data at @paradigm github.com/sslivkoff
- Tweets 3,551
- Following 2,226
- Followers 16,190
- Likes 19,005
Pinned Tweet
new research from @paradigm: How to Raise the Gas Limit ⛽⛽
lots of people want to raise the gas limit, but so far most of the discourse has lacked detailed quantitative evidence
that ends today. we are making the scientific case for scaling Ethereum using ultra-high-res data
104
331
1,385
404,898
PSA if you haven't used python in the past 5 years it's a completely new language now
- uv solved the package management problems entirely
- mypy type checks have become actually useful
- cpython is now much faster
- polars + fastapi + ruff replaced pandas + flask + black
180
302
6,748
488,480
major narrative violation:
hyperliquid ADL actually INCREASED PnL for the vast majority of shorts last week
many ppl are sour that ADL closed shorts right before prices nuked. but the data reveals that most ADL hit near the price bottom, locking in near-optimal profits
122
218
1,331
348,866
~68% of Base chain state has been taken over by a single protocol of questionable utility
thanks to cheap fees + storage mispricing, a tiny number of users have monopolized the majority of network resources
this has implications for every cheap fee blockchain. some thoughts 👇
129
205
1,240
613,180
today I'm releasing spice 🌶️ my python client for @DuneAnalytics
this is the easiest + fastest way to export data out of Dune
you can now extract any query's data with a single python call
56
112
785
68,925
🟩📊🌀excited to announce the launch of the Paradigm Data Portal 🟩📊🌀
TLDR: we are assembling a bunch of public crypto datasets for researchers and tool builders
short thread
66
188
719
178,896
juicy 🧃 crypto 🧮 data 📊 thread 🧵
ctc 0.3.0 is out today
this is a large release with lots of new features
i want to share some of them with you in an oversized megathread (cc @adamcochran)
many examples and screenshots incoming...
1 / 48
25
139
636
excited to share that I recently joined @Paradigm as their first data associate 🧮 📊
70
51
587
EPIC ACTUAL DATA on how much the gas limit can be raised
initial results suggest that all 5 node clients can already support a substantially higher gas limit
mainnet is currently 1.25MGas/s...the 5 node clients are supporting 80x - 280x times that. still need to comb through edge cases but looking very promising
and now that we are collecting **actual data**, instead of flying blind and performing vibes-based optimization, the client teams can precisely identify bottlenecks and raise their performance even higher
massive praise to Kamil and the nethermind team
Ethereum L1 PerfNet up and running 🔥
5 EL clients
4 CL clients
4 GigaGas #gaslimit
4 seconds slots
Full blocks already there 📈
High txs blocks (20k txs) 📈
Tough conditions but chain keeps going 🦾
Several 🪲 reported to ELs to make results even better
#ScaleL1 #PumpTheGas
55
62
505
109,611
amazing news. ethereum is about to lower the storage burden of nodes by 80%, via expiring history
this will switch from the dark blue line to the light blue line in this chart
this will free up many years of runway for state growth
History-expiry update:
More than 80% of Ethereum full node disk space is occupied by history, so data not needed to validate new blocks.
With the latest 1.31.* releases, @nethermind is fully ready to drop pre-merge history via ERA files.
Dropping history requires cross-client consensus, and all teams have agreed to remove pre-merge history from full nodes on 1st May. History expiry is desired for further scaling.
Once activated, your full node will require ~320 GB of pre-merge blocks and receipts history. A full Nethermind execution node without dropping pre-merge history today takes around 700 GB.
If we drop all blocks and all receipts and keep only what is needed to verify the latest block, it would be less than 200 GB for full node.
Nethermind team keeps working on history expiry (portal-network integration, rolling history expiry, ERA files, EIP-7801).
Node without pre-merge history 👇
26
54
497
56,086
probably no one other than VC's cares about this, but at paradigm I'll be changing roles from data associate to research partner
expect more of the same from me. work ain't done until the world is drowning in open source tools and data
Excited to announce @notnotstorm will be stepping up as Research Partner at @paradigm!
His OSS tools like Cryo and Spice power some of the industry's biggest data platforms, and his research work has helped us think analytically about the future of Ethereum.
