December 11, 2023

Classification of blockchains: understanding the nuances

Article Reading Time:
16 min.

Classification of blockchains: understanding the nuances

Opinion resident expert Vladimir Popov aka Menaskop explores the gradation of blockchain solutions on a variety of grounds: from well-known to rarely mentioned.

Why is it worth reading this article? 

Believe me, classification is much morea fascinating activity than, for example, looking at the stars and looking for constellations among the myriads of constellation points glowing in the night sky. Especially if this is your first time coming from the Northern to the Southern Hemisphere (or vice versa). 

After all, Bitcoin in 2023 has passed its second decade since the first hype (2013), and 15 years have passed since paper as a concept. And during this time, a lot has changed. 

Among other things, for me, as a crypto enthusiast, it is extremely important to present an approach where there is no place for standard classifications that are not standard at all. I hope it is for you too.

Summary of previous episodes

To begin with, let me remind you of the basic things, without which you can’t go anywhere:

  1. Bitcoin is a cryptocurrency based on blockchain No. 01, but it was immediately followed by altcoins for various reasons. I talked about this in one of the previous articles.
  2. The word “blockchain” itself is not mentioned in the Bitcoin White Paper: this concept, like the concept of cryptocurrency, appeared thanks to the media to facilitate understanding of the technology.
  3. In addition to blockchain, there are already severaltools that automate work in an untrusted environment, such as: DAG; IPFS and similar databases and so on. But this article will focus specifically on blockchains, and only them. 

So, let's get started? Perhaps yes. 

Base No. 01. Gradation by generation

I’ll highlight it first with a list (and will continue to adhere to this approach):

  1. Firstgeneration: Bitcoin and its altcoins (such as Litecoin).
  2. Secondgeneration: Ethereum and other chains with smart contracts.
  3. Thirdgeneration: I call them specialized. 

Is there a fourth generation? I'm sure yes, but first let me say a few words about the first three. 

Bitcoin's blockchain is evolving: Therefore the Bitcoin blockchain of 2009 and 2023— not exactly different blockchains, but still different in many respects. Especially after Taproot, from which Ordinals and other layers grew. Yet the first generation bet that every blockchain solves the blockchain trilemma in its own way: Litecoin is faster than Bitcoin; DogeCoin - cheaper; X11 (Dash) - more anonymous and so on. 

In this sense, there is no clear time limit,since such blockchains are still emerging and are alive (see Monero or, for example, the Mimblewimble family, which includes a variety of generations). And yet this is approximately 2013, the year of the first ICO, which was followed in 2014 by one of the most important ICOs in the entire crypto industry. ICO project of Vitalik Buterin Ethereum. 

Concerningsecond generation, then the easiest thing to do with it is:the bet was placed on full-fledged smart contracts. And, as the confrontation with Bitshares (Graphene) showed, it was precisely this, and not the consensus (since in the end, the ether switched, although not to DPoS, but still to the modernized PoS) that lay the stumbling block for the next ten years. 

What exactly are smart contracts for?It is true that the blockchain can “sing, dance, and sell a little.” That is, it became not specialized, but universal, so in the EVM subfamily we see DAOs, DeFi, DeSoc, DeSci, NFT hypes and much more. 

Third generationin this sense, it went again according to specialization:Solana pursued speed, Near pursued scalability (sharding). And then Aptos and others. Here, in terms of dates, everything will be much more complicated: 2018 can be considered the starting point, because... crypto-winter was upon us, but the projects received quite a lot: let’s say, the same Solana ended up grabbing more than $500M. 

But the battle for TPS and similar parameters— this is a battle with phantoms. The recent overtaking of the very same Visa on the sharp turn of Bitcoin, which has been inadvertently cited as an example for more than ten years, trying to point out the negative sides of the blockchain, proved this best of all. 

And yet there is a certain, even marketing, flair to thisspecializations remained: Cardano is a scientific blockchain (although there are clearly no less scientific developments in the EVM family); Solana — “if you want to work quickly”; Near - “if you want to work simply” and so on. 

And here we move on to the question asked above: “So is therefourth generation? I'm sure so.And here’s why: if you look at Polkadot parachains, Avalanche subchains, Cosmos hubs and other similar examples, it becomes clear: specialization has again retreated a little and passed the baton to versatility. But only now the focus is no longer on smart contracts (the internal aspect of the blockchain), but on multi- and cross-chain mechanics (the external aspect). 

I will say more:thanks to EVM, it was Ether who first tested this path. Judge for yourself: EvmOs in Cosmos, Moonbeam in Polkadot, Avrora in Near, Neon in Solana, EVM EOS, EVM for XRP Ledger (Ripple), EVM on Avalanche. And this is not counting BSC (BNB chain), Harmony, ETHPoW, Cronos, Polygon and L2 solutions on rollups (Arbitrum, Optimism; zkSync, etc). All this is precisely a huge multi-chain network with a single operating module. The full list can be found, for example, here. 

