Other L1s

:::section{mins=2} Why Other L1s Exist :::

:::lead If [[gloss:ethereum|Ethereum]] and [[gloss:solana|Solana]] already cover both ends of the design spectrum — maximum [[gloss:decentralization|decentralization]] on one side, maximum speed on the other — what's left for everyone else? Plenty, it turns out: dozens of [[gloss:l1|L1s]] have launched since [[gloss:ethereum|Ethereum]] first arrived in 2015, and as of April 2026, a small set of them still hold real ground. :::

:::p{html} The standard story is that every [[gloss:blockchain|blockchain]] has to choose two out of three from the [[gloss:blockchain|blockchain]] trilemma (→ 1.3 §1 "[[gloss:consensus|Consensus]]") — speed, security, [[gloss:decentralization|decentralization]]. Other [[gloss:l1|L1s]] are essentially attempts to find a different combination than the two we already covered. Cosmos says: give every application its own sovereign chain. Avalanche says: let anyone spin up a custom subnet with their own rules. Polkadot says: share security across many chains so newcomers don't have to bootstrap their own [[gloss:validator|validator]] set. TON (The Open Network — a chain originally started by the Telegram team) says: forget the developer ecosystem race, plug into Telegram's user base — over a billion monthly active users as of 2026. :::

:::p{html} These bets came in waves. The first wave (2015-2018) was the original [[gloss:smart contract|smart contract]] platforms — [[gloss:ethereum|Ethereum]] (live since 2015, → 2.2 "[[gloss:ethereum|Ethereum]]"), and the alt-[[gloss:l1|L1s]] that came out of 2017's ICO boom and shipped mainnets in 2018: EOS, Tron, NEO. Most of them either folded into [[gloss:ethereum|Ethereum]]'s gravity well or kept living as niche networks. The second wave (2020-2021) was the "[[gloss:ethereum|Ethereum]] killers" — [[gloss:solana|Solana]] (→ 2.3 "[[gloss:solana|Solana]]"), Avalanche, Fantom, Harmony, Near, Algorand — funded by frustration over $40-50 average [[gloss:transaction|transaction]] fees on [[gloss:ethereum|Ethereum]], peaking together with the broader market in November 2021. Most of them lost 90%+ of their [[gloss:tvl|TVL]] by 2023 (Fantom went from $8B to under $500M, Algorand from $183M to under $100M). The third wave (2022-2024) brought the [[gloss:move|Move]]-based chains — [[gloss:aptos|Aptos]] ([[gloss:mainnet|mainnet]] October 2022) and [[gloss:sui|Sui]] ([[gloss:mainnet|mainnet]] May 2023), both founded by ex-Meta engineers using [[gloss:move|Move]], the language originally built for Meta's discontinued Diem project — and the Cosmos ecosystem catching its second wind. The current wave (2025-2026) is parallel-[[gloss:evm|EVM]] bets like [[gloss:monad|Monad]] ([[gloss:mainnet|mainnet]] November 24, 2025) and consumer-onboarding plays like Berachain ([[gloss:mainnet|mainnet]] February 6, 2025) and TON. :::

:::p{html} What survives, in 2026, is uneven. Cosmos has a healthy long tail of sovereign chains and a working IBC (→ §2 Cosmos). Avalanche is alive but smaller than at its 2021 peak — AVAX (Avalanche's native token) trades around 90% below its November 2021 all-time high — with subnets driving most of its growth. Polkadot is technically impressive but has struggled with adoption: parachain auctions priced out most teams, and the Agile Coretime reform that replaced auctions only landed in September 2024. TON is having a real moment thanks to Telegram, but the depth of its [[gloss:defi|DeFi]] is still shallow. The "[[gloss:ethereum|Ethereum]] killers" of 2020-2021 are mostly ghost towns. The lesson — easier to see now than at the peak — is that a faster TPS number is not a moat. Ecosystems take a decade to grow; many of these projects were trying to skip that step. :::

:::callout{icon=💡} Think of [[gloss:l1|L1]] blockchains as countries. [[gloss:bitcoin|Bitcoin]] and [[gloss:ethereum|Ethereum]] are the major economies — old, slow, but with deep institutions, currency, and trade networks that everyone has to deal with. [[gloss:solana|Solana]] is a fast-growing modern city-state that bet on speed and consumer experience. Cosmos is more like a federation of small republics, each writing their own laws but sharing a common diplomatic protocol. Avalanche is a country with a federal core and many semi-autonomous regions. TON is a country built around its messaging infrastructure: it doesn't care about competing on technology, it cares about reaching everyone you already talk to. The [[gloss:l1|L1]] landscape is not a single race; it's a map of different bets on what makes a [[gloss:blockchain|blockchain]] economy work. :::

:::keypoints{label="Key Points"}

• Fantom went from $8 billion [[gloss:tvl|TVL]] in 2021 to under $500 million by 2023 — a 95% drop in two years
• Most "[[gloss:ethereum|Ethereum]] killers" of 2020-2021 are now ghost towns — survivors all found a niche, not just better TPS
• Most native tokens of the 2020-2021 "[[gloss:ethereum|Ethereum]] killers" haven't reclaimed their all-time highs even in the current cycle — the market voted with its feet :::

:::details{badge=Advanced title="The Fat Protocol Thesis and the motivation for new L1s"} :::p{html} Joel Monegro's "Fat Protocols" essay (2016) argued that in Web3, value accumulates at the protocol layer (the [[gloss:l1|L1]]) rather than at the application layer — the inverse of Web2, where Google and Facebook capture most of the value while TCP/IP captures none. This thesis was a major motivator for launching new [[gloss:l1|L1s]]: every team wanted to be the next [[gloss:ethereum|Ethereum]] and capture protocol-level value through their token. :::

:::p{html} The thesis turned out to be only half true. Yes, [[gloss:ethereum|Ethereum]]'s market cap is much larger than any single application built on it. But applications like Uniswap, Aave, and MakerDAO have captured significant value too — through fee revenue, governance tokens, and ecosystem effects. And [[gloss:layer 2|L2s]] like Arbitrum and Base are now capturing value that was supposed to flow to [[gloss:l1|L1]]. The clean "protocols are fat" framing doesn't survive contact with the modular reality of 2026. :::

:::p{html} What's left of the thesis: [[gloss:l1|L1s]] still compete on a smaller axis — execution environment, [[gloss:data availability|data availability]], ecosystem grants, and developer tooling. The token narrative shifted from "buy the protocol" to "buy the network with the best fee-burn mechanics" or "buy the [[gloss:l1|L1]] whose [[gloss:validator|validators]] capture [[gloss:mev|MEV]]." The motivation to launch a new [[gloss:l1|L1]] in 2026 is rarely "be the next [[gloss:ethereum|Ethereum]]"; it's usually "find a defensible niche where you can build a sticky ecosystem." ::: :::

:::quiz{id=q-2-4-s1 correct=1 explanation="Correct. The [[gloss:l1|L1]] landscape is a long tail: [[gloss:ethereum|Ethereum]] and [[gloss:solana|Solana]] sit at the top, a small group of survivors (Cosmos, Avalanche, Polkadot, TON, [[gloss:aptos|Aptos]], [[gloss:sui|Sui]], BNB Chain, Tron) hold defensible niches, and a long graveyard of "killers" from 2020-2021 lost almost all activity. There's no single winner and no convergence — different bets on the trilemma keep the landscape diverse."} What's the most accurate description of how the [[gloss:l1|L1]] landscape settled by 2026?

