head of research @l2beat, native rollups @ethereum โ scaling is caring, paperclip maxi. | name: gwei.domains
- Tweets 8,282
- Following 397
- Followers 8,060
- Likes 175,902
since i've seen a lot of bad takes regarding L2s and their censorship resistance, here's how i bypassed @soneium's sequencer and bought a banned token with a forced tx through L1 ๐งต
.@soneium rugged users >$100k in ETH on launch day
I've never seen anything like it
Want to be a permissioned chain? fine. Instead of whitelisting developers, they chose to instead freeze contracts (effectively rugging users)
Amazing launch!
105
211
1,178
333,379
Congrats to @zksync ๐ฅณ the first project to process more txs in a month than Ethereum itself
Despite the Inscriptions craze, no project managed to surpass Ethereum's 30 days transactions count
who'll be the first? ๐
l2beat.com/scaling/activity?โฆ
45
988
704
168,084
- @blast 3/5 msig ($1.45B)
- @0xMantle 6/13 msig ($1.44B)
- @LineaBuild 4/6 msig ($849M)
- @Starknet 2/5 msig ($676M)
- @MetisL2 4/9 msig ($303M)
- @fraxfinance 3/5 msig ($170M)
- @taikoxyz 3/4 msig ($100M)
- @loopringorg 4/6 msig ($45M)
- @KintoXYZ 3/5 msig ($36M)
PSA: Radiant finance is being exploited live.
Withdraw your fund and revoke approvals.
Seems like their 3/11 multisig got compromised, ownership transferred and then rekt. Another key management failure.
77
148
810
223,606
if youโre interested in building @l2beat for DeFi, stables, socials, or anything really, DM me!
Making Ethereum alignment legible
vitalik.eth.limo/general/202โฆ
189
132
612
360,392
> ethereum should do this ethereum should do that
bro why donโt you do it. some of you could really spend more time on github than yapping here. you can just do things.
23
38
449
28,266
ethereum blobs have hit capacity for a full hour for the first time
demand is up only and im very much looking forward to see more capacity after fusaka with peerdas ๐ซก
30
68
488
42,517
we just finished reviewing the project, so here's a short thread on @Lighter_xyz ๐งต
had a quick look at @Lighter_xyz contracts and first impression is very good! very exciting. they feel a little like a mix of zksync and the good old dydx. weโll hopefully have the full review on @l2beat next week. the advantages of being a rollup vs HL are quite evident here :)
52
60
439
130,182
had a quick look at @Lighter_xyz contracts and first impression is very good! very exciting. they feel a little like a mix of zksync and the good old dydx. weโll hopefully have the full review on @l2beat next week. the advantages of being a rollup vs HL are quite evident here :)
28
25
438
106,478
โdedicated blockspaceโ here literally means posting DA to a AWS S3 bucket.
for the sake of our industry, please use @celestia or @AvailProject instead.
Base Appchains make Base more powerful
These L3s (powered by the op-enclave framework) provide apps with dedicated blockspace + custom gas tokens and permissions, enterprise-grade support, builder friendly pricing, and seamless integration with Base builder tools
49
49
391
208,298
given the requests, @HyperliquidX has been added to @l2beat. TLDR: the bridge is controlled by 3/4 permissioned โhotโ validators, that can challenged by 3/4 permissioned โcoldโ validators within ~3 mins. Some other actors can finalize withdrawals and others pause the bridge. fin
Hm, if the code is not public, but one can see the flow going from arbitrum, and itโs the only bridge, is it not enshrined?