Congrats Dr Storm!
59
480
30,304
there’s a new EIP for scaling ethereum. it:
- increases blockspace by FIFTY percent
- does NOT increase the gas limit or DoS risk
- is TRIVIAL to implement
details 👇
30
63
451
108,620
still stuck using csv? well there’s a new tool for anyone that enjoys rust, parquet, or crypto data…
❄️🧊cryo🧊❄️
cryo is the easiest way to extract crypto data to parquet (and other formats)
details 👇👇
38
63
422
132,260
excited to release some new data tooling today
🌊🌊 flood 🌊🌊 is a load testing tool for benchmarking EVM nodes over RPC
paradigm.xyz/2023/06/flood
we believe flood will be a huge unlock for optimizing your crypto infra
let me show you how 🔑🔓👇
10
66
358
98,514
🌋🔥 very exciting breakthrough 🔥🌋
if you are doing rpc data extraction in python, cryo will make your code faster
it can do 10,000+ requests per second in rust and return results as a python dataframe. up to 16x faster than a highly-optimized native python client
details 👇
20
46
356
67,623
⌨️🖥️⌨️
new set of tools for performing eth_call's from the command line
❌🖱️❌
[a thread with many screenshots and examples]
1 / 16
11
55
342
big update: EOF is probably dead due to lack of rough consensus being reached on the core dev call
this is a **MASSIVE** milestone. it is symbolic of ethereum turning a new page and evolving toward maximal consideration of User Impact
there will be growing pains. some explicit statements heard on the call:
1. "On this issue we should only listen to core devs, not users"
2. "We should ship EOF just to respect the process"
3. many different versions of sunk cost fallacy
thankfully these voices did not win the debate as they did in the past
everyone should be excited to see what this new chapter of Ethereum will bring
24
20
340
28,829
this is my understanding/prediction for the ethereum scaling roadmap
continual improvements to hardware, software, and protocols will support gas+blob increases without compromising network health
Imho we should aim (grain of salt etc...) to at least 3-4x the pectra blob throughput in peerdas, so 6-8x from today, and maybe more. Two things we know:
1. Most of the network can handle *a lot* more than today
2. Peerdas requires the weakest nodes to use 8x less bandwidth
24
33
330
134,596
how many solana data analysts does it take to change a lightbulb?
no one knows and it would be too difficult to check. so everyone else will just remain in the dark
How many L2s does it take to change a lightbulb?
We should be able to work the answer out but first we have to agree on a shared interoperability standard so that the rollups can communicate with each other...
15
12
316
51,365
FYI to all indexooors / analyzooors:
not enough people know about the power of parquet
over the past few months I've been converting everything I do to a parquet-based architecture and it's a complete game changer
20
25
306
63,796
you can query each address's nonce per block. so you can use this with something like a binary search to efficiently gather an complete transaction history of an address
ctc uses this technique. for example for vitalik.eth takes about 20 secs to find 811 tx's with alchemy node
is going block by block the only way to get every transaction from an address? sitting here working through 14 million blocks right now 💀
5
33
282
excited to release a new cli tool today: tbl
the goal = make it effortless to read, edit, and manage parquet datasets
tbl is now the easiest way to do these things from the command line
let me tell you about its features
11
31
281
28,318
lots of ethereum lovers are concerned with making the world a better place
but few of them take this fact to heart:
ethereum will not make the world a better place if it fades into irrelevance
28
16
272
17,082
𝓐dvent of 𝓒rypto 𝓓ata, DAY 0️⃣1️⃣
excited to share a new tool called **etop**
just like how htop + btop monitor the state of your local computer, etop monitors the state of The Ethereum World Computer
etop enables checking the chain with unprecedented visibility + intuition
10
42
241
41,457
how to raise the gas limit, part 2 ⛽⛽
1. the goal = raise the gas limit
2. the approach = use lots of data
3. the focus today = 𝐇𝐈𝐒𝐓𝐎𝐑𝐘 𝐆𝐑𝐎𝐖𝐓𝐇
very excited to release our latest research on the LARGEST + LEAST DISCUSSED factor holding back Ethereum's gas limit
17
50
242
55,586
everyone is forming an opinion about glamsterdam EIP’s, but almost no one understands the technical implications of each proposal
here is a direct comparison of the headliner EIP’s along the dimensions that matter:
23
48
239
27,328
super excited to release mesc 0.2.0 today
mesc will make it **effortless** to operate your tools in a multichain world