• A single "winner" [[gloss:l1|L1]] absorbed most of the activity, leaving [[gloss:ethereum|Ethereum]] and [[gloss:solana|Solana]] as legacy systems
• Most "[[gloss:ethereum|Ethereum]] killers" lost their [[gloss:tvl|TVL]]; survivors are those who found a real niche (sovereignty, subnets, distribution)
• All [[gloss:l1|L1s]] converged on the same architecture, only the names differ
• [[gloss:l1|L1s]] are dead, all activity moved to [[gloss:layer 2|L2s]] :::

:::section{mins=3} Cosmos: Internet of Blockchains :::

:::lead Cosmos is the answer to a different question. Not "how do we make one fast [[gloss:blockchain|blockchain]]," but "what if every application got its own [[gloss:blockchain|blockchain]], and they could talk to each other like servers on the internet?" Through that lens, Cosmos is less a competitor to [[gloss:ethereum|Ethereum]] and more an alternative model of how the multi-chain world should be organized. :::

:::p{html} The Cosmos philosophy is called sovereignty: every application deserves its own chain, with its own [[gloss:validator|validators]], its own token, its own governance, and its own rules. On [[gloss:ethereum|Ethereum]], every dApp lives as a [[gloss:smart contract|smart contract]] on shared infrastructure — competing for the same [[gloss:block|block]] space, paying the same [[gloss:gas|gas]], ruled by the same upgrades. On Cosmos, you don't write a [[gloss:smart contract|smart contract]]; you launch a [[gloss:blockchain|blockchain]]. The clearest example is dYdX, a leading perpetual-futures DEX, which migrated from an [[gloss:ethereum|Ethereum]] [[gloss:layer 2|L2]] (StarkEx) to its own Cosmos chain in October 2023 specifically to escape [[gloss:gas|gas]] costs and gain control over its [[gloss:validator|validator]] set and order-matching logic. :::

:::p{html} To make this practical, Cosmos provides three components: :::

:::p{html}

• CometBFT (formerly Tendermint Core) — the [[gloss:consensus|consensus]] engine. A Byzantine Fault Tolerant protocol with instant [[gloss:finality|finality]] at around 6 seconds, capped at roughly 200 active [[gloss:validator|validators]].
• Cosmos SDK — the modular Go framework you assemble your chain from. Pick the modules you need (staking, governance, banking, IBC), and you have a working [[gloss:blockchain|blockchain]].
• IBC (Inter-[[gloss:blockchain|Blockchain]] Communication) — the protocol that lets Cosmos chains exchange tokens and messages trustlessly. TCP/IP for blockchains.

:::p{html} By 2025 IBC was moving several billion dollars in monthly volume across 115+ chains — "chain" here just means a standalone [[gloss:blockchain|blockchain]], each Cosmos chain having its own [[gloss:validator|validator]] set and its own rules. And — notably — IBC has never been exploited, unlike about half the projects on the cross-chain bridge Rekt leaderboard. :::

:::p{html} The Cosmos Hub itself is a chain in this ecosystem, holding the ATOM token (the native token of the Cosmos Hub) and serving as a routing point for IBC. But ATOM is not "the Cosmos token" the way [[gloss:ethereum|ETH]] is "the [[gloss:ethereum|Ethereum]] token" — every Cosmos chain has its own native token, and ATOM stakers secure the Hub itself plus a small set of opt-in "consumer chains" through Interchain Security (Stride is one current example). Most Cosmos chains run their own [[gloss:validator|validator]] set independently. :::

:::p{html} So why would an application use Cosmos at all if it can just launch its own [[gloss:blockchain|blockchain]]? Because "your own [[gloss:blockchain|blockchain]]" written from scratch means writing your own BFT [[gloss:consensus|consensus]], P2P networking, staking, governance, [[gloss:slashing|slashing]], indexing — easily two or three years of distributed systems work. Cosmos hands you all of that as ready-made modules, plus IBC connectivity to 115+ existing chains. It's the cheapest way to ship a real [[gloss:l1|L1]]: a small team can have a working sovereign chain in months instead of years. :::

:::p{html}

:::callout{icon=💡} [[gloss:ethereum|Ethereum]] is one big megacity: everyone lives, works, and trades in the same place, sharing roads and utilities and rules. Cosmos is more like a federation of cities connected by highways. Each city writes its own laws (its own rules), prints its own currency (its own token), and has its own police (its own [[gloss:validator|validators]]). They all use a shared road standard (IBC) to trade, but no city is in charge of the others. The trade-off: you have full control of your city, but if it fails, no neighboring city is obligated to help. :::

:::keypoints{label="Key Points"}

• Cross-chain bridges have collectively lost over $2.5 billion to exploits since 2021 — IBC has lost zero in the same period
• Cosmos chains often have a Nakamoto coefficient in the single digits — sovereignty is the trade for less [[gloss:decentralization|decentralization]] than [[gloss:ethereum|Ethereum]]
• Many projects that read as standalone [[gloss:l1|L1s]] in the news — Celestia, [[gloss:sei|Sei]], Injective, dYdX — actually run on the Cosmos SDK; the ecosystem hides inside the infrastructure :::

:::details{badge=Advanced title="CometBFT, Tendermint, and the cost of instant finality"} :::p{html} To understand why Cosmos chains feel different from [[gloss:ethereum|Ethereum]], it helps to look at how their [[gloss:consensus|consensus]] actually works. CometBFT (a [[gloss:fork|fork]] of Tendermint Core, maintained since February 2023 by a company called Informal Systems) is the engine that decides which [[gloss:block|block]] is next. Every few seconds it runs a small "election." One [[gloss:validator|validator]] is chosen as the proposer and sends out a candidate [[gloss:block|block]]. The other [[gloss:validator|validators]] look at it and vote in two rounds: first a "pre-vote" (do I think this [[gloss:block|block]] is valid?), then a "pre-commit" (am I willing to lock in?). When more than two-thirds of the staked tokens have signed off in both rounds, the [[gloss:block|block]] is committed. This whole dance takes about six seconds. :::

:::p{html} The interesting word here is committed. On [[gloss:bitcoin|Bitcoin]], a [[gloss:block|block]] is never truly final — there's always some tiny chance a longer chain appears and rewrites history, which is why exchanges wait for several confirmations before crediting a deposit. [[gloss:ethereum|Ethereum]] is similar: [[gloss:block|blocks]] become "more final" the deeper they are buried. CometBFT is different. The moment two-thirds of [[gloss:validator|validators]] have signed a [[gloss:block|block]], it cannot be reverted. There is no "wait six confirmations" on a Cosmos chain. This is what people mean by instant [[gloss:finality|finality]], and it's a real practical difference — exchanges can credit deposits in seconds, and bridges have a clean cut-off for what's confirmed. :::

:::p{html} The price of this design is the [[gloss:validator|validator]] count. Every [[gloss:validator|validator]] has to send a vote on every [[gloss:block|block]] to every other [[gloss:validator|validator]] — twice per round, in fact. The number of messages flying around grows roughly with the square of the [[gloss:validator|validator]] count, so the network gets clogged long before you can fit a million [[gloss:validator|validators]]. In practice CometBFT tops out around 200 active [[gloss:validator|validators]] (the Cosmos Hub keeps 180). [[gloss:ethereum|Ethereum]] sidesteps this entirely with a different design: it has roughly 900,000 [[gloss:validator|validators]] as of April 2026 (Casper FFG, → 2.2 §4 "[[gloss:the merge|The Merge]]"), but only a small randomly-chosen committee votes on each [[gloss:block|block]], with the rest sampling in over time. The two systems are solving the same problem with opposite trade-offs. :::

:::p{html} This is the deepest reason Cosmos chains are usually called sovereign rather than decentralized. A 200-[[gloss:validator|validator]] network is faster, finalizes sooner, and is easier to upgrade — but it leans on a smaller group of operators behaving honestly. Many app-chains in Cosmos have a Nakamoto coefficient (the number of top [[gloss:validator|validators]] you'd have to bribe to halt the chain) in the single digits. For applications that want their own rules and don't need [[gloss:bitcoin|Bitcoin]]-grade censorship resistance, that's an acceptable trade. ::: :::

:::quiz{id=q-2-4-s2 correct=2 explanation="Correct. Sovereignty is the practical motivation: dYdX wanted to remove [[gloss:gas|gas]] friction for users, run a custom on-chain matching engine that wouldn't fit in [[gloss:evm|EVM]] [[gloss:gas|gas]] limits, and have [[gloss:validator|validators]] that answer to dYdX governance. The Cosmos SDK and CometBFT made it feasible to launch a dedicated chain rather than live as another contract on shared infrastructure."} Why did dYdX migrate from [[gloss:ethereum|Ethereum]] to its own Cosmos chain?