If an L2 goes live, but L2Beat never heard of it, did it even go live in the first place ๐ค
33
37
309
92,476
then i wait a couple of minutes and... the L2 tx magically appears. you can even see the buy on @dexscreener. and as it shows, many people are doing the same. so yes, L2s are censorship resistant, as they are supposed to be. thanks for playing
tx: soneium.blockscout.com/tx/0xโฆ
16
18
300
11,158
. @Lighter_xyz, a ZK rollup using Ethereum blobs, is doing 4k+ TPS posting just ~100MB per day, all thanks to state diffs and the app-specific architecture.
for comparison:
- @arbitrum: 37 TPS @ 436MiB/d
- @base: 155 TPS @ 2.11GiB/d
- @soneium: 39 TPS @ 241 MiB/d
Just in from L2BEAT - 20x growth in rollup UOPS (user operations per second)๐
All of that because we added @Lighter_xyz. Congrats @vnovakovski and the whole team ๐ซก
70
46
293
43,791
here i say it. @0xPolygon PoS was so ahead of its time. it was so underappreciated because we all were too ambitious thinking we could have state validation 4 years ago. Polygon PoS tech is far more secure than most โrollupsโ launching today and many are reinventing their tech.
thatโs exactly how polygon pos has been working for the past 4 years. hereโs their stake manager etherscan.io/address/0x5e3Efโฆ
27
35
288
55,729
so that you understand how bad crypto marketing is nowadays: @AbstractChain is not using @celestia. they do not post to Celestia. theyโre literally posting Ethereum blobs.
@AbstractChain *will* *maybe* use Celestia in a few months. why is it so difficult to use the future tense?
107
13
277
59,586
one of the 5 whitelisted "lockers" on @HyperliquidX just voted to pause the bridge. given the locker threshold of 2, another vote is needed.
does anyone know what's going on? either they'll pause the bridge soon or they want to effectively lower the threshold to 1
36
32
256
121,866
soo i spent some more time reading @hibachi_xyz contracts to better understand their "if we ever rug (we won't), you can still get your funds" claim
tldr: they are lying and the EOA upgrader is not the only issue ๐
๐งต
Introducing the Wynn Upgrade
A privacy-first architecture for Hibachi, built with @celestia underneath and @SuccinctLabs
A high-performance CLOB with ZK proofs, the first to publish encrypted state on Celestia Private DA
if we ever rug (we wonโt), you can still get your funds out
13
18
258
39,022
. @arbitrum often switches back to calldata because blobs are now too expensive lol
blobs min fee pricing drama has always been overblown. it literally only took one application in two different instances (@Scroll_ZKP airdrop Oct 22, @flippyflop_gg on @Starknet yesterday) to get to significant blob prices
stop blaming L1 and just get some activity duh
21
10
246
43,975
just deployed inclusion.watch, a website to easily check how much you have to wait to include a blacklisted tx with the current % of censoring nodes or a custom amount
a weekend project by @emilianobonassi and me ๐๏ธ
13
57
220
. @Taikoxyz currently pays an insane onchain cost of almost $1M per **week** even with blobs, and with only ~358K txs, or around ~$2.50 per L2 tx
for comparison, every project except Scroll pays <$60K per week
19
19
213
79,628
congrats @Mantle_Official & @eigencloud, but this just false. EigenDA currently has the *least* economic security ($0) compared to other DA layers.
moreover, the ETH stake cannot be slashed even when slashing will be activated, so it would still have less security than @celestia
โ Ethereum-backed security
Mantle is building the future of onchain finance for enterprises, which means they need enterprise-grade security.
EigenDA has more economic security than any other DA provider, with 4.3M restaked ETH and 391.6M restaked EIGEN.
12
15
235
51,807
Only 30 days left before the Great Recategorization goes live on @l2beat. All projects without a proper proof system, or without a proper DA bridge, will be not classified as Rollups or Validiums/Optimiums anymore, but "Others". ๐งต
16
25
223
24,455
apparently @real_rwa has fully halted block production and state updates for their Orbit stack chain without any announcement. $500K in stETH & USDC are still locked in the chain with no way to force exits
exit windows are not a meme ๐
33
25
218
34,635
TLDR of @EclipseFND architecture: users deposit into a 2/4 multisig โ no other functionality.
