Today we saw a 5.22ETH blob fee paid by
@Blast_L2
.
As shown, blob fees can be extremely volatile. Wouldn't it be nice to be able to reserve a fixed blob fee in advance?
@eigen_da
supports blob throughput at a fixed rate annually. DM to learn more.
Over time as builders launch creative applications that are unhindered by data constraints, EIP4844 blob space will rapidly saturate and trigger EIP1559-style auctions.
Cost-effective alt-DA solutions like
@eigen_da
exist for this world, a timeline that’s fast approaching.
Just caught up with the
#CryptoZombies
lessons on Solidity and smart contract development. Thanks and congrats to
@loomnetwork
for building such a fun and edifying tutorial. Check out my hot CryptoZombie Bob and his spawn Bobson:
One of the things I’ve enjoyed during my free time this break was watching VODs of the past World Chess Championship between
@MagnusCarlsen
and
@FabianoCaruana
. In particular, on
@chesscom
’s channel on Twitch, shout out for high production value.
Finally got to try out the
@magicleap
One Creator edition! Fave applications: Project Create, where I could persist goofy hats on my coworkers’s heads in AR and the NBA Screens app, where I could watch a virtual
@KingJames
dunk on a 3D court in sync with game footage.
Ethereum will need non-native data availability to properly scale its rollup-centric roadmap
What is non-native data availability, why does it matter, and how does it work? Let’s find out
Like many others I thoroughly enjoyed reading “How the Internet Happened”, a history of the Internet from Netscape, AOL, Yahoo, Priceline, eBay, Wikipedia, Gooogle, Amazon, Facebook to the iPhone. Lots of fun stories and addresses major themes.
Some observations:
@sreeramkannan
is truly a genius, and not only has an amazing breadth of quantity and quality of ideas, but is also great at sharing them with the world. (while keeping some in reserve of course 😉)
Favourite quote from Robert Sapolsky's Behave so far: "Humans...delay gratification for insanely long times...No gerbil works hard at school to get good SAT scores to get into a good college to get into a good grad school to get a good job to get into a good nursing home."
I originally met
@sreeramkannan
over a year ago as an investor interested in participating in their upcoming round. Over time, I met more of the organization and got increasingly excited about the mission and team.
By ushering in an era of bootstrapped shared security and cost-effective scalability for rollups, we hope EigenLayer and EigenDA play important roles in enabling a new universe of previously impossible services and applications.
The data availability problem is one of the most urgent scaling problems in crypto, and
@eigen_da
offers the cheapest, most performant, and most scalable solution to the crypto industry’s bottleneck problem of data availability, throughput, and storage.
We are building towards an important vision of enabling open innovation. Shared security through restaking solves a real problem faced by protocols and applications looking to bootstrap new networks.
He has surrounded himself with a complementary team, including
@chrisdury
and
@cjliu49
, and many others focused on execution and strategy. The sum of the entire organization up and down comprises one of the smartest, hardest working, and nicest teams I’ve ever worked with.
It is intended to be the endgame for data availability, and the key to eliminating Ethereum blockspace constraints. It's also integral to the long-term EigenLayer vision: in order to penalize bad behaviour, the protocol must access data associated with related transactions.
I bought a piano recently. I’ve been playing for over 20 years, but in the years after college have only been able to sneak sparse sessions whenever there happened to be one lying around. Best decision ever: nothing beats having a great set of keys to hammer on at will.
It's a big idea that the market has resoundingly responded to, with many new and existing projects exploring ways to build on top of the shared security AVS (Actively Validated Service) model.
Over time as builders launch creative applications that are unhindered by data constraints, EIP4844 blob space will rapidly saturate and trigger EIP1559-style auctions.
Cost-effective alt-DA solutions like
@eigen_da
exist for this world, a timeline that’s fast approaching.
L2s pay both a gas fee and a blob fee for posting their batched transaction data to Ethereum.
Usually, blob fees are tiny, and gas is the majority of the cost. Yesterday, however, blob fees skyrocketed and made up 21% of L2 batch posting costs that day.
This is because the
Favorite Hacker News thread of the week is an Ask HN: How you decide what problems to solve in your lifetime? There was a great variety of interesting responses. Of course, YMMV.
The companion
@a16z
podcast with
@cdixon
and author
@brianmcc
is also a must-listen as they expand on many of these narratives in conversation. It’s helpful for entrepreneurs and builders to understand what’s been tried before and how we got to today.
@saraduit
The typical headwinds still exist. Salaries are still relatively low which means you can just throw headcount at problems, and businesses are generally still penny pinching. But companies like Salesforce and Veeva have a presence here and are doing well.
It's been a week since OpenAI blessed some ChatGPT pro users with GPT-4V.