• To avoid high [[gloss:l1|L1]] fees and the complexity of deploying on Arbitrum or Optimism
• To get IBC interoperability with other DEXes and share liquidity across Cosmos chains
• To remove [[gloss:gas|gas]] costs for traders, control the [[gloss:validator|validator]] set, and own its order-matching infrastructure
• Because dYdX's matching engine couldn't run inside the [[gloss:evm|EVM]] regardless of layer :::

:::section{mins=2} Avalanche: Subnets and Speed :::

:::lead Avalanche is a fast [[gloss:evm|EVM]]-compatible chain with a clever [[gloss:consensus|consensus]] algorithm and a peculiar three-chain structure. Its real bet, though, was on subnets — letting any team spin up a customized chain that uses Avalanche's [[gloss:validator|validator]] infrastructure but writes its own rules. :::

:::p{html} Internally, Avalanche is not one [[gloss:blockchain|blockchain]] but three, each tuned to a different job: :::

:::p{html}

• C-Chain — the [[gloss:evm|EVM]]-compatible chain (→ 2.2 §2 "[[gloss:evm|EVM]]"). This is where Avalanche's [[gloss:defi|DeFi]] lives: Trader Joe (now LFJ), GMX, Aave's Avalanche deployment. For most users, "Avalanche" means this chain.
• P-Chain — manages [[gloss:validator|validators]] and subnet metadata.
• X-Chain — handles native AVAX transfers using a DAG (directed acyclic graph) data structure rather than a linear chain.

:::p{html} The [[gloss:consensus|consensus]] algorithm — Avalanche [[gloss:consensus|Consensus]] — is genuinely novel. Instead of all [[gloss:validator|validators]] voting on every [[gloss:block|block]] (the standard BFT approach), each [[gloss:validator|validator]] repeatedly samples a small random subset of about 20 [[gloss:validator|validators]], asks what they think, and updates its own opinion based on the responses. If at least 14 of those 20 agree (70%, called α=14 in the code), the [[gloss:validator|validator]] becomes more confident. Repeat until confidence is high enough. The result is [[gloss:finality|finality]] in roughly 0.8 to 1.5 seconds that scales with the [[gloss:validator|validator]] count rather than degrading as more [[gloss:validator|validators]] join — the opposite of CometBFT's behavior. :::

:::p{html} AVAX, the network's native [[gloss:gas|gas]] token, gets burned on every [[gloss:transaction|transaction]] in the [[gloss:eip|EIP]]-1559 style, and the total supply is capped at 720 million — closer to [[gloss:ethereum|ETH]] economics than [[gloss:solana|Solana]]'s perpetual emission. :::

:::p{html} But the real story of Avalanche in 2026 is subnets, formally renamed Avalanche [[gloss:l1|L1s]] in the Etna upgrade on December 16, 2024, under the ACP-77 reform. The same reform cut the cost of launching an [[gloss:l1|L1]] by more than 99%: instead of a mandatory 2,000 AVAX stake per [[gloss:validator|validator]], [[gloss:validator|validators]] now pay a flat continuous fee of about 1.33 AVAX per month. A subnet (now an [[gloss:l1|L1]]) is its own independent chain, with its own [[gloss:validator|validator]] set, rules, and virtual machine, that opted to live within Avalanche's broader network rather than as a fully sovereign Cosmos chain. :::

:::p{html} The most active examples: :::

:::p{html}

• Beam — a gaming [[gloss:l1|L1]] from Merit Circle DAO with 4.5M+ unique [[gloss:wallet|wallets]].
• DFK Chain — the chain behind [[gloss:defi|DeFi]] Kingdoms; uses JEWEL as [[gloss:gas|gas]] instead of AVAX.
• Dexalot — a fully on-chain order-book DEX.

:::p{html} Where Cosmos says "every chain stands alone," Avalanche says "every chain stands together — choose your level of integration." :::

:::callout{icon=💡} Avalanche [[gloss:consensus|Consensus]] is like a public opinion poll done well. Instead of asking all 200 [[gloss:validator|validators]] every time (slow, expensive), each [[gloss:validator|validator]] asks 20 random ones. If 14 or more agree, that's already counted as the position of the majority. Repeat the poll a few times to be sure. Each poll is fast, the math compounds, and the network as a whole moves quickly. Subnets, meanwhile, are franchise locations: same brand, same operating standards, but each franchise gets to set its own menu, prices, and hours. :::

:::keypoints{label="Key Points"}

• AVAX trades around 90% below its November 2021 all-time high — most of the recent growth comes from subnets, not from C-Chain [[gloss:defi|DeFi]]
• Launching an [[gloss:l1|L1]] on Avalanche now costs less than a corporate SaaS subscription — ~1.33 AVAX per month per [[gloss:validator|validator]]; before, you needed roughly $150K of stake
• Avalanche finalizes when [[gloss:consensus|consensus]] converges — no fixed confirmation count to wait for; exchanges can credit AVAX in real time :::

:::details{badge=Advanced title="Avalanche Warp Messaging and the two consensus protocols"} :::p{html} Avalanche Warp Messaging (AWM, formally renamed Avalanche Interchain Messaging — ICM — after the Etna upgrade, with a contract layer called Teleporter) is the native cross-chain communication protocol between Avalanche [[gloss:l1|L1s]] — the project's answer to IBC. Chains sign messages with BLS signatures aggregated from their [[gloss:validator|validators]], and any other chain in the Avalanche system can verify those signatures without trusting an external bridge. This is structurally similar to IBC but limited to the Avalanche universe; you cannot, for example, ICM your way to a Cosmos chain or to [[gloss:ethereum|Ethereum]] without going through a separate bridge. :::

:::p{html} Avalanche actually runs two related [[gloss:consensus|consensus]] protocols, not one. Snowman is the linear version used on C-Chain and P-Chain — it produces a totally ordered chain of [[gloss:block|blocks]], which is what [[gloss:smart contract|smart contracts]] need (they require deterministic ordering to compute results). Snowball is the DAG-based version used on X-Chain, where [[gloss:transaction|transactions]] don't need to be globally ordered (since they're simple value transfers, not state-modifying calls). The DAG variant is faster but cannot host [[gloss:evm|EVM]]-style contracts. :::

:::p{html} Both protocols share the same core sampling-based primitive ("ask 20, see if 14 agree"), which is what gets called "Avalanche [[gloss:consensus|consensus]]" in marketing. The trade-off vs CometBFT is interesting: CometBFT gives you instant [[gloss:finality|finality]] for ~200 [[gloss:validator|validators]] with classic BFT message complexity; Avalanche gives you sub-second probabilistic [[gloss:finality|finality]] (mathematically very high probability, just not instant) and scales to thousands of [[gloss:validator|validators]] in theory. As of April 2026, Avalanche has roughly 1,300 active [[gloss:validator|validators]] on the Primary Network — more than any single Cosmos chain, but still less than [[gloss:ethereum|Ethereum]]. ::: :::

:::quiz{id=q-2-4-s3 correct=1 explanation="Correct. Avalanche [[gloss:consensus|Consensus]] uses repeated random sampling: each [[gloss:validator|validator]] queries about 20 random peers and accepts the result if 14 of them agree (around 70%, the parameter α=14 in the code). Repeating this poll a few times yields very high confidence. The clever property is that it scales with the number of [[gloss:validator|validators]] rather than getting slower as more join — the opposite of classic BFT."} What's the key idea behind Avalanche [[gloss:consensus|Consensus]]?