👇 mini thread on all the new features 👇
14
26
232
38,623
a raspberry pi barely matches the hardware capabilities of a desktop pc from 15+ years ago
hobbling ethereum to work on rpi's is the wrong point in tradeoff space
you can either build The Future of Finance, or you can build a network that runs on toasters. can't have both
20
9
222
25,905
📊 📊 calling all crypto data enthusiasts 📊 📊
excited to release **ctc** today
ctc is an EVM tool specifically designed for historical data collection and analysis
github.com/fei-protocol/chec…
🧵⬇️ [1/7]
15
32
215
eth gas limit roadmap:
2021-Q3: 30M
2025-Q1: 36M
2025-Q3: 45M (today)
2026-Q1: 100M (modexp repricing)
2026-Q2: 300M (ePBS)
15
24
217
25,084
over the past few weeks I’ve added 11 new trace datasets to cryo, bringing the total to 17
this required a ton of research, so I gathered all the info I found into the
🌌🪐Ultimate EVM Tracing Reference🪐🌌
a quick overview...
7
38
211
33,208
a year ago, the rise of Cursor made me feel a sense of loss. it meant I'd have to give up vim and all my favorite CLI tools if I wanted to keep up with the AI frontier
but instead the opposite happened. CLI's are now stronger than ever
1. the best agentic frameworks (e.g. claude code) target the CLI because of the diversity of tools it has access to
2. almost any question related to the install/configure/debug of CLI tools can now be one‑shot by LLM's. hours of googling no longer needed, learning curve is now flat
3. copilot works really well in vim
4. AI-powered CLI tools like @simonw's llm or @willmcgugan's toad give new ways to interact with models directly from the CLI
5. building CLI tools is easier and more expressive than building plugins for Cursor. so most of the future AI tooling innovation will probably happen on CLI's
8
9
222
21,844
one recent vibe shift is that we have received royal scaling decrees from the high chancellors of ethereum:
- @VitalikButerin proposes 48 blobs/block by EoY
- @tkstanczak proposes 100M gas/block by EoY
it shall be done my lords 🫡🫡
Have heard that from a few folks that there is already a vibe-shift in the EF.
Bullish the @tkstanczak and @hwwonx arc.
9
16
210
34,524
⚠️📊⚠️ earlier this week was a major milestone for crypto data tooling
the birth of a new, open-source, multi-chain, collaborative database of contract sourcecode + ABI’s
behold: the VERIFIER ALLIANCE, @verifalliance
18
27
202
27,585
each blockchain selects technical + cultural tradeoffs that make it better suited to some use cases and not others
the reason I'm an ethereum fan = it selected better tradeoffs than other blockchains for building The Future of Finance. that's the usecase I find most interesting
16
14
206
10,989
the future of sovereign crypto data = you have a local copy of the whole chain on your laptop. it support very fast queries and doesnt take up much space because it's natively compressed. made possible by modern data eng best practices
polars and parquet is just crazy. this query finds all vyper 0.3.4+ contracts and extracts the compiler versions. done in just 6.66s while running locally. hope @notnotstorm updates his datasets soon.
6
12
196
27,675
I love uv so much. setting up a new laptop right now. *zero* use of pip. everything is completely squeaky clean
for the first time ever, my jupyter env is completely reproducible, not fucked up or weird at all
I can finally treat python envs as cattle instead of pets
11
5
202
10,453
rust-based tooling fixed these python issues for the first time in 34 years (python was released 1991)
3
1
196
10,795
rust's problems are 1) it has an insane learning curve and 2) it takes way too long to compile. python doesnt have these problems at all
python's problems are that it's 1) slow, 2) brittle, and 3) an absolute horrorshow to distribute. rust doesnt have these problems at all
33
17
181
98,408
the goal with ❄️🧊 cryo 🧊❄️ is to make an ENDGAME evm data extraction lego
cryo version 0.2.0 is out today 🌱➡️🌳
here are 5 very badass new features:
☝️✌️🤟🖖🖐️
🐍📊📜🔗🪵
7
27
187
42,022
if scrapping EOF means we can push fusaka to july/august instead of Q3/Q4 then that is yet another compelling reason to scrap it
we don't have the luxury of spending dev time and resources on low impact features
Important Clarification on EOF debate
There's a significant EOF debate today, but one crucial misunderstanding needs to be addressed clearly:
The current EOF discussion is NOT related to the upcoming Pectra upgrade, scheduled for May 7th. The Pectra upgrade does not include EOF, nor intended to include EOF. Everything on Pectra is going as planned for the May 7th release. (eips.ethereum.org/EIPS/eip-7…).