It's so insane what you can do with multi-modal ChatGPT.
Here are 13 incredible examples:
A friend told me about two types of lists he keeps: a list of everything he’s learned as an entrepreneur and a list of traits observed in others to emulate or avoid. Keep these running lists to remind yourself of the experiences you’re having, and to have them all in one place.
Thankful for the time I was able to spend over the holidays with friends and family, skiing, snowboarding, training, watching movies, and reading lots of good books. Rest is important, folks! Let’s go 2019.
AI Story time! Given the recent breakthroughs in LLMs and generative AI, I’ve been reflecting on how algorithms, research techniques, and public perception have evolved over the past decade-plus I’ve been working in the machine intelligence space. (1/n)
I’ve always been an advocate for living in the Peninsula especially if you work in the South Bay and have the option to be there. If you work in SF and want to live there, you should do that, and consider hanging with us sometime :)
🚀 New 🚀 Blog 🚀 Alert 🚀
I’m starting a blog! It’ll hopefully be a fun way to document what I’m learning, get feedback and use as a tool to think through ideas and observations by writing essays.
It felt much more immersive than mobile AR and more interactive with the real world than traditional VR. Super excited to see what creative devs come out with here.
@Kasparov63
@OpenGovInc
@martin_casado
@a16z
The case of how the deep learning breakthroughs of the 2010s affected seemingly unrelated research problems (such as Computer Go) is also a good example of unanticipated second and third order effects,
(25/n)
A great example of content that produces video “exhaust” that’s valuable even after the main production is done, in this case the live event. What keeps it valuable are the live reactions of the community which still apply, and edifying for those who wish to improve at the game.
Watching World Chess Championship 2018 commentary takes me all the way back to my competition days. Highly recommend watching some YouTube / Twitch commentary reruns whether you play chess or not! Congrats
@MagnusCarlsen
@Kasparov63
@OpenGovInc
@martin_casado
@a16z
where breakthroughs in one area can be applied and lead to unlooked-for triumphs in other fields. Expect this to continue to occur as we collectively explore new applications of our recent breakthroughs in LLMs and generative AI.
More to come!
(26/fin)
First, I’m a big believer in living closer to where you work. Having a short commute improves your quality of life dramatically, putting time back in your day.
@saraduit
I’m in China right now. It does feel like enterprise and SaaS will eventually be a thing, but it’s still very hard to make it as as an enterprise startup at the moment.
One month into my new iPad Pro with Face ID, and I’m loving it. I previously had the first iPad Air from 2013, which was honestly amazing too and aged well. But I’m loving the snappy responsiveness of the interface that comes from Apple’s continued innovation on chips.
In breaking down what made it so fun, I should start with the great job
@DanielRensch
and
@GM_Hess
did in hosting , keeping it fun and providing great commentary. As a former competitive player that’s super rusty, I found myself learning a lot.
@0xASK
@DeanEigenmann
@eigen_da
@nickwh8te
Thanks for asking:
@eigen_da
currently doing 0.6 MB/s on mainnet. EigenDA mainnet has supported up to 10 MB/s with peak load tests. For comparison, ETH EIP-4844 blob throughput is 0.032 MB/s and Celestia's mainnet is 1/6 MB/s (2MB blobs every 12s).
EigenDA currently continuously generates synthetic load of 0.6 MB/sec and we have been running load tests on mainnet at peak throughput up to 10 MB/s. For comparison, Ethereum’s 4844 blob throughput is 0.032 MB/s and Celestia mainnet is ⅙ MB/s.
Even though I had the end result of each game spoiled ahead of time, (I watched the tiebreak live) there were several things that kept me engaged. First was the educational aspect. Next, was the entertainment value both from the hosts and the high quality guests they brought on.
Anyways, I love SF as well, and am there usually every week (catch me on the weekends when there’s no traffic!), but often find that living in the Peninsula is underrated among young people. There’s lots of hidden gems here folks! Come down and run/bike/climb/hang with us. :)
@Kasparov63
Though our research did not result in a Go champion, interestingly DeepMind’s AlphaGo combined these reinforcement learning Monte Carlo Tree Search ideas with the deep learning breakthroughs of the time where convolutional neural networks (CNNs) were used to train the 1) policy
Recently enjoyed reading “The Upside of Stress”. Contrary to popular belief, stress isn’t necessarily bad for you, aside from extremely low levels or extremely high levels of stress or trauma, according to recent studies.
The best use of “Hey Siri” is asking the time on a lazy Saturday so you can go right back to sleep without opening your eyes, reaching for your desk, or needing to move at all. 🙌🏻
Really enjoying reading Dear Founder. Lots of practical advice for entrepreneurs in letter-sized snippets on all sorts of topics. And all in one place too!