14
15
213
60,335
random list of personal beliefs:
1/. L2s have always been a short/medium term solution to L1 scalability issues
2/. the long term plan has always been to scale L1 via zkfication, DAS (for L1) and statelessness
3/. the rollup centric roadmap is mainly a psyop to outsource the development of the necessary tech (ZK) to scale L1
4/. long term, thereโs no place for vanilla (i.e. just-EVM-equivalent) rollups
5/. the only rollups that might make sense long term are the non-EVM equivalent ones
6/. there is a non-zero chance that ethereum also decides to enshrine risc-v or similar in some form and natively move beyond EVM
32
14
227
31,214
whatโs going on on @base? in 20 days the number of ETH deposited went from 1.8M to 800K (-55%) and itโs still dropping
37
11
209
44,013
The Polygon vs ZKsync drama is so tiring. Please copy each other. I beg you. After reading the code of all major systems, I guarantee that we would have saved millions of hours if teams just looked at each other code instead of reinventing the same thing over and over but worse
13
13
182
21,361
popular opinion: all EVM L2s should have a website that clearly describes all code diffs wrt L1.
op-geth.optimism.io/
12
13
185
10,455
happy to announce that iโve started deeply researching native rollups to make them a reality, and that iโll be speaking about it at the next @bankless Summit II in Buenos Aires, alongside other amazing speakers :)
if you wanna chat about native rollups, DMs are open!
31
15
180
37,460
Differences and similarities between @optimismFND Bedrock and @arbitrum Nitro deposits:
scalability.guide/posts/bedrโฆ
8
34
157
Happy to share the progress done in the past month on native rollups! A new shiny website has been created to collect the spec draft and design ideas, with the goal of providing a starting point for feedback from community members and L2 devs.
link at the end of the thread ๐งต๐
18
25
153
21,974
The first L2 to migrate to Celestia modular DA...
...in the least secure way possible, let me elaborate ๐งต
6
23
144
103,892
thatโs exactly how polygon pos has been working for the past 4 years. hereโs their stake manager etherscan.io/address/0x5e3Efโฆ
8
6
150
53,705
i believe @cartesiprojectโs Dave to be the most advanced fraud proof system out there, outcompeting both @arbitrumโs BoLD and @Optimismโs OPFP
i highly recommend reading the research post. iโd love to see OP and Orbit stack chains integrating it!
Excited to share a new post on ethresearch about the Dave Fraud Proof Algorithm.
Read here: ethresear.ch/t/the-dave-frauโฆ
14
33
146
18,383
Unfortunately this is the case. Almost all L2s are still centralized and can pull the plug and turn off the chain at any moment.
All, except @MetisL2, the real #decentralized #Layer2.
14
4
141
20,585
vibe-coded a simple website that shows a live feed of ethereum's gas limit changes to track the progress towards 45M
(link in comments)
19
14
138
14,580
agreed. luckily theyโre very serious people, iโve already spent several weeks helping them designing stage 1, which is very close to be deployed. bullish on @base
4
8
134
8,962
hey @AcrossProtocol, any plan to remove the 3/5 multisig? ๐ cuz if not, itโs time to get serious and build a permissionless alternative.
6
6
132
13,237
i'm considering aggregating internal @l2beat in-depth knowledge in the form of code walkthroughs, would anyone find it useful? these would be written exclusively based on prod code with a focus on security
example: @arbitrum's new buffered forced txs: lucadonnoh.github.io/l2-codeโฆ
30
3
134
17,553
two things that i donโt understand:
1. how a bug in the implementation of an opcode of the fault proof VM can be considered low severity
2. how it is considered acceptable that the fault proof system is purposely outside the scope of external audits
Today, @OPLabsPBC posted an upgrade proposal detailing findings from a recent series of community-driven audits on the Fault Proof System, including the plan to fix the bugs identified as part of the audits.
gov.optimism.io/t/upgrade-prโฆ
5
16
130
20,588
in the last 3 months, @base only paid $0.0041 in total to publish the data of ~400 *million* transactions onchain. or in other words, $0.00000000001 per tx on avg
17
5
127
25,457
soo what happens to @HyperliquidX (a self-proclaimed L1) if Arbitrum has a 24h reorg? would it reorg as well? asking for a friend
18
4
123
24,622
how can we properly scale blobs if top builders are censoring, artificially pushing down max throughput and pricing? cc @ralexstokes @fradamt
h/t @bertcmiller for the dashboard
28
9
124
39,215
congrats @Starknet both for joining the Stage 1 squad and the top 5 leaderboard, which now only contains Stage 1 projects.
wen Stage 1 for all top 10 projects? @zksync @worldnetwork @LineaBuild (...@blast)
Another ZK Rollup joins the Stage 1 squad - Starknet! ๐
With this update, the top 5 rollups by TVS on L2BEAT are now Stage 1!