Today's EOF discussion pertains to the subsequent network upgrade, Fusaka, which has not yet been scheduled, but we're aiming for a Q3/Q4 timeline (approximately September/October) (eips.ethereum.org/EIPS/eip-7…).
7
3
162
8,088
once ethereum scales the L1 by 10x, the ultrasound deflation threshold will fall from 20 gwei to 2 gwei
17
6
189
16,987
useful plugin for anyone using polars to process evm data:
polars_evm just implemented binary_to_float()
it uses native polars operations to convert u256/i256 to f64. 10x-20x faster than native python. this makes it trivial to decode most onchain events 🐻❄️🐻❄️
3
17
184
15,233
tomasz is not fucking around
Where are the shorter slot times @dankrad? 6 seconds to start with and how we plan for 1s?
6
3
186
9,711
no more browser extensions for wallets. they just work in the webpage now
the ithaca team has optimized every layer of the stack to create a qualitatively new tools for building crypto apps
one of the most exciting projects on ethereum. even bigger milestones otw
Introducing the Porto Developer Preview!
What if passwords did not exist?
What if the web was built natively for payments?
Porto is the open-source Typescript library we've built @ithacaxyz that gives accounts the superpowers to answer these questions.
Learn more below 🧵
11
7
175
45,420
some detailed data on ethereum read/write patterns for balances/codes/nonces/slots
details in thread ethresear.ch/t/proper-disk-i…
4
32
147
14,824
ethereum has pivoted+reprioritized in leadership, in roadmap, in pragmatism, in maturity, in competitiveness, and in hunger to **actually** build the Future of Finance
and these are all very good things
vocab choice of "pivot" vs "reprioritization" is irrelevant
We discussed this at a Token2049 panel--most call it a pivot, but insiders frame it as reprioritization.
I honestly think it's better for Ethereum to call it a pivot. The market *wants* Ethereum to call it a pivot. Calling it reprioritization seems like saving face. Showing some humility that it made a strategic mistake is critical to building confidence in the asset IMO.
11
8
158
9,693
spice is the easiest + fastest way to extract dune data
it's a modular lego that enables using dune as a data source in your internal pipelines and dashboards
Reminder that Spice exists github.com/paradigmxyz/spice
An excellent way to extract data from @DuneAnalytics
2
9
159
15,044
spice v0.2.0 out today
big highlight = 🌶️🌶️ the spice cli🌶️🌶️
with this new cli it's quick+easy for anyone to extract data from @dune_analytics, not just python users
9
23
158
14,172
ethereum is a trojan horse
when institutions adopt ethereum, they unwittingly hegemonize its core principles of decentralization, transparency, and open source software
13
6
154
21,698
your daily reminder that state growth will not be an issue for many years, even with large increases to the gas limit
raising limit to 150M gas/block would only give state growth of 180GB-360GB per year (incl indices)
state is <10% the size of current 4TB drive recommendation
Ethereum contract devs: if we could 10x the gas limit but, to do so, we had to 10x the gas cost of CREATE, CREATE2 and SSTORE, would that still be a net improvement? Why or why not?
IOW, if state growth is our main bottleneck to scale L1 gas, should we just fix it with pricing?
16
17
156
18,215
I'm giving a couple talks at devcon this year. lmk if youll be around
1: raising the gas limit scientifically
2: sovereign data: how and why
13
3
159
5,260
rust is the main reason why python is usable in 2024
you don't realize how bad python devex is until you use cargo. it's like a modern sportscar vs a horsedrawn carriage. and thanks to @astral_sh much of the rust experience is coming to python
One might say that uv is a fairly popular within the Python packaging space 👀👀👀🫢 here is an image demonstrating it via GitHub stars.