• All [[gloss:validator|validators]] must vote on every [[gloss:block|block]], like classic BFT
• Each [[gloss:validator|validator]] polls a random small subset and updates opinion based on majority — repeat until confidence is high
• A VRF picks a small committee per [[gloss:block|block]], like Algorand — only the committee votes
• A rotating leader proposes [[gloss:block|blocks]], [[gloss:validator|validators]] just verify and the chain moves on :::

:::section{mins=3} Polkadot: Shared Security :::

:::lead Polkadot solves the same problem as Cosmos — many chains that talk to each other — but with a different bargain. Instead of letting every chain bring its own [[gloss:validator|validators]], Polkadot rents you the [[gloss:validator|validators]] of the entire network. Build your chain, plug into the relay, and security is handled. :::

:::p{html} Polkadot's architecture has three pieces: :::

:::p{html}

• Relay Chain — the central spine. It does only [[gloss:consensus|consensus]] and security; you cannot deploy applications directly on it.
• Parachains — application chains that connect to the Relay Chain. Each has its own runtime (its own logic, written in Substrate, Polkadot's chain-building framework), but borrows the Relay Chain's [[gloss:validator|validators]] for [[gloss:finality|finality]].
• [[gloss:validator|Validators]] — stake DOT (Polkadot's native token) and validate [[gloss:block|blocks]] across all parachains. As of early 2026 the active set is around 600 [[gloss:validator|validators]], up from 297 in 2024 (after a series of referenda to expand the set). Fewer than [[gloss:ethereum|Ethereum]], but more than any single Cosmos chain.

:::p{html} Polkadot's [[gloss:consensus|consensus]] is called Nominated [[gloss:proof of stake|Proof of Stake]] (NPoS — proof-of-stake with nomination): regular DOT holders nominate [[gloss:validator|validators]] they trust, and the protocol uses an optimization algorithm to pick the active set with the most evenly distributed stake. The result is more concentrated than [[gloss:ethereum|Ethereum]]'s permissionless [[gloss:validator|validator]] set but more distributed than the typical Cosmos chain. The big trade-off vs Cosmos (→ §2 Cosmos) is exactly the security model: Polkadot gives shared security — every parachain inherits the full economic security of the Relay Chain — at the cost of less flexibility in [[gloss:consensus|consensus]] and runtime upgrade rules. Cosmos gives sovereign security — each chain runs its own [[gloss:validator|validators]] — at the cost of having to bootstrap that security yourself. :::

:::p{html} The hard part for Polkadot has been adoption. Parachain auctions launched on November 11, 2021; the first batch of winners (Acala, Moonbeam, Astar, Parallel, Clover) connected on December 17-18, 2021. The slot mechanism was an auction that locked DOT for two years per slot, which priced out most teams. :::

:::p{html} The Agile Coretime reform, activated on September 19, 2024, replaced auctions and crowdloans with a marketplace where you buy [[gloss:block|block]] compute time as needed: cheaper, more flexible, much friendlier to small projects. But the reform came late, and by then a lot of teams that might have built on Polkadot had already chosen Cosmos, an [[gloss:evm|EVM]] [[gloss:layer 2|L2]], or [[gloss:solana|Solana]]. :::

:::p{html} As of early 2026 there are roughly 65 actively producing parachains out of 216 registered. The leaders: :::

:::p{html}

• Hydration — DEX and [[gloss:defi|DeFi]] liquidity.
• Acala — [[gloss:defi|DeFi]] hub with the aUSD [[gloss:stablecoin|stablecoin]].
• Moonbeam — [[gloss:evm|EVM]]-compatible chain.
• Astar — multi-VM platform.
• Phala — privacy compute.

:::p{html} Cross-parachain communication uses XCM (Cross-[[gloss:consensus|Consensus]] Messaging — Polkadot's analog of IBC), and is trustless because it routes through the Relay Chain. :::

:::callout{icon=💡} Cosmos is a strip mall: each store hires its own security guards, sets its own hours, and runs its own cash register. The mall gives you a parking lot and a sign on the highway, but if your store's security is bad, no one helps. Polkadot is a luxury shopping center: there's a central security team that patrols all the stores, climate control, a shared sound system. You pay rent (in DOT, for coretime), and you must follow the building's rules — but you don't have to hire your own guards or run your own HVAC. Two business models, two trade-offs. :::

:::keypoints{label="Key Points"}

• Of 216 registered parachains, only about 65 are actively producing [[gloss:block|blocks]] — the gap between "registered" and "alive" is the real story
• Substrate is the most useful part of Polkadot's legacy — the sovereign-chain framework gets used well beyond Polkadot itself, including by teams with Cosmos backgrounds
• Agile Coretime arrived three years late — by September 2024 most builders had already chosen Cosmos, an [[gloss:evm|EVM]] [[gloss:layer 2|L2]], or [[gloss:solana|Solana]] :::

:::details{badge=Advanced title="GRANDPA, BEEFY, and Polkadot vs Cosmos in practice"} :::p{html} Polkadot uses two separate [[gloss:finality|finality]] protocols. GRANDPA (GHOST-based Recursive Ancestor Deriving Prefix Agreement) is the main one — it finalizes [[gloss:block|blocks]] across all parachains at the Relay Chain level using a Byzantine fault-tolerant voting scheme. BEEFY is a lighter-weight protocol designed specifically for verifying Polkadot's [[gloss:finality|finality]] on external chains; it's how bridges from Polkadot to [[gloss:ethereum|Ethereum]] prove that a Polkadot [[gloss:block|block]] was actually finalized without forcing the receiving chain to run a full Polkadot [[gloss:node|node]]. :::

:::p{html} The split is a clean separation of concerns. GRANDPA is for Polkadot-internal use; BEEFY is for the outside world. This matters because it makes Polkadot's bridge story more honest: the receiving chain only needs to verify a small set of [[gloss:validator|validator]] signatures, not replay the whole [[gloss:consensus|consensus]] process. Cosmos solves the same problem differently with light clients and IBC's relayers; both approaches work, but BEEFY's compact-proof model is specifically designed to be cheap to verify on [[gloss:ethereum|Ethereum]]. :::

:::p{html} In practice, despite Polkadot's technical sophistication, the Cosmos ecosystem (→ §2 Cosmos) ended up with more diverse adoption. Cosmos hosts Celestia (the leading modular DA layer), dYdX v4, Osmosis, Injective, and [[gloss:sei|Sei]] — projects with real on-chain volume. Polkadot's biggest parachains (Moonbeam, Acala, Astar) have struggled to attract comparable activity. Several explanations are plausible: the early auction structure was hostile to small teams, the Substrate framework has a steeper learning curve than Cosmos SDK, and Polkadot's narrative was harder to compress into a one-line pitch. The technical comparison and the adoption comparison don't always agree, and Polkadot is the cleanest current example. ::: :::

:::quiz{id=q-2-4-s4 correct=2 explanation="Correct. Shared vs sovereign security is the central design decision. A parachain inherits the Relay Chain's economic security automatically — you don't have to convince stakers to back your chain. A Cosmos app-chain has full sovereignty but must build its own [[gloss:validator|validator]] set, which is harder for a small project but allows complete freedom over [[gloss:consensus|consensus]] rules and upgrades."} What's the core difference between Polkadot's parachains and Cosmos's app-chains?

• Polkadot caps the number of parachains; Cosmos can host unlimited chains
• Polkadot parachains all run the same language (Substrate); Cosmos chains use a mix
• Parachains share the Relay Chain's security; Cosmos app-chains run their own [[gloss:validator|validators]] and bootstrap security themselves
• Cosmos app-chains can run [[gloss:smart contract|smart contracts]], parachains cannot :::

:::section{mins=2} TON: The Telegram Blockchain :::

:::lead Open Telegram. Tap the [[gloss:wallet|wallet]] button. Send your friend $20 in USDT — no app to install, no [[gloss:seed phrase|seed phrase]] to copy, just a chat. That ordinary moment is what TON is actually selling: not a faster [[gloss:blockchain|blockchain]], but a [[gloss:blockchain|blockchain]] that meets users where they already are. :::

:::p{html} TON's history is unusual. The original Telegram Open Network was a Pavel Durov project from 2018 with a $1.7 billion ICO. The [[gloss:sec-regulator|SEC]] blocked the launch, the project was settled, and Durov handed the codebase to the community as the TON Foundation. The project then went mostly quiet. It came back to life in 2023, when Telegram itself started integrating crypto features directly into the messenger: :::

:::p{html}

• The @[[gloss:wallet|wallet]] bot — an in-chat [[gloss:wallet|wallet]] that turns Telegram itself into a crypto [[gloss:wallet|wallet]], no separate app required.
• USDT in chat — native support for Tether's [[gloss:stablecoin|stablecoin]], send a friend $20 with one tap inside a conversation.
• Mini Apps (TMA) — Telegram Mini Apps, JavaScript apps that run inside the chat as WebViews. Their potential audience: the same 1B+ messenger users.