What changed? ๐ค๐
7
9
123
7,928
the cool thing about L2s is that you can have the exact same properties as if you were an L1 *plus* additional stronger ones from the base layer
I fully disagree with the finality argument. If you're going to have the same decentralized staked capital as Ethereum then yes, otherwise your finality is just trust me bro finality. If it's going to be trusted anyway you may as well trust it on an L2 with 1/2
27
12
119
12,376
remainder to all alt-DA L2s that their TVL should never go over the economic security provided by the external DA, otherwise it is profitable to confirm an unavailable block and steal everything. As an example, the *sum* of TVL of all projects using Celestia shouldnโt be over $2B
17
9
124
14,402
2025 will be the year that reveals whoโs truly willing to move beyond stage 0 and whoโs just full of bullshit marketing. no more excuses
12
15
119
11,156
risk assessment aside, seeing a zk op stack on top of agglayer using a shared bridge with polygon cdk chains and potentially any stack with accounting proofs and an atomic interop protocol is just insane
congrats @katana @optimism @agglayer @succinctlabs @0xpolygon @conduitxyz
wake up anon, a new ZK Rollup has dropped @katana
(yes, still Stage 0, but WIP to make it)
l2beat.com/scaling/projects/โฆ
34
18
118
14,682
. @Starknet is the first rollup to ever hit >100 daily avg TPS on prod. Congrats
Starknet #1 ๐๐บ
- max 24h sustained TPS ever recorded for an L2: 127 TPS
- max peak TPS ever recorded on mainnet for an L2: 857 TPS
7
12
120
10,863
we also just released our *smart* smart contract differ, which compares flattened sources, can ignore comments and reorgs sections to prevent phantom diffs
as an example, here's @katana adding one line (compared to standard op stack) to censor forced transactions ๐
link below
one of the most annoying thing of reviewing projects through etherscan is when contracts import other contracts just to use them as an interface, and etherscan gives you the full code as if any of it is actually relevant, misleading you
worry not, we have a fix ๐
20
7
114
12,334
fortunately, @soneium is a standard OP stack chain, meaning that it implements a forced txs mechanism on L1 to bypass sequencer censorship. so let's try to use it.
2
7
109
11,575
ah yeah, this might get lost in the narratives, but one of the main reasons polygon pos was not being listed as an L2 is because they claimed they were an L2 because of the plasma bridge, but if you looked at the onchain contract there was literally nothing but a TODO comment. all their docs were extremely misleading and straight up lying
6
4
111
23,484
if you expect a forced tx mechanism to bypass censoring sequencers on a good L2, then you should also expect a forced tx mechanism to bypass censoring builders on a good L1.
FOCIL For Glamsterdam
It's nice to see more people advocating for FOCIL inclusion in Glamsterdam, knowing there is a good chance it becomes too difficult to include in subsequent forks.
For the record, I think it would be a pretty big mistake to gamble the censorship resistance properties of the network for the sake of shipping a fork two months earlier.
FOCIL For Glamsterdam
22
11
104
8,967
Ma'am, your project is literally upgradable by a 4/9 multisig. Stop spreading BS to mislead your users.
Nice picture! Did you draw it yourself?
Or did you copy-paste from someone?
Is this source reliable?
Did it warn you that any Layer 2 at the moment can pull the plug and turn off their chain?
That almost any #Layer2 can be switched off at any moment on the request of any party like founders, court order, police, SEC, and even AWS/MS Asure.
What information does this source have that almost ALL of the #Layer2 are #Centralized in this way?
Thank you for the drawing!โค๏ธ
12
2
108
24,848
TLDR on @redstonexyz: OP stack chain with offchain DA with no attestations, but with DA challenges. They claim this mechanism is a "Plasma" implying that the system is as secure as a rollup, but it's trivial for the sequencer to outspend honest challengers and steal funds.