11
6
155
12,837
how to acquire the highest temporal resolution crypto price data:
- for tokens trading on binance, you can download binance OHLC data with 1 second resolution from their website
- for tokens with chainlink oracles, you can get per-block historical pricing, for ~12s resolution. you can scrape this data in bulk efficiently with ctc or cryo
- similar story for tokens trading on DEX's
you can get years of data for free from each of these sources
What is the best api for getting historical crypto prices? As fine of granularity as possible. Coingecko is insufficient.
7
12
148
56,567
compared to polars, pandas is is 1) slower, 2) more memory-hungry, 3) more fragile, and 4) has a significantly worse api
(4) is the most important reason. compared to pandas, polars has an actually sane + consistent + intuitive API where you don't have to look at the docs any time you want to try something mildly complicated
with pandas, even after years of experience, I was always hesitant to use complex data operations for fear of some random thing breaking. with polars I dont worry, I just trust it will work
making complex data transforms fast + effortless changes the way you think about data. you will explore your data with more breadth and depth. and this means you will understand your data better
13
13
147
91,637
are you trying to scrape large amounts of data from an EVM RPC node? 🤔🤔
it's a simple task in theory, but in practice it can require a lot of book-keeping, boilerplate, and fault-tolerant infra
allow me to explain how ctc's new RPC client makes this process easy and efficient
5
14
146
lately there's been a lot of chatter about open source licenses. do you know the major differences between all of these?
- permissive: MIT, BSD, Apache
- gpl: GPL, AGPL, LGPL
- unrestricted: public domain, unlicense
- source-available: BSL
TLDR summary of each license 👇
5
19
140
31,228
someone in a @molly0xFFF reply thread asked me, what are 10 plausible/beneficial use cases for blockchains?
here are 10:
1. @Bitcoin functions as a store of value similar to gold, while at the same time being more cheap+secure to store+transport than gold
2. smart contract chains like @ethereum or @solana allow transmitting $ and other currencies much quicker than traditional bank transfers, seconds instead of hours/days
3. peer-to-peer lending platforms like @aave allow ppl to safely+easily take out loans, or get paid to loan to others
4. stablecoins like @circle have become one of the most significant (top 15) purchasers of USA bonds, which lowers the inflation of the dollar
5. crypto enables donations to regions with geopolitical strife, like when @Ukraine received >$200M in crypto donations despite banking restrictions and destroyed infrastructure
6. prediction markets like @Polymarket create 24/7 information streams that often reveal the outcomes of important events hours/days before traditional news media
7. services like @chainlink can provide verifiable randomness to ensure that competitive/adversarial environments are fair, like ensuring that the superbowl coinflip isnt rigged
8. crypto art marketplaces like @foundation are one of the only places for digital artists to make significant money from their work, and these marketplaces also foster a culture of explicitly valuing artists instead of just using them to train AI
9. storing your crypto in a hardware wallet like @Ledger prevents your government from seizing your assets when you exercise your right to protest, like what the canadian government recently did during the trucker protests
10. blockchain-based DNS services like @ensdomains provide a better system for securing domain names and they cut out the value-extracting domain registrar middlemen
ok im listening. please give me 10 plausible and beneficial use cases for cryptocurrency (or blockchain tech in general). im v curious. write them yourself, no AI, one per tweet. if you do this right now i will give you $500 out of my own pocket. this is my promise to you
10
15
138
25,639
the solana ecosystem needs to do a 100x better job making their data accessible and visible
being poorly understood is the obvious consequence of this underinvestment
1/ I sat down to read the @a16zcrypto State of Crypto report this weekend. I learned a lot! I’m mostly focused outside of the EVM world, so I always appreciate opportunities to learn about innovation happening elsewhere in the world of self custody.
That said: I noticed quite some implicit EVM bias throughout the report. A few observations:
8
6
143
19,571
zkvm's are very exciting
however I still think you should explicitly disavow the stance that "the L1 is not for users", even if you believe most activity will happen on L2's
L1 quality is critical. messaging needs to shift from "L2-centric roadmap" to an "L1-and-L2 roadmap"
3
142
16,741
what tools do you like using for extracting data from ethereum/evm nodes?
