper ZK rant was data availability culture: Ethereum-world developers who do not want to run nodes end up dumping availability onto someone else's chain.

"The whole point of going offchain is to not have to use blockspace," Shinobi agreed. Putting the data back onchain as the scaling story is an oxymoron with a research budget. For the STARK end of that conversation from the same season, I already recorded Eli Ben-Sasson on STARKs, rollups & privacy (article still catching up). Covenant food fights live next door in Andrew Poelstra on OP_CAT vs OP_CTV.

Samourai, BitVM Bills & JoinMarket's Ceiling

Shinobi's Samourai roast was less about fixed five-person Whirlpool folklore (surge cycles later widened that) and more about culture. Years earlier he proposed a light-wallet path that would stop leaking balance data to backend servers for mixed coins: isolate a sub-derivation, cache your own outputs, never ship the master xpub for the mixed branch. "They just completely melted down… and got mad when I proposed this to them," he said, instead of shipping a small design change with a massive privacy win for people who will not run Dojo. Optional Tor kill-switches for "street" convenience got the same side-eye. Privacy software that invites you to disable Tor because Tor is flaky is solving the wrong outage.

BitVM's sin is price. Super Testnet: getting through a logic circuit can mean on the order of forty on-chain transactions, or fewer but still enormous transactions in later versions, so you still pay a fat multiple of normal spend cost. "It's freaking expensive, bro," he said. Shinobi added the structural limit: amazing expressivity inside a closed participant set, still dependent on keyholders who must sign to progress, cooperative or not.

Super Testnet's UniSAB attempt to bolt on unilateral exit made it more expensive on purpose. Hedgehog, his newer async Lightning direction after ZapLocker faceplanted into force-closes, was deliberately kept off the suck list for now. Robin Linus can take the expensive-but-honest critique; the longer BitVM origin story is S15 E13.

JoinMarket's complaint was stagnation plus structure. Self-hosted Tor order books finally landed (Shinobi had to scratch that gripe mid-sentence). The remaining itch is maker-taker topology: the taker coordinates, makers supply liquidity, and privacy compounding feels capped. Shinobi wanted proxy/smuggled inputs and more layered scrambling. Super Testnet wanted maker auctions instead of posted rates, plus blinding so the taker cannot map maker inputs to outputs. "That's why everything in Bitcoin sucks," someone laughed, right before I noted they had somehow skipped Ordinals and Runes. Shinobi's verdict: those are not Bitcoin projects. They dump data on our chain. "Get your own blockchain."

Have Fun Staying Poor

Follow Super Testnet on Twitter as SuperTestnet, on Nostr as two words, on Telegram without the underscore, on YouTube as High Level Bitcoin, and tip him on Stacker News. Shinobi is brian_trollz, still publishing at Bitcoin Magazine, still willing to call you names if you poke him enough. Block Digest, he admitted, is probably dead for good: too busy, too much research debt if you refuse to talk out of your ass. I told him the bar is too high and that Saylor-recap shows print money. "I would rather die poor," Shinobi said. Super Testnet agreed. They closed with a BTC Minstrel sea shanty chorus: have fun staying poor.

That is the episode in one room: two builders who use the tools enough to hate them correctly, a host who keeps inviting heretics, and a Pizza Day reminder that Bitcoin culture is healthiest when the roast is specific. Not "everything is captured." Not "everything is sacred." Just: here is the missing coin selection, the missing key rotation, the missing unilateral exit, the missing y-value, the rent-seeking Loop-shaped hole, the forty-transaction BitVM bill. Fix those or stop pretending the brochure shipped.

Listen to Super Testnet & Shinobi on Spotify, Apple Podcasts, YouTube, and More

Time stamps:

00:00:00 – Introducing Shinobi & Super Testnet
00:01:06 – Recording at Paralelni Polis
00:01:49 – Why Taproot Sucks
00:05:06 – Lightning Network
00:08:19 – Wasabi Wallet
00:16:15 – Nostr Social Media
00:17:09 – Cryptographic Key Management
00:18:19 – Cryptography Misconceptions
00:21:01 – Nostr's Lack of Key Management
00:21:47 – Proposed Solutions for Key Management
00:24:21 – Criticism of SeedSigner
00:28:23 – Alternative ESP32 Hardware Wallets
00:28:50 – LNBits and Iterative Improvements
00:30:04 – Drivechain Debate
00:34:42 – Random Ordering for Sidechains to Prevent MEV
00:34:45 – Drivechains and Miners
00:38:12 – Drivechains and Memes
00:39:02 – Core Lightning and Plugin Issues
00:42:04 – Bitcoin Rollups & The Unilateral Exit Issue
00:44:25 – RGB vs ZK Rollups
00:52:25 – Samourai Wallet Criticisms
00:57:42 – BitVM and Expensiveness
01:01:59 – JoinMarket Limitations and Criticism
01:05:05 – Non-Bitcoin Projects Dumping Data on the Blockchain
01:07:33 – Block Digest Reboot?
01:09:09 – Bitcoin Advocacy and Criticism