:::p{html} Suddenly the platform's distribution became its selling point. The 2026 numbers tell a lopsided story: :::

:::p{html}

• Telegram reports more than 1 billion monthly active users — that's the addressable audience.
• Toncoin (TON's native token, used for [[gloss:gas|gas]], staking, and a quirky DNS-style name system for .ton domains) had passed 100 million unique on-chain holders by late 2024, mostly driven by @[[gloss:wallet|wallet]] integration and tap-to-earn airdrops.
• [[gloss:defi|DeFi]] [[gloss:tvl|TVL]] on TON has been volatile: $15 million at the start of 2024 → a peak of roughly $600 million in late 2024 → back down to $150-300 million across 2025 by various measures. Modest by [[gloss:defi|DeFi]] standards, but a huge leap from where the chain was two years earlier.

:::p{html} The big 2024 story was mini-apps. Simple games and tap-to-earn loops embedded directly in Telegram chats. Notcoin hit roughly 35 million players and ran its TGE in May 2024; Hamster Kombat claimed around 300 million users and launched in September 2024. They spawned millions of new [[gloss:wallet|wallet]] addresses and very real attention, but converting that attention into real [[gloss:defi|DeFi]] activity has been slow: [[gloss:tvl|TVL]] did not keep up with the user count. :::

:::p{html} TON's bet, in the language of marketing, is a distribution play: get users on board through a familiar interface first, build the actual [[gloss:defi|DeFi]] economy second. Whether that "second" is years away or never is the open question. :::

:::callout{icon=💡} TON is the WeChat Pay of crypto. WeChat Pay is not the most sophisticated payments system in the world — Stripe, Visa, and Square all have richer feature sets — but it's built into the messenger that everyone in China already uses, which made it the dominant payment rail in the country basically overnight. TON is making the same bet on Telegram: the technology underneath is fine but unremarkable, the killer feature is that you don't have to download anything, sign up for anything, or learn a new app. You're already in the chat — the [[gloss:wallet|wallet]] is just one tap away. Distribution beats technology, sometimes. :::

:::keypoints{label="Key Points"}

• TON is the only major [[gloss:l1|L1]] whose user base existed before the chain — Telegram already had a billion users to inherit
• Hamster Kombat reached ~300 million users in 2024 — more than Instagram and Twitter combined acquired in their first years
• TON is the only top-30 [[gloss:l1|L1]] whose fate is tied to a single private company — [[gloss:bitcoin|Bitcoin]], [[gloss:ethereum|Ethereum]], and [[gloss:solana|Solana]] have no equivalent dependency :::

:::details{badge=Advanced title="Architecture, async messaging, and the centralization question"} :::p{html} TON's architecture is a multi-chain design with three layers: :::

:::p{html}

• Masterchain — the coordinating chain, storing protocol state and [[gloss:validator|validator]] scheduling.
• Workchains — independent application chains anchored to the Masterchain. Currently only one workchain is active, though the design allows up to 2³².
• Shardchains — each workchain can dynamically split into shards to parallelize load. In theory the network scales by spawning more shards as demand grows; in practice almost all activity sits on a small set of active shards.

:::p{html} [[gloss:consensus|Consensus]] uses a Catchain BFT protocol on top of [[gloss:proof of stake|PoS]], [[gloss:block|blocks]] finalize in roughly five seconds. The original [[gloss:smart contract|smart contract]] language was FunC — low-level, intentionally close to the [[gloss:bytecode|bytecode]] of the TON Virtual Machine (TVM). A more developer-friendly language called Tact has emerged since, but the developer ecosystem remains an order of magnitude smaller than [[gloss:solidity|Solidity]]'s (→ 2.2 §2 "[[gloss:evm|EVM]]") or Rust's (→ 2.3 §3). :::

:::p{html} The most unusual technical property is TON's asynchronous contract model. On [[gloss:ethereum|Ethereum]] and [[gloss:solana|Solana]], contract calls are synchronous: contract A calls contract B, gets a result, and the caller's [[gloss:transaction|transaction]] succeeds or fails atomically. On TON, contracts pass messages to each other, and each message is processed in a separate [[gloss:transaction|transaction]], asynchronously. This is good for shard scalability (shards do not have to coordinate in real time), but it is significantly harder for [[gloss:defi|DeFi]] composability. The "flash loan" pattern that's routine on [[gloss:ethereum|Ethereum]], where one [[gloss:transaction|transaction]] borrows, swaps, and repays atomically, is much harder to express in TON's model. Most TON [[gloss:defi|DeFi]] (STON.fi, DeDust) works around this rather than embracing it. :::

:::p{html} The other major question is centralization. The TON Foundation and Telegram have outsized influence on the protocol's direction; [[gloss:validator|validator]] distribution is improving (around 340 active [[gloss:node|nodes]] by late 2025), but it remains concentrated. Pavel Durov's August 2024 arrest in France over alleged moderation issues on Telegram had a knock-on effect on TON, both reputationally and in token price. He was released under judicial supervision in early 2025 and continues to lead the company. In May 2026, an even more direct plan called Make TON Great Again (MTONGA) launched: Telegram itself became the chain's largest [[gloss:validator|validator]], and fees were cut sixfold. For an [[gloss:l1|L1]] that markets itself as "the people's [[gloss:blockchain|blockchain]]," the dependence on a single corporate counterparty is a structural risk that does not have a clean answer in 2026. ::: :::

:::quiz{id=q-2-4-s5 correct=2 explanation="Correct. The bet is the inverse of the standard [[gloss:l1|L1]] playbook. Most chains spent years bootstrapping developer ecosystems and hoping users follow. TON started with over a billion Telegram users and a [[gloss:wallet|wallet]] inside the messenger, betting that [[gloss:defi|DeFi]] can be built on top of that user base later. Whether the conversion from "Telegram user" to "[[gloss:defi|DeFi]] user" actually happens is the open question."} What's the strategic difference between TON and most other [[gloss:l1|L1s]]?

• TON optimizes for [[gloss:stablecoin|stablecoin]] payments and remittances, minimizing [[gloss:defi|DeFi]] functionality
• TON was launched by Telegram rather than the open crypto community — it's the first corporate [[gloss:l1|L1]]
• TON bets on distribution first (Telegram users), [[gloss:defi|DeFi]] second — opposite of [[gloss:l1|L1s]] that built developer ecosystems before chasing users
• TON's architecture splits the network into dynamic shards, unlike any other top-30 [[gloss:l1|L1]] :::

:::section{mins=2} Other Notable L1s :::

:::lead Beyond the four ecosystems we just covered, the [[gloss:l1|L1]] landscape has half a dozen more chains that matter — each one alive because it found a real niche. Some are recognizably "another [[gloss:evm|EVM]]" (BNB, Sonic), some are workhorses with one killer function (Tron as the USDT rail), some are bets on a new programming model ([[gloss:aptos|Aptos]], [[gloss:sui|Sui]], [[gloss:monad|Monad]]), and some are economic experiments (Berachain). Worth knowing, even at a glance. :::

:::p{html} Start with the heavyweights nobody names as innovative but everyone uses: :::

:::p{html}

• BNB Chain — the Binance ecosystem chain. Cheap [[gloss:evm|EVM]], very large [[gloss:tvl|TVL]] (top-five among all [[gloss:l1|L1s]]), and an active set of 45 [[gloss:validator|validators]] (21 Cabinet with priority [[gloss:block|block]] production plus 24 Candidates rotated daily by stake). It's among the most centralized of the major chains, which most users don't care about because the [[gloss:gas|gas]] is low and the bridges work.
• Tron (TRX — Justin Sun's [[gloss:l1|L1]], [[gloss:mainnet|mainnet]] 2018) is the USDT highway. About half of all global USDT supply (roughly $80–85 billion in circulation) lives on Tron, and Tron and [[gloss:ethereum|Ethereum]] have been trading the top spot back and forth for over a year. The dominance grew out of remittance corridors built around Tron's low fees and its fixed "energy and bandwidth" model instead of variable [[gloss:gas|gas]].
• Cardano (ADA — [[gloss:l1|L1]] from 2017, founded by [[gloss:ethereum|Ethereum]] co-founder Charles Hoskinson) took the academic-first path. Ouroboros [[gloss:consensus|consensus]], Haskell-based contracts, formal verification at every step. The technology is well-formalized, but [[gloss:defi|DeFi]] adoption has been notoriously slow.