- etl: cryo, ethereum_etl
- cli: cast, seth, chifra, ethereal
- libs: ethersjs, ethersrs, ctc, ape, web3py
- frameworks: thegraph, subsquid
what am I missing? looking for self-hosted bulk data extraction
14
16
133
18,707
while we're listing people, you should follow all of these ppl for quality crypto data content
- @nero_eth
- @hildobby_
- @andrewhong5297
- @dataalways
- @sui414
- @0xKofi
- @jconorgrogan
- @MSilb7
- @smyyguy
- @brianisbland
- @tomwanhh
- @sealaunch_
- @artemis
why not check for yourself? It's not that hard to keep up with the activities of EF teams, as most of it happens in the open:
- Applied Research Group hosts ACDC and posts actively on ethresearch, and both @ralexstokes and @mikeneuder are active here and in various other community forums
- The Account Abstraction team contributes to many repos & EIPs (e.g. 4337, 7702), and their contribution will be obvious to anyone who works on wallets or has touched AA over the years - @yoavw is the right person to follow here
- Consensus R&D is also very active (contributing to community calls and on ethresearch), and people on this team like @dankrad and @drakefjustin are similarly pretty vocal and open about their work
- Cryptography Research publishes their work here: crypto.ethereum.org/research
- Devcon's work product probably speaks for itself. This week some of the team are busy with site visits for next year.
- @austingriffith's work on developer growth is similarly very public, follow @buidlguidl as well if you want to keep up.
- The Ecodev coordinators team is more of a support function for other teams, and their outputs are usually bundled with the work of other teams (e.g. helping raising money for important initiatives like the recent security attackathon, or helping improve EthMagicians, etc)
- @EF_ESP manages the entire grants apparatus of the EF, including regular inbound through the open program and they publish quarterly recaps on the blog + info on their website regularly (esp.ethereum.foundation/)
- @ethdotorg's work is similarly visible and easy to follow along with on github github.com/ethereum/ethereum…
- @go_ethereum's work is also all available on github github.com/ethereum/go-ether…, and obviously you will see geth team members on many ACDE calls and similar forums
- @teamipsilon's work is also very easy to follow on github github.com/ipsilon
- @EFJavaScript is the same github.com/ethereumjs
- @EFNextBillion publishes their work here nxbn.ethereum.foundation/
- @ethPandaOps is well known for their high rate of output and value to the ecosystem, you can learn more about them here: ethpandaops.io/
- Portal Network you can also follow along on github github.com/ethereum/portal-n…
- Privacy & Scaling Explorations (more of a group of teams than a single team) publish tons of info about their work here pse.dev/, and you can find the github repos of various PSE projects on that site
- Protocol Security's work and output is also very obvious to people involved in this domain - more about the team here github.com/ethereum/protocol…, and follow @fredriksvantes who leads the team if you want to see regular updates
- Protocol Support's work is also very visible (e.g. running ACDE, running the Ethereum Protocol Fellowship (github.com/eth-protocol-fell…), and lots more) and @TimBeiko is obviously the easy follow here.
- @EthereumRemix's work is also very public github.com/ethereum/remix-pr…
- Robust Incentives Group publishes info about their work on their website here rig.ethereum.org/, follow @barnabemonnot if you want to keep up
- Snake Charmers (i.e. Python) team also works in the open, blog posts here snakecharmers.ethereum.org/ and their work is spread across a few different repos like github.com/ethereum/web3.py, github.com/ethereum/eth-util…, and several others
- The Specs & Testing team manages much of the extensive testing and spec work to ensures that each hard fork upgrade to Ethereum goes off without any issues - their work is most visible to those close to protocol upgrades, but you can learn more here github.com/ethereum/tests and here blog.ethereum.org/2023/08/29…
That's every team at the EF, aside from ones that are already spinning out (see @argotorg) and operational functions like legal, finance, people ops, devops, and opsec which are together ~20 people. We have a pretty thin operational layer for an org of ~250 people.