It also drew live audiences of 30k-80k+ for the ones I’ve seen, which made it exciting. Because it was a somewhat niche audience compared to mainstream Twitch content, the community in chat added to the entertainment factor. And some of the Twitch memes applied to chess were fun.
My friend
@jeanqasaur
got a shout out with her tweet storm here: to make a decision and commit hard. Doesn’t matter if you switch later. Avoid the paralysis by analysis.
Been having many conversations with people in their 20s about the paralysis that can arise from having too many life choices. My advice: commit to something and commit hard. Doesn't matter if you switch later. It's easier to prove yourself if you've had to do it once before.
I find it interesting that some of today’s founders, particularly in the “prosumer tools” space find inspiration from the early computing pioneers of the 70s and 80s such as Alan Kay, Doug Engelbart, JC Licklider, etc.
Currently watching The Bachelor for the first time with my roommates 😂. Good reminder that human interests don’t change very much over time. In terms of the Seven Deadly Sins we’ve got almost all of them: pride, greed, lust, envy, wrath, and (literally) a sloth.
@Kasparov63
Computer chess was more easily solved because 1) chess’s branching factor, the number of available moves to analyze per board state, averages 30-35, and is thus easier to brute force variations computationally compared to Go’s average branching factor of 250. This gap compounds
@Kasparov63
We were studying ideas in reinforcement learning such as Monte Carlo Tree Search, modeling the Go board as a multi-armed bandit, and techniques like Upper Confidence Bounding applied to Trees. These were exploration and exploitation strategies to “exploit” computation on
Enjoyed
@danwwang
's 2017 reflection, in particular the reminder that learning is cumulative with positive returns to scale (learning more facts increases the value of facts you already know, and the easier it is to learn). Great music/book recs as well
My favourite current-events example of typoglycemia is the ‘asymptotic’ and ‘asymptomatic’ pair. Edit distance of 2, begin and end with the same letters, and both relevant in epidemiology. I find myself reading one as the other all the time when skimming along.
#covid19diaries
Interestingly, the way you THINK about stress impacts how it affects you. Viewing stress as positive, a growth mechanism, and your body giving you energy and adrenaline to address the issue at hand makes you more likely to perform, grow, and recover from it.
@Kasparov63
@OpenGovInc
@martin_casado
Though I generally feel that doomer sentiment about AI / AGI is overblown, these series of events have taught me that the development of AI, and technology in general, can advance a lot faster than we may expect. Sometimes when it rains it pours. (20/n)
@Kasparov63
In fact, the reinforcement learning / Monte Carlo Tree Search portion of AlphaGo’s algorithm was not a major change from previous attempts: the breakthroughs came through a much better move suggestion engine and board evaluation function trained through supervised deep learning
@Kasparov63
Of course, AlphaGo was famously victorious against Fan Hui, a professional player, in 2015 and Lee Sedol, one of the strongest players in history, in 2016. DeepMind went on to develop the stronger AlphaZero in 2017, trained only using self-play and MuZero in 2019 to master games
As I finished the book it also occurred to me that a history of how enterprise technology companies evolved would also be incredibly interesting as a parallel to the consumer Internet. How did the rise of enterprise giants like Oracle, SAP, Salesforce, Cisco, etc happen?
@RickBullotta
Devcon finished before the recent Parity wallet issues surfaced actually! It wrapped up Nov 4. Thanks for bringing it up though, the discussion would have been interesting
@Kasparov63
@OpenGovInc
@martin_casado
and combining deep learning breakthroughs with the ideas we were investigating in deep reinforcement learning, this achievement happened a short 4 years later in 2015 with AlphaGo. (22/n)
My first post, titled The Logistics Of Startup Ideas, is based on my experiences as a VC and founder. Before a founder enters the idea maze and searches for product-market fit, there’s a whole other maze of personal foundations to explore.
COVID-19 Diaries: Watching videos online with your friends in 2020 still sucks. We've tried Netflix Party, Metastream, Watch2gether. Ended up counting down 3, 2, 1, GO over Zoom and each individually starting our Netflix streams ¯\_(ツ)_/¯
“Between 1967 and 1979, bad economic surprises led to big and unexpected price declines... Living through that taught me that while almost everyone expects the future to be a slightly modified version of the present, it is usually very different.” - Ray Dalio, Principles
On the list of traits, it can be attributes of others, such as a co-worker, parent, boss that you want to avoid when you one day fill that role, or traits to aspire to.
@Kasparov63
In 2011, computers were playing at a low-dan Go level, equivalent to a high level amateur, far from professional play, let alone sniffing competition against top players. The consensus at the time was that due in part to the high branching factor of the 19x19 Go board compared to