Jul 1, 2017 · 2h 14m · y-combinator
IPFS, CoinList, and the Filecoin ICO with Juan Benet and Dalton Caldwell · Y Combinator
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this Y Combinator Podcast episode, Protocol Labs founder Juan Benet joins host Craig Cannon and YC Partner Dalton Caldwell to discuss the technical foundations of IPFS and Filecoin, the evolution of peer-to-peer networks, and novel economic frameworks for funding open-source protocol development.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The partners hold 7.8% of the talking time here. How this is scored →
The partners as informed peer 2.4 Guest teaching 5.5 Guest disagreement 1.3 The partners pushing back 1.6
speaking balance: gold is the partners, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 50:08 Sharp disagreement over Google Brain's commercial utility
Juan immediately interrupts Dalton's assertion that Google Brain has no practical commercial use, asserting strong disagreement based on massive internal cost reductions.
Hardest push from the partners ▶ 1:29:26 Dalton challenges crypto's wealth concentrationDalton directly questions Juan's idealism around cryptocurrency value distribution by highlighting that actual network data shows extreme wealth concentration among early buyers.
Biggest teaching moment ▶ 29:36 Dismantling the consumer-product lens on blockchainsJuan systematically educates Dalton on why Silicon Valley lagged Europe and New York by viewing crypto through a consumer lens instead of recognizing its role in programmatic finance and contract settlement.
The partners hold their own ▶ 13:09 Dalton establishes early peer-to-peer lessonsDalton cites his personal experience founding an early P2P startup to demonstrate deep familiarity with NAT traversal, bandwidth limits, and UX trumping protocol elegance.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The partners as informed peer | Guest teaching | Guest disagreement | The partners pushing back | Why |
|---|---|---|---|---|---|---|
| Protocol Labs Mission and IPFS Content Addressing | 0 | 6 | 1 | 0 | Host Craig Cannon introduces the episode and asks Juan to define basic terms. Juan delivers an extended solo breakdown explaining the founding of Protocol Labs, the limitations of university and Google research models, and IPFS content addressing. | |
| The Challenge of Human-Readable Names | 1 | 5 | 1 | 0 | Craig asks about progress on human-readable naming. Juan explains the computer science tradeoffs between relative trust graphs, centralized DNS, and global consensus protocols. | |
| Incentivizing Storage with Filecoin | 0 | 7 | 0 | 0 | Juan monologues on Filecoin's architecture, contrasting traditional resource sharing with market-mediated storage and detailing Proof of Replication proofs of work. | |
| Transition to Dalton Caldwell Interview | 6 | 3 | 1 | 4 | Dalton Caldwell joins and shares his personal background founding a peer-to-peer startup, arguing that usability historically trumped architectural elegance in cases like YouTube vs BitTorrent. Juan agrees on usability while emphasizing underlying economic differences and missing peer-to-peer tooling. | |
| Historical Peer-to-Peer vs Modern Blockchain Architecture | 6 | 6 | 4 | 4 | Dalton brings up running Mojo Nation in 1999 and presses Juan on why this cycle will succeed when previous attempts failed. When Dalton asks for a consumer demand killer app like Napster, Juan pushes back firmly, stating Silicon Valley fundamentally misjudged crypto by viewing it through a consumer lens rather than institutional financial automation. | |
| Harnessing Latent Storage and Financial Incentives | 4 | 5 | 2 | 3 | Dalton seeks clarity on what single motivation drives miners and users. Juan clarifies that money is the primary driver for storage providers and explains how software is beginning to programmatically replace legal and financial intermediaries. | |
| Smart Contracts as Programmable Finance and Law | 4 | 6 | 1 | 3 | Dalton asks for non-conceptual, real-world use cases for smart contracts. Juan details algorithmic flight insurance and decentralized commerce like OpenBazaar, explaining that mobile UX remains the temporary bottleneck. | |
| The 2017 ICO Boom and Deep Tech Capital Gaps | 5 | 7 | 5 | 5 | Juan criticizes Silicon Valley's shallow interest driven solely by ICO fundraising rather than deep foundational infrastructure. When Dalton interjects asking why venture capital should fund basic research instead of government bodies, Juan directly challenges Dalton's claim that projects like Google Brain lack immediate commercial utility. | |
| Bell Labs Model vs Decentralized Protocol