7
18
138
18,605
𝓐dvent of 𝓒rypto 𝓓ata, DAY 2️⃣5️⃣
to many ppl, rpc config probably seems like the dryest + most boring topic imaginable
that ends NOW
my new rpc config standard will drastically boost the interoperability of the crypto tooling ecosystem
rpc config megathread ⬇️
11
15
125
18,467
there is a widespread misconception that pushing for ethereum's relevance implies compromising its core values
the opposite is true: the best way to respect ethereum's values is 1) to share them with the world and 2) to help give them a significant+lasting role in world affairs
It won't make the world a better place if it compromises on its values. Fortunately the future is oh so bright and we're still very cypherpunk. Pectra makes me erectra.
18
10
135
16,970
the reth issue tracker is so insane
I filed a bug report. when I woke up 7 hours later, the bug was not just fixed, but there had been two PR's by two separate ppl fixing the bug
it's like throwing a dead chicken into a pool of piranhas
10
4
130
15,477
nethermind now supports 27 gigagas/s in some cases
this corresponds to an ethereum gas limit of 324 billion gas per block, which 9000x higher than the current limit of 36 million
So I wanted 10 GGas/s but by mistake I made almost 28 GGas/s.
Was thinking I can't go above 10 and now it is 28 sooo...
100GGas/s next? 📈
10
14
130
10,242
what is the problem?
1) thanks to XEN, 0.3% of Base users are monopolizing 67.8% of the chain's resources
2) state is permanent. Base nodes now need to store + propagate this spam XEN data forever
3) storage abuse can continue until gas pricing is redesigned
12
5
125
20,342
iykyk
new research from @paradigm: How to Raise the Gas Limit ⛽⛽
lots of people want to raise the gas limit, but so far most of the discourse has lacked detailed quantitative evidence
that ends today. we are making the scientific case for scaling Ethereum using ultra-high-res data
4
5
130
10,445
python started off as a sketchy sidechain but now it's becoming a secure L2 using the rust L1
7
16
117
9,788
an unappreciated cost of EOF is that it breaks a lot of ecosystem tooling. EOF solidity can't compile popular smart contracts like solady or OZ. rewrite required
EOF also increases the testing burden for the already overstretched core dev teams and EF ops team
I was one of EOF's biggest supporters.
I now think we might be making a mistake.
Speaking for Reth:
1. Nix it altogether. Replace with EIP7912.
2. If we cannot rally core dev teams on (1), do EOF w/ code & gas introspection restored.
Screenshot w/ full context.
13
12
126
25,421
cryo currently collects 36 types of datasets from EVM nodes
does anyone have ideas for more datasets that would be nice to have?
15
10
121
15,043
𝓐dvent of 𝓒rypto 𝓓ata, DAY 0️⃣2️⃣
how many times have you told yourself “today is the day I learn Uniswap v3 math”?
well today might finally be that day...thanks to @aadams's new Uniswap lib, there’s now an exact implementation of v3 math in python
𝓐dvent of 𝓒rypto 𝓓ata, DAY 0️⃣2⃣
Happy to release the package behind the ever popular @Uniswap v3 liquidity distribution gifs!
Written in polars, anyone can maintain multichain Uniswap v3 data, calculate arbitrary swaps at arbitrary times, and generate state variables from v3
1
12
124
32,746
the paradigm data tools chat is the largest and most active place to discuss crypto data science + data engineering
we cultivated this chat by building sovereign data tools. sovereign data goes beyond open source. it means you can have full control over the entire data pipeline
13
7
120
6,506
amazing news...the previous owner of the `cryo` package on pypi (@cesarizu) has allowed us to have the name
now you can just do `pip install cryo` instead of `pip install cryo_python` 🎉🎉
7
3
122
6,412
in case you missed it, many members of the EF (such as @dankrad, @adietrichs, @nero_eth) and many of the core dev teams, have publicly voiced support for scaling the L1 over the past month
this is a shift from the L2-centric roadmap to a more comprehensive scaling strategy
@jon_charb and @maxresnick were the two main people that convinced me & others over the past 6 months that we disregarded L1 scaling too much as part of the rollup centric roadmap.
Now there is work ongoing to change that - as L1 scales, it will be in part because of these two.