:::p{html} The new generation of "[[gloss:smart contract|smart contract]]" platforms is built on the [[gloss:move|Move]] language, originally developed at Meta for the discontinued Diem [[gloss:stablecoin|stablecoin]] project. Two main [[gloss:move|Move]] bets: :::

:::p{html}

• [[gloss:aptos|Aptos]] ([[gloss:aptos|APT]] — [[gloss:mainnet|mainnet]] October 2022) bets on classic parallel execution at the language level. [[gloss:move|Move]]'s resource model makes it possible to know which [[gloss:transaction|transactions]] touch which assets without executing them, similar in spirit to [[gloss:solana|Solana]]'s account-pre-declaration model.
• [[gloss:sui|Sui]] ([[gloss:sui|SUI]] — [[gloss:mainnet|mainnet]] May 2023) uses the same [[gloss:move|Move]] language but with an object-centric data model and a focus on gaming and consumer-app use cases.

:::p{html} The shared trade-off for [[gloss:move|Move]]: a new language with far fewer developers than [[gloss:solidity|Solidity]], and many [[gloss:solidity|Solidity]] idioms don't translate cleanly to it. :::

:::p{html} Three more bets from 2025-2026 worth flagging: :::

:::p{html}

• [[gloss:monad|Monad]] (MON — [[gloss:mainnet|mainnet]] live November 24, 2025) is a parallel [[gloss:evm|EVM]], fully compatible with [[gloss:solidity|Solidity]], executing [[gloss:transaction|transactions]] in parallel and optimistically rolling back conflicts. The claimed ~10,000 TPS held up in the first weeks of [[gloss:mainnet|mainnet]]: by early February 2026 the network had processed over 140 million [[gloss:transaction|transactions]], with 0.4-second [[gloss:block|block]] times and 0.8-second [[gloss:finality|finality]]. If [[gloss:monad|Monad]] sustains that pace under load, it gets the speed of [[gloss:solana|Solana]] with the developer ecosystem of [[gloss:ethereum|Ethereum]].
• Berachain (BERA — [[gloss:mainnet|mainnet]] February 6, 2025) launched with "Proof of Liquidity": [[gloss:validator|validators]] secure the network not just by staking the native token but by providing liquidity to [[gloss:defi|DeFi]] pools, tying network security directly to ecosystem depth.
• Sonic is the successor to Fantom: Sonic Labs rebrand announced August 1, 2024; [[gloss:mainnet|mainnet]] went live December 18, 2024; FTM-to-S token swap completed by January 2025. Sub-second [[gloss:finality|finality]], a new SonicVM virtual machine, and a SonicDB database with live pruning — [[gloss:node|nodes]] drop old data they no longer need on the fly, keeping the database compact without halts or maintenance windows. Technically interesting, commercially still small.

:::p{html}

:::callout{icon=💡} Picture the [[gloss:l1|L1]] landscape as a car market. [[gloss:ethereum|Ethereum]] is Toyota — reliable, everywhere, the safe default. [[gloss:solana|Solana]] is Tesla — fast, polarizing, ahead on user experience. BNB Chain is the Chinese EV that nobody on the coast respects but that outsells everything in volume. [[gloss:move|Move]]-chains ([[gloss:aptos|Aptos]], [[gloss:sui|Sui]]) are concept cars from a designer who used to work at a top brand: clever, beautiful, possibly the future, possibly never mass market. Berachain is the experimental hybrid where the engine is wired into the suspension somehow. Most people just want the Toyota. :::

:::keypoints{label="Key Points"}

• Tron has almost no native [[gloss:defi|DeFi]] — and that's not a bug, it's the niche: pure USDT settlement at scale
• [[gloss:move|Move]]'s developer pool is roughly 100× smaller than [[gloss:solidity|Solidity]]'s — better safety guarantees, fewer people who can build with them
• Berachain inverts the usual axis: stake more = more governance + more yield, tying network security to ecosystem depth :::

:::details{badge=Advanced title="Move VM, parallel EVM, and what really moves L1 metrics"} :::p{html} [[gloss:move|Move]]'s defining property is its resource-oriented type system. In [[gloss:solidity|Solidity]], a token is a number stored in a mapping, and you can write code that creates or destroys tokens accidentally if you're not careful. In [[gloss:move|Move]], a token is a resource — a value with linear types, meaning it cannot be implicitly copied or dropped. The compiler enforces that every resource has exactly one owner at every moment. This kills entire bug classes: double-spends, [[gloss:reentrancy|reentrancy]], accidental token creation. The trade-off is a steeper learning curve, a much smaller developer ecosystem, and the loss of certain [[gloss:solidity|Solidity]] idioms that don't translate cleanly. :::

:::p{html} The parallel [[gloss:evm|EVM]] bet is the opposite trade-off: keep [[gloss:solidity|Solidity]], rewrite the execution engine. [[gloss:monad|Monad]]'s approach is called optimistic concurrency, and it works roughly like this. Run all [[gloss:transaction|transactions]] in a [[gloss:block|block]] in parallel, as if none of them touch each other. Then check: did any two [[gloss:transaction|transactions]] touch the same account? If not — great, you're done. If yes — conflict; roll one of them back and re-run it. All of [[gloss:monad|Monad]]'s engineering boils down to making that rollback cheap; otherwise the parallelism gain gets eaten by retries. :::

:::p{html} [[gloss:monad|Monad]] [[gloss:mainnet|mainnet]] launched on November 24, 2025, and the claimed 10,000 TPS didn't stay on the slide deck — they held up under the first weeks of real load. That's the strongest argument yet against learning [[gloss:move|Move]]: any [[gloss:solidity|Solidity]] contract deploys unchanged and runs an order of magnitude faster. If that's true, why switch languages? :::

:::p{html} So what actually moves [[gloss:l1|L1]] metrics in 2026? Looking at Cosmos vs Polkadot and the "[[gloss:ethereum|Ethereum]] killers" graveyard, the answer is almost never raw TPS. Four other parameters correlate with survival: :::

:::p{html}

• Organic [[gloss:tvl|TVL]] — the kind that doesn't depend on incentive grants. When the program ends, the liquidity stays.
• Full-time developer growth — Electric Capital publishes the tracker. Counted are sustained contributors, not one-off hackathon commits.
• Unique active addresses — separate from bots and sybil accounts, to the extent that's measurable at all.
• Real fee revenue — paid by actual users, not subsidized by the team out of the treasury.

:::p{html} A chain with 100,000 TPS in a benchmark and five active dApps is a marketing number. A chain with 50 TPS and a working economy is an [[gloss:l1|L1]] that will still be alive in five years. ::: :::

:::quiz{id=q-2-4-s6 correct=2 explanation="Correct. Tron's role isn't [[gloss:defi|DeFi]] or [[gloss:smart contract|smart contracts]] — it's USDT settlement. About half of all global USDT supply (~$80 billion in circulation) lives on Tron, and Tron and [[gloss:ethereum|Ethereum]] have been trading the top spot for over a year. The combination of low fees and a network of remittance services built around it made Tron one of the two main rails for global [[gloss:stablecoin|stablecoin]] transfers. This is a real economic moat that doesn't show up in [[gloss:tvl|TVL]] rankings or developer counts but matters for actual usage."} Why is Tron a major chain even though it's not a "[[gloss:smart contract|smart contract]] platform" in the usual sense?