Of course, they will object to me thatThe Tendermint family has its advantages, and the so-called ether killers (more correctly, suicides) are a dime a dozen (from the same ICO on Ethereum - Tron to the giant in terms of fees, but not infrastructure, EOS). And I myself am not an Ethereum maximalist, but it is obvious that the approach with EVM turned out quite well. 

And therefore, such a vector of development is definitely appropriate. And from there we will move on to the next foundation. 

Base No. 02. Division by levels

If in previous examples we analyzed, letand very conditionally, the gradation is horizontal (since, for example, in the EVM subfamily there are both vertical and horizontal connections), now let’s consider the vertical one. 

At first glance, everything is simple:

  • L0 — cross-chain level (ZeroLayer, EYWA, etc); 
  • L1 — blockchains of a standard type (this can be Bitcoin, Ethereum, Cardano, and many others); 
  • L2 — here again there will be many branches: sidechains and Plasma-like solutions (Polygon), optimistic (Optimism, Arbitrum, etc) and zk-rollups (zkSync, etc) and so on;
  • L3 — the level is still more theoretical,but already being developed from two sides at once (from the side of applications and high-level protocols - see, for example, Envelop, and from the position of the blockchain itself and low-level protocols - a solution from the same zkSync). 
  • Will itL4? Perhaps, but let's leave him alone for now. 

    Vertical gradation solves the same provisionstrilemma, like the horizontal one, but here security, as a rule, is not inherited from the main chain, but is assigned to it. However, the options may be different. 

    The main thing is that L0-L3 is developed again and again inthe entire spectrum is fully functional in the EVM family: while the same Polkadot (or Cosmos – or whichever you like best: choose for yourself) focused on L0, and even L0.5, Vitalik and the projects of the Ethereum ecosystem provided the possibility of non-linear development and on L0 and L2 at the same time. 

    And for this reason we see a battle of walletstoday (and very soon - with the SEC, of ​​course, but this is an economic, not a technical aspect, although it is precisely important for the development of L2). Massadoduction has been the mantra of the last ten years. And the one who makes simpler and simpler solutions... does not necessarily win: MetaMask is still ahead of Trust, although the interface part clearly irritates usability engineers and UX/UI designers. 

    For me personally, the L0-Ln circuit is anothera level of decentralization and abstraction at the same time that is worth fighting for. And for you? Answer the question - and move on to the next, third, basis. 

    Base No. 03. Differentiation by Consensus

    Let me make a reservation right away that on the one handI have researched this issue more than once, and I have something to say. But, on the other hand, the classification of consensuses itself is not the subject of this article, so I will try to present it as briefly and clearly as possible. 

    So, for me, consensuses at the lowest level can be divided into four basic levels (although they can be combined into two mega-levels):

    1. Those where you need to do something with the physical world,to reach consensus (it doesn’t matter whether it’s the digitized energy of power plants or your steps): and in this sense, not only PoW, but also PoC and many others are suitable here.
    2. Those where you need to put something as collateral in order to find technical associates  (PoS, LPoS, DPoS, naive PoS and so on). 
    3. Those where a certain transactional reputation is calculated, whatever that means (PoI was one of the first, but has not been the only one on the list for a long time).
    4. Mixed (from banal PoS/PoW, to morecomplex ones, such as PoS+PoH, where PoH is not really about consensus, but at a fundamental level it will still ultimately not be achievable without this element, which practice has proven; PoW/PoI and many others). 

    It is clear that you can go simpler and listconsensus, as has been done more than once. But if you are really interested in this, I recommend a free lecture by a colleague from The Power, Igor Belousov. 

    Many Bitcoin maximalists seeThe PoW system is an ideal, forgetting that decentralization does not have the right to this: the right to a monopoly. Therefore, for me, the diversity of consensuses and pseudo-consensuses is the same norm as the constant growth of tokens and coins (and now programmable assets) on CoinMarketCap or any other aggregator. 

    Base No. 04. TPS, or transaction speed 

    For me, this is the most merciless and at the same time useless basis for gradation. But since so many people talk about it, we will apply it too. Can be divided into the following categories:

    1. The slowest (up to 10 TPS).
    2. Medium-slow (up to 100 TPS).
    3. Medium (up to 1000 TPS).
    4. Medium-fast (up to 10,000 TPS).
    5. Fast (up to 100,000 TPS).
    6. Ultra-fast (over 100,000 TPS). 

    Despite the fact that once even a fork(Golos) showed 10,000 TPS, and Bitshares gave out a theoretical 100,000; in practice, few people need this. The best example is Solana: a blockchain with a lunch break of 19+ hours, that is, unstable in its operation, no matter how fast, it is not in demand as expensive as Ethereum (or even Bitcoin), which is very reliable in many respects. ;

    However, there is still more to come:The era of sharding is just beginning. And for AI (AI), ultra-fast blockchains are a necessity the same as air, water and sunlight are for you and me. In the meantime, let’s move on to the next basis for differentiation, which has already been discussed in this publication. 

    Other grounds

    I will say a little less on these points, so I decided to combine them. 