5
2
120
11,079
at ethereum rust con there will be a working group on using rust to transform the crypto data ecosystem
the future is looking very rusty
Excited to share @paradigm will be hosting the Rust x Ethereum Day on August 26th, in San Francisco, right before SBC!
Join us for a day full of Rust x Ethereum, for the engineering and academic community.
Fully technical. No exhibit booths. No distractions.
RSVP Below!
5
13
114
36,930
🌊🌊flood🌊🌊 is the best way to understand the performance of a EVM node or RPC endpoint
new version out today:
- faster
- better docs
- more RPC methods / parameters
- bug fixes
- better docker image
it makes outputs like this ⬇️ or reports like this ➡️datasets.paradigm.xyz/notebo…
1
12
117
10,455
vacations are the best time for coding because you have the time + inspo to work on the crazy ideas
over the past couple weeks in Turkey I had a super fun time rearchitecting cryo
it's not ready yet, but just merged major pr: cryo now collects 24 types of datasets, up from 11
9
4
117
10,791
installing dev tools is now a mostly solved problem
I still remember the bad old days of fighting compilers/libs/drivers/envvars for hours just to install basic stuff
these days you can robustly install almost any tool using a single call to brew, cargo, npm, or uv
5
5
120
7,377
is this specific to Base?
not at all. XEN has left a large mark on every major EVM chain
for example, XEN is now the largest contract on Ethereum. it has been responsible for 8.8% of all Ethereum state growth over the past year and it now takes up 3.5% of all state
8
22
108
22,206
ruff does formatting in addition to linting! and it's nearly instantaneous in speed compared to black. check it out
2
116
13,181
𝓐dvent of 𝓒rypto 𝓓ata, DAY 0️⃣6️⃣
despite CEX <> DEX arbitrage being an "obvious" form of MEV, its theoretical underpinnings remain a highly active area of research
@0xcchan’s new work on CEX <> DEX arbs compares how the prevailing models hold up against empirical data
3
21
107
27,373
at frontiers I gave a talk on four tools I've been working on: 1) mesc, 2) cryo, 3) tbl, and 4) lbl
slides and the video recording available
Most lore filled slides award @paradigm frontiers defo goes to @notnotstorm. Whoever makes these animations/designs is killing it 🔥
8
3
114
7,992
one of the most important steps to scaling eth L1 is "performance-focused testnets"
same featureset as mainnet, just a much higher gasrate
that's the only way to find all the things that will break. not sure why this approach hasn't been used so far but we really should start
Execution layer client teams - do you think it would be all fine to target 100 Mgas blocks on L1 by the end of year?
Consensus layer client teams - any challenges with reorgs?
Community - bad or good idea?
L2s - what are your Mgas/s (or equivalent) targets
2
6
110
9,103
to live a peaceful easy life, you want to write all of **your** code in python and you want **everyone else's** code to be written in rust
2
10
109
19,781
here is the roadmap of datasets that ❄️🧊cryo🧊❄️ will collect
have any more ideas for datasets?
link to spreadsheet with more details below
6
9
108
11,383
hope you’re ready for a special crypto holiday event…
📊🎁 ADVENT OF CRYPTO DATA 🎁📊
25 builders have sync’d up to bring you 25 days of releases: datasets, dashboards, tooling, and more
5
16
111
29,340
“I love the art of programming, so I ignore AI tools and just keep rawdogging my code the old fashioned way”
in medieval warfare the introduction of the crossbow was seen as dishonorable and disruptive to the martial hierarchy
a crossbow enabled a peasant with minimal training to kill a heavily armored knight who had spent their entire life mastering swordsmanship, horsemanship, and combat tactics
this new weapon violated the knights' chivalric codes that idolized bravery and skill in hand-to-hand combat. some churches banned crossbows because a powerful knightly class was necessary for maintaining the feudal order
crossbows eventually became commonplace. warfare didn't end but it did evolve
all arenas of human thought are about to become arenas of AI-assisted or AI-dominated thought. it’s fine if you still enjoy riding horses or swinging around a sword. but if you don’t also learn the new ways, you’re going to get a bolt in the chest
12
8
113
7,815