• Tron has the lowest [[gloss:gas|gas]] fees of any [[gloss:l1|L1]], undercutting everyone on cost
• It became the leading chain for meme-coin launches
• About half of all USDT supply lives on Tron (~$80 billion in circulation), mostly via remittance corridors
• Tron was the first successful [[gloss:ethereum|Ethereum]] [[gloss:fork|fork]], inheriting its developer base :::

:::section{mins=2} Blockchain Trilemma in Practice :::

:::lead "Pick two of three: speed, security, [[gloss:decentralization|decentralization]]." That framing, popularized by Vitalik Buterin around 2017, is the cleanest summary of why no [[gloss:l1|L1]] has ever pulled clear of the others. By 2026 the trilemma is less a theoretical problem and more a map of who chose to give up which axis. :::

:::p{html} The trilemma has a simple shape but a sharp edge: any technique you use to make a [[gloss:blockchain|blockchain]] faster tends to either reduce [[gloss:decentralization|decentralization]] (fewer [[gloss:validator|validators]], higher hardware requirements) or weaken security (smaller stake, faster reorg windows, less economic [[gloss:slashing|slashing]]). Take three points on the spectrum to see the frame: :::

:::p{html}

• [[gloss:bitcoin|Bitcoin]] and [[gloss:ethereum|Ethereum]] [[gloss:l1|L1]] (→ 2.2 [[gloss:ethereum|Ethereum]]) — one end: kept maximum [[gloss:decentralization|decentralization]] and security, accepting slow throughput as the cost.
• [[gloss:solana|Solana]] (→ 2.3 [[gloss:solana|Solana]]) — the opposite end: kept high throughput and security but accepted a smaller, more concentrated [[gloss:validator|validator]] set.
• BNB Chain (→ §6) — same trade-off as [[gloss:solana|Solana]], pushed further: an active set of only 45 [[gloss:validator|validators]], elected daily by stake.

:::p{html} Cosmos chains (→ §2 Cosmos) and Avalanche subnets (→ §3 Avalanche) sit in a different spot: they keep speed and instant [[gloss:finality|finality]] but reduce shared security in favor of per-chain sovereignty. Each Cosmos chain has its own [[gloss:validator|validator]] set, often with a Nakamoto coefficient in the single digits — meaning a small number of operators could collude to attack it. The justification is that for many use cases (gaming, niche [[gloss:defi|DeFi]], enterprise) the level of [[gloss:decentralization|decentralization]] [[gloss:ethereum|Ethereum]] [[gloss:l1|L1]] provides is overkill, and the speed and customization is worth the trade. :::

:::p{html} [[gloss:ethereum|Ethereum]]'s particular position is unique. As of 2026, it's the only [[gloss:l1|L1]] with all of these properties holding at the same time: :::

:::p{html}

• around 892,000 active [[gloss:validator|validators]] on the [[gloss:consensus|consensus]] layer;
• five-plus production execution clients (Geth, Nethermind, Besu, Erigon, Reth) plus five [[gloss:consensus|consensus]] clients;
• hardware requirements low enough to validate from a Raspberry Pi-class machine;
• no single entity over a third of stake — the largest, Lido, sits around 23-24%, and the community is actively working to keep it from crossing the 33% line.

:::p{html} Most of the survival arguments for "[[gloss:l1|L1]] still matters" are really survival arguments for that specific combination — and the bet is that scaling without breaking it is possible through [[gloss:layer 2|L2s]], [[gloss:blob|blob]] storage, and statelessness rather than through hardware-driven monolithic scaling. :::

:::p{html}

:::callout{icon=💡} The trilemma is the engineer's classic "fast, cheap, good — pick two." A construction firm can build you a house quickly and well, but it won't be cheap. Or quickly and cheaply, but it won't be good. Or cheaply and well, but not quickly. Blockchains play the same game on a different axis: fast, secure, decentralized — pick two. [[gloss:solana|Solana]] picked fast and secure, sacrificing [[gloss:decentralization|decentralization]]. [[gloss:ethereum|Ethereum]] [[gloss:l1|L1]] picked secure and decentralized, sacrificing speed. The [[gloss:layer 2|L2]]-augmented [[gloss:ethereum|Ethereum]] stack tries to take all three by splitting the work between layers, the way construction firms split work between general contractors and specialists. :::

:::keypoints{label="Key Points"}

• Vitalik's "two of three" framing came in 2017 — almost every [[gloss:l1|L1]] in this card was launched after that essay was written
• Until December 2025 [[gloss:solana|Solana]] ran on one independent client; the full Firedancer release gave it a second — [[gloss:ethereum|Ethereum]] has five-plus
• If [[gloss:validator|validator]] hardware costs $50K, [[gloss:decentralization|decentralization]] is gated by capital, not by interest — this is [[gloss:solana|Solana]]'s structural ceiling :::

:::details{badge=Advanced title="The trilemma is a simplification — what's actually multidimensional"} :::p{html} "Speed, security, [[gloss:decentralization|decentralization]]" is a useful storytelling device but compresses a much messier picture. The real trade-off space includes at least twelve axes: :::

:::p{html}

• Latency (time-to-inclusion) — how fast a [[gloss:transaction|transaction]] lands in a [[gloss:block|block]];
• Throughput — sustained TPS under load, not the peak benchmark;
• [[gloss:finality|Finality]] time — when can a [[gloss:transaction|tx]] be treated as irreversible;
• Per-[[gloss:transaction|tx]] cost — [[gloss:gas|gas]] or its equivalent;
• Composability — can dApp A call dApp B atomically (so that either both operations succeed together, or both roll back)?
• Developer experience — how fast can a team build and ship;
• Censorship resistance — can a [[gloss:validator|validator]]/miner drop your [[gloss:transaction|tx]];
• [[gloss:mev|MEV]] fairness — who captures the value of [[gloss:transaction|transaction]] ordering;
• Geographic distribution of [[gloss:validator|validators]] — across countries and jurisdictions;
• Client diversity — how many independent implementations exist;
• Hardware ceiling — where performance growth runs out of headroom;
• Economic [[gloss:finality|finality]] — what it costs to attack the network.

:::p{html} Each of these axes can pull against the others, and almost none of them are captured by the trilemma. :::

:::p{html} Three concrete examples of how composability cuts across the chains we just covered: :::

:::p{html}

• [[gloss:solana|Solana]] — very low latency and high throughput, but composability is constrained by leader scheduling: your [[gloss:transaction|transaction]] is at the mercy of whoever is leader for the current 400ms slot.
• [[gloss:ethereum|Ethereum]] — worse latency and throughput, but better composability across its [[gloss:l1|L1]] ecosystem, because every dApp shares the same global state.
• Cosmos chains — low latency within a chain, but cross-chain composability requires relayers and IBC packets.

:::p{html} The most useful framing in 2026 is to think of [[gloss:l1|L1s]] as different points in a multi-dimensional design space, each optimizing for the bottleneck its target use case actually has: :::

:::p{html}

• High-frequency trading dApp prioritizes latency and throughput above everything else.
• Real estate registry prioritizes [[gloss:finality|finality]], censorship resistance, and economic security.
• DAO prioritizes governance composability and [[gloss:gas|gas]] costs.

:::p{html} The trilemma is the headline; the real engineering happens in the fine print. ::: :::

:::quiz{id=q-2-4-s7 correct=1 explanation="Correct. [[gloss:decentralization|Decentralization]] isn't a single number — it's a combination of properties that have to hold simultaneously. Lots of [[gloss:validator|validators]] alone don't help if they all run the same client ([[gloss:solana|Solana]]'s pre-Firedancer history), if they all need expensive hardware (which would push validation toward big operators), or if a few entities hold most of the stake. [[gloss:ethereum|Ethereum]]'s claim is that all four conditions are met at once."} What does [[gloss:ethereum|Ethereum]]'s claim to "maximum [[gloss:decentralization|decentralization]]" actually rest on?