    Blockchain, like many other Web3 phenomena, is interoperable by definition. But here you can clarify:

  • there are multi-blockchains (Polkadot, Avalanche, Cosmos, etc);
  • there are cross-chains (EVM-compatible, say);
  • there are single-blockchains (Cardano, etc);
  • there are partially externally interoperable (due to L0-connections). 
  • Or you can gradate blockchains relative to sharding:

  • sharded;
  • not sharded.
  • However, if you go deeper into this aspect(you can start here and continue here), you can easily see: sometimes sharded blockchains are so similar to DAG solutions that the gradation itself will require eternal clarifications and clarifications. Which is not good in scientific (objective) classification. 

    For me, it is extremely important to divide blockchains into the following types:

  • anonymous (Monero, many from the Mimblewimble family);
  • conditionally anonymous (Zcash: only about 5% of all transactions are started with the desired flag);
  • pseudonymous (Bitcoin, Ethereum, etc);
  • non-anonymous (they are still a long way off, and, as we know, there are no closed blockchains). 
  • Speaking of the last remark:There are no closed, corporate, private or similar blockchains. This is something different, but not blockchain. Therefore, I do not adhere to such a gradation and I urge you to protect yourself from it. And all these Unpermissioned public ledgers, Permissioned public ledgers, Permissioned private ledgers, as well as the Consortium blockchain, fully private blockchain, hybrid blockchains are pure simulacra created for the standard control scheme through suppression. Therefore, if you enter “blockchain classification” or even more so “blockchain classification” into the search, the results will not please you: the gradation into open, closed and hybrid blockchains is almost the only one in 6 out of 10 cases. And in 8 out of 10 - the main one. And this is unfortunate if we remember Satoshi’s will. 

    And finally, in practice, it is also important to differentiate by language:

  • solidity;
  • Rust;
  • Go;
  • others. 
  • See the link for the full list, but I am sure that the layouts may change greatly in the coming years. 

    Have I forgotten anything important? Of course! Nakamoto coefficient or level of decentralization:

    1. Decentralized.
    2. Conditionally decentralized.
    3. Slightly decentralized.
    4. Not decentralized. 

    By what criteria to determine?For full and archived nodes, for example. By the number of wallets/accounts. Or by the level of validators/miners and/or other super nodes. And according to other criteria: better - at least ten. And here there will be no unity of opinions, believe me.

    Conditionally additional criteria can also be identified:

  • according to TVL - the entire ecosystem;
  • by the development of this very ecosystem: how many solutions are there and on which blockchain (see examples on DeFiLama);
  • by community: how many developers (git monitoring tools are a great help here); active users; HODLs and so on;
  • for other reasons. 
  • I'll try to make a consolidated list now. 

    Summary classifier of blockchains

    So, here’s what we got for today (the list is, of course, open):

    1. By grassroots division, or zero basis: blockchain and non-blockchain (DAGs; distributed databases without transfer of value, etc).
    2. Gradation by generation: first, second, third and, possibly, fourth.
    3. Division by levels: L0, L1, L2 and to some extent L3.
    4. Differentiation by consensus (architecture).
    5. Based on the Nakamoto coefficient (aka decentralization levels).
    6. In terms of publicity: there are open blockchains, they are federated, and so on (there are no others).
    7. TPS (or high-speed and non-high-speed).
    8. According to the level of anonymity.
    9. By programming languages.
    10. By TVL-blockchain ecosystem.
    11. According to the development of the community.
    12. By the number of dApps. 

    Want to highlight other criteria? Write to the forum or
    chat in telegram, tagging me (@menaskop). 

    Additional materials

    If you liked my research, here are some materials to continue studying:

    1. Classification of blockchain solutions for business.
    2. Summary list of consensuses.
    3. Selection (2022) based on them.
    4. Brief summary of the basis of classification.
    5. Classification of EVM blockchains (written before the zkEVM classification).
    6. About blockchain scaling and how classifications are related to it.
    7. L0 bridges (only archived version available).
    8. Not my approach to grading blockchains.
    9. And my approach: why there are no private blockchains.
    10. About the difference between DLT and blockchain.
    11. Criteria for choosing a DLT solution.

    Instead of a conclusion

    In fact, the beauty of classification isthe same thing as travel: “It’s a dangerous thing, Frodo, to go beyond the threshold: once you step on the road, and if you give free rein to your legs, you don’t know where it will take you.” In the sense that one can always classify more broadly, more deeply, and even simply in more detail.

    For example, the L2 classifier has greatly improved over the yearshas increased: Plasma solutions, “true sidechains”, “pseudo sidechains”, optimistic rollups (and here, too, several completely different approaches can be distinguished), zk-rollups, state channels and various kinds of L2 in exotic networks, and so on. Or the already mentioned zkEVM: gradation is also appropriate.

    Therefore, if you are a crypto enthusiast,IT entrepreneur, netstalker, scientific researcher or anyone else who is interested in the topic, please let me know through the forum or telegram channel and perhaps the article will have a continuation, but for now that’s all