7
7
104
33,049
then i go to etherscan (on L1!), to their OptimismPortal (contract used to send L1->L2 msgs), and craft a depositTransaction with such calldata. i want to call the exchange contract on L2 (0xE1A...3Ab) with the appropriate value to swap and the calldata above. check the tx yourself here: etherscan.io/tx/0x8a8748acb5โฆ
8
3
98
10,254
I believe it's time to seriously discuss Ethereum's strong censorship resistance guarantees and the level of security we want optimistic rollups to provide for their users.
ethresear.ch/t/optimistic-roโฆ
9
10
98
13,345
itโs insane how many people are not used to read onchain contracts, even very technical people. on the other hand, tooling for it is terrible.
donโt trust verify is not a meme. running a node doesnโt verify enough. having an @l2beat for everything doesnโt scale. tooling does
15
4
100
7,465
it is only posting state diffs and most txs are done to open or cancel orders which do not touch state so require zero data to be posted
3
3
105
5,805
Just published the english version of my thesis about @optimismFND and @StarkWareLtd:
github.com/lucadonnoh/lucadoโฆ
here's the main topics discussed: ๐งต๐ (1/4)
6
19
97
daily reminder: if you think the EF isnโt focusing enough on something, you can literally do it yourself. you can just do things.
3
7
96
5,237
given the positive feedback, we'll start writing L2 code walkthroughs under docs.l2beat.com, with the goal of lowering the bar to self-verification for everyone.
first pages are about Arbitrum's new proof system, and here are some nice illustrations i've drawn for it ๐งต
i'm considering aggregating internal @l2beat in-depth knowledge in the form of code walkthroughs, would anyone find it useful? these would be written exclusively based on prod code with a focus on security
example: @arbitrum's new buffered forced txs: lucadonnoh.github.io/l2-codeโฆ
6
9
96
10,945
. @taikoxyz had 84 upgrades in total since launch, with an average of one upgrade every two days
11
4
95
13,537
am i the only one who thinks that these new L1s do not mean anything for anyone? weโll just see the effects when they will want to interoperate with one another and will need to decide where this should happen
23
7
94
3,922
while the bitcoin community spends its time arguing whether they should like NFTs or not, @ethereum is building a post-quantum stack so good and flexible that all blockchains and bitcoin itself will want to adopt it.
youโre not bullish enough
lean ethereum thread:
PQ workshop (Cambridge, UK) started hours ago. this thread will be updated with presentation recordings (and slides) as they occur
๐งต
12
6
98
8,509
spot the backdoor, @l2beat edition
@Scroll_ZKP's "importGenesisBatch" function vs @MorphL2, a Scroll fork
7
1
94
12,711
- stage 1 and a credible path to stage 2
- composability with other chains
- credible path to zk
- L2 light client
5
1
94
6,367
the vast majority of alt-L1s really underestimate the psychological impact of being able to autonomously and trustlessly verify a blockchain state at all times on any device.
we need a slur for blockchains that rely on trusted rpcs
18
5
92
4,376
fusaka will allow to increase blob capacity, raising rollup max throughput from 380 to 817 TPS by the end of the year. if everything goes well the short term goal is to reach ~3K TPS, with a long term target of 5-6M TPS (1 GB/s). of that, ~10K TPS will be consumed by L1 alone
We are <24hrs away from the first testnet to transition to Fusaka. It sounds to me that a Fusaka release this year is quite likely .
11
13
93
5,954
I'm speaking at @EFdevcon!
See you in Bangkok, November 12-15
devcon.org/en/sea/schedule/Uโฆ
6
5
92
4,726
Over 50% of @Optimism TVL is comprised of $OP tokens, whereas $ARB constitutes only ~20% of @Arbitrum TVL
l2beat.com/scaling/tvl
9
15
82
27,960
gm, @aztecnetwork's old.zk.money has been resurrected from the dead (i.e. the archived page) as it is still technically possible to advance the rollup state, and people shouldn't forget that we had a private stage 2 payments rollup on prod since 2021.
4
2
91
4,427
A fundamental trilemma seems to affect optimistic rollups, making it impossible to design safe systems with fixed delays that are simultaneously protected by a 1-of-N honest assumption - we analyzed it in our latest article! ๐
medium.com/l2beat/fraud-prooโฆ
5
7
84
15,379
just read the docs, and there is a *huge* opportunity to compete and do better with actual objective slasing rather than intersubjective, which works with any token (e.g. ETH) rather than just EIGEN
verifiable AI is a perfect use case for fraud proof systems like cartesiโs dave
The ecosystem is taking notice! ๐ฅฐ
EigenAI + EigenCompute are live, bringing verifiability to AI and compute.
Hereโs what the builders, founders, and thinkers are saying ๐
24
5
89
10,214
very much looking forward to see some prividiums going live and to check what guarantees they can provide vs tradfi. this has the opportunity to really popularize zk tech
The financial industry requires private, incorruptible systems connected as one global network.
Citi, Deutsche Bank, Mastercard, and 30+ top global institutions joined us to explore the power of Prividiums.
Unveiling The Prividium Breakthrough Initiative.
20
8
86
5,758
itโs both true that stage 2 rollups can keep significant parts of their system centralized (sequencing, delayed upgrades) and that they need to decentralize others (tx inclusion, proving). saying that a system is centralized because some components are is extremely reductive.
13
9
86
9,618
the world computer canโt run DOOM. i repeat, the world computer cannot run DOOM. whoโs fixing this
30
4
87
17,937
gm, just published a proposal to decouple escrows from the stage assessment and provide instead more nuanced escrow-specific assessments. feedback is welcome! this affects projects like @LineaBuild and their upcoming native yield bridge. forum.l2beat.com/t/stages-upโฆ
23
14
88
22,730
lots of people think at @l2beat we just write html pages, while most of the time is spent building tools to better inspect and monitor projects. we believe these tools can be useful to everyone and weโre working to make them more publicly accessible. let us know what you think!
Itโs time to DISCOver ๐ชฉย every secret of L2 architecture!
Our in-house evaluation tool is now open to everyone - shine a light on contract graphs, source code and key interactions in a UI built specifically for the job.
3
4
89
2,696
if @l2beat had a documentation quality metric @0xPolygon zkEVM would be first
8
8
82
12,758
wow! super impressive. it has been interesting to see how non-superchain op stack chains that want proofs prefer zk proofs over economically uncertain fraud proofs. now superchain chains want to join the zk gang too ๐
6/ After these investments, we successfully delivered end-to-end proving on a parallel bridge contract on World Chain's mainnet: etherscan.io/address/0xa244dโฆ
17
2
83
7,010
i carefully craft the calldata to call the swap on the exchange on the L2 with the appropriate swap amount, token address and deadline. in particular i'm trying to buy at least 102.39 $aiko tokens (0x5Ee...edd) with ETH (0x420...006).
4
2
80
8,442
inb4 paradigm deploys a single sequenced L2 on tempo as their main offering to overcome the challenges of a โdecentralizedโ validator set
5
6
89
6,089
โno one cares about canonical bridges security because most L2 users use external bridges and permissioned apps anywayโ
The idea that users donโt care about bridge security is wrong
Most just express this by not bridging at all
The delta in my comfort holding BTC on Base vs Hyperliquid is small
The delta in my comfort holding BTC on Base vs Coinbase Prime or self-custody on Bitcoin is huge
15
3
83
8,798
apparently @BuildOnBase deleted 3 output roots using the centralized challenger a couple of months ago
6
4
84
14,784
fwiw, presumably because of the exploit, ZKsync Era has added an unverified "TransactionFilterer" contract that can censor forced transactions from L1, without any onchain approval from the Security Council. This is one of the main reason ZKsync Era is currently not Stage 1
Update: further investigation has confirmed yesterdayโs findings that the compromise was contained to the airdrop distribution contracts and no additional ZK tokens can be minted from this contract.
User funds are secure and were never at risk. The ZKsync protocol, ZK token contract, all three governance contracts, and all active Token Program capped minters are not impacted by this incident.
The investigation and recovery efforts are on-going. We will share material updates as we have them, and will have a comprehensive incident report to share with the community once the investigation is fully completed.
2
2
83
9,717