• [[gloss:ethereum|Ethereum]] has more [[gloss:validator|validators]] than any Cosmos chain, which by itself makes it more decentralized
• A combination of [[gloss:validator|validator]] count, client diversity, low hardware requirements, and no single entity over 33% stake
• [[gloss:ethereum|ETH]] is the largest proof-of-stake asset by market cap, so attacking it is economically prohibitive
• The [[gloss:ethereum|Ethereum]] Foundation doesn't control the protocol; all upgrades are decided by [[gloss:validator|validator]] vote :::

:::section{mins=2} How to Evaluate a New L1 :::

:::lead A new [[gloss:l1|L1]] launches roughly every month. Most of them won't be alive in three years. The hard part is picking the few that will — and the harder part is doing it without falling for the marketing. Here's a checklist that survived contact with the 2020-2021 graveyard. :::

:::p{html} Start with the red flags. If even one of the six lights up, dig deeper before you commit: :::

:::p{html}

• "[[gloss:ethereum|Ethereum]] killer" in the marketing — itself a red flag in 2026; the surviving [[gloss:l1|L1s]], even the strongest ones, no longer market themselves that way.
• TPS numbers from a testnet — meaningless; what matters is the throughput the network actually sustains in production.
• No real applications six months after launch — a very bad sign; strong networks bootstrap real dApps in the first three to six months, even if usage is small.
• 90%+ of [[gloss:tvl|TVL]] coming from incentive grants — the apparent activity vanishes the day the incentive program ends.
• Fewer than five independent [[gloss:validator|validators]] — structurally that's a federated database, not a [[gloss:blockchain|blockchain]].
• Anonymous or first-time-founder team — fine for a small experimental project, not for a chain asking for serious capital.

:::p{html} The green flags are inverted: :::

:::p{html}

• A genuinely novel technical bet — not "faster" or "cheaper" but something with a clear architectural insight: [[gloss:solana|Solana]]'s PoH, Avalanche's repeated sampling, [[gloss:move|Move]]'s resource types.
• Organic developer growth tracked publicly by the Electric Capital Developer Report; paid bounties don't count, sustained interest does.
• Real fee revenue from real users, visible on Token Terminal or DefiLlama, separate from token incentives.
• Live applications with real users, not just deployment counts on the explorer.
• Client diversity, or a credible plan for it — even one alternative client in development is much better than zero (→ §7).
• Audits and bug bounties with reputable firms, not just internal reviews.

:::p{html} One practical detail that determines how easy it is for a user to get on: :::

:::p{html}

• [[gloss:evm|EVM]]-compatible [[gloss:l1|L1]] (BNB Chain, Avalanche C-Chain, Sonic, [[gloss:monad|Monad]], any Cosmos chain with [[gloss:evm|EVM]] support) — your existing [[gloss:wallet|wallet]] (MetaMask, Rabby, etc.) just works; you only need to add the network's RPC.
• Non-[[gloss:evm|EVM]] [[gloss:l1|L1]] ([[gloss:solana|Solana]], Cosmos chains without an [[gloss:evm|EVM]] module, TON, [[gloss:aptos|Aptos]], [[gloss:sui|Sui]]) — you'll need a separate [[gloss:wallet|wallet]] for that ecosystem.

:::p{html} This is one of the simplest practical reasons [[gloss:evm|EVM]]-compatible chains have an easier time gaining adoption: zero friction for users who already have an [[gloss:ethereum|Ethereum]] [[gloss:wallet|wallet]]. :::

:::callout{icon=💡} Evaluating an [[gloss:l1|L1]] is closer to startup due diligence than to picking a savings account. A polished landing page is not a working product. Testnet TPS is a demo, not a track record. Investor names on the deck are fine but they don't validate the product. The questions that actually matter are mundane: who's using it today, what are they actually doing, who's building on top of it without being paid to, and how would you know if those numbers were faked? "I read the [[gloss:whitepaper|whitepaper]] and it sounds technically novel" is the start of the analysis, not the end. :::

:::keypoints{label="Key Points"}

• Day-one [[gloss:tvl|TVL]] on a new chain is almost always paid for by the foundation — the leading indicator is whether that activity persists six months after the program ends
• "[[gloss:ethereum|Ethereum]] killer" became a self-disqualifying phrase around 2023 — every survivor in this card stopped using it
• A chain a year old with no dApp you can name from memory has already failed the test — it just hasn't been declared dead yet
• The [[gloss:lindy-effect|Lindy effect]] applies hard to [[gloss:l1|L1s]]: the burden of proof sits on the new project, not on the chain that's been alive for a decade :::

:::details{badge=Practical title="Concrete metrics to watch and how to use them"} :::p{html} Six metrics that, taken together, are reliable enough to actually use: :::

:::p{html}

• Daily Active Addresses — but only if you can roughly correct for sybils, which most chain explorers don't.
• [[gloss:tvl|TVL]] trend, not absolute [[gloss:tvl|TVL]]: a chain at $500M with steady growth is healthier than a chain at $1B that's bleeding 5% per month.
• Developer count from Electric Capital's report (publicly available, updated quarterly) — focus on the growth rate of full-time developers, not headline counts that include drive-by commits.
• Fee revenue from Token Terminal or DefiLlama, separated from inflation rewards: the cleanest measure of "are users actually paying for blockspace."
• Nakamoto coefficient — how many distinct entities does an attacker need to compromise to control 33% of the network?
• Uptime history — any unplanned halts in the last 12 months, and how long.

:::p{html} None of these is decisive on its own. A chain can have a high developer count from a few enthusiastic teams; growing [[gloss:tvl|TVL]] can come from one whale moving funds; fee revenue can be inflated by a single high-volume dApp. The pattern is in the combination — the chains that look healthy on multiple axes simultaneously are the ones that tend to survive. The 2020-2021 "[[gloss:ethereum|Ethereum]] killers" mostly looked good on TPS and [[gloss:tvl|TVL]] while being weak on developer growth, fee revenue, and Nakamoto coefficient. Those second-tier metrics turned out to be the leading indicators of survival. :::

:::p{html} One final framing — the [[gloss:lindy-effect|Lindy effect]]: the longer something has existed, the longer it's likely to keep existing. It applies hard to [[gloss:l1|L1s]]. [[gloss:bitcoin|Bitcoin]] has 17+ years of continuous uptime and economic survival. [[gloss:ethereum|Ethereum]] has 10+. [[gloss:solana|Solana]] has 5+. A new [[gloss:l1|L1]] starting in 2026 has zero. This doesn't mean new is bad; it means the burden of proof on a new project is high. To displace existing infrastructure, a new [[gloss:l1|L1]] needs to be roughly an order of magnitude better on a dimension that real users care about — not a single metric in isolation, and not just a faster benchmark. Most [[gloss:l1|L1s]] that try to clear that bar do not. ::: :::

:::quiz{id=q-2-4-s8 correct=2 explanation="Correct. All four are reasonable questions, but C is the one to ask first. Day-one [[gloss:tvl|TVL]] on a new chain almost always comes from incentive programs — the foundation pays for liquidity, dApps redeploy with grants, and the [[gloss:tvl|TVL]] number looks great until the program runs out. dApp counts (A) and unique addresses (D) are easy to inflate in the first weeks; VC backing (B) doesn't validate the technology. The leading indicator of survival is whether the activity persists six and twelve months later, when the subsidies end. Almost every "[[gloss:ethereum|Ethereum]] killer" of 2020-2021 looked great on day one and was nearly empty two years later."} A new [[gloss:l1|L1]] launches with a flashy testnet showing 50,000 TPS and a $200M [[gloss:tvl|TVL]] on day one. What's the first question to ask?

• "Which dApps are already deployed on [[gloss:mainnet|mainnet]]?"
• "Which top crypto-VCs are backing the team?"
• "Where does the [[gloss:tvl|TVL]] actually come from, and what happens when the incentive program ends?"
• "How many unique addresses transacted in the first week?" :::