lots of discussions lately on building on L1s directly or building on current L2s/rollups/sidechains/subnets/parachains/fancynames. some thoughts


Why consider using L2s/rollups/sidechains/subnets/parachains/fancynames? because you believe you have a problem with your L1 that needs a solution.


For the most part, the L1 problem that these solutions solve is a real (or perceived) problem of scale/perf/cost. and, these solutions have done a reasonably good job alleviating these problems.


But, any solution to a problem brings new problems. The question is always, would you rather have your old problems or your new problems?


What are the new problems these particular solutions bring? They fragment ecosystems (and sometimes security), they introduce technical and security complexity which makes reasoning hard, they can break previous invariants, and they can make user experience confusing.


Unfortunately, we’ve now seen this play out in the broader blockchain ecosystem in the past few days. Existing contracts designed for an L1 without consideration for an L2, were used on an L2 and a nuance of the added complexity was missed resulting in a loss of funds.


To be clear, I believe this is a risk of the model, not of the various implementations or teams etc - I’ve been in similar situations in my career and my heart goes out to them.


What is the alternative? There’s no reason to bring in the problems above if you have a performant L1. It provides a single ecosystem, single security model, single user experience, etc.

@algorand is an example.


Algorand handles ~1100 TPS today, 6k in about a month, 10k soon after. 100% uptime with no degradation or stalling since genesis 3 years ago. Avg costs for all txs ~.001 Algos (currently $.0004). Final in 4.4s today, under 4s in about a month and ~2.5s soon after.

アルゴランドは現在、最大1,100TPS(毎秒トランザクション数)を処理し、その数は約1ヶ月後に6,000、その後すぐに10,000になります。3年前のGenesis以来、劣化も停止もなく稼働率100%です。全Txの平均コストは、.001 Algos (現在価値では$.0004)です。ファイナリティ(決済最終確定)は現在4.4秒、約1ヶ月後に4秒以下、その後すぐに2.5秒になります。

Algorand is actually final in 4.4s. not maybe final, pretty final, mostly final, kind of final, final here but not there, etc.


Algorand’s growing and thriving ecosystem is nodding their heads right now, but others might say, “yes, sounds amazing, but this brings a new problem too. my users are [there] and not currently on Algorand”


To which I say your users are there for your value prop, not the underlying blockchain. They care about the chain your solution is built on for only one reason: the user experience it creates.


How long does it take to bridge your asset? Finality of the chain. How quickly will fees escalate? Capacity of the chain. How much downtime will your dapp have? Historical uptime of the chain. How responsive is the dapp? latency of the chain. and on and on

あなたの資産をブリッジするのにかかる時間は? チェーンの最終性の問題です。料金が上昇する速さは? チェーンの容量の問題です。あなたのアプリケーションがダウンする時間は? チェーンの過去の稼働時間の問題です。アプリケーションの応答性は? チェーンのレイテンシーですね。などなどたくさんあります。

Do the work to understand how what you’re building *on* impacts the experience of you’re building; Your users should be laser focused on your value prop, not your technology stack.


Finally, am I saying that current L2s/rollups/sidechains/subnets/parachains/fancynames are bad? definitely not - when you have problems you need solutions. but, you must ask yourself, do you like the new problems they bring more than the current problems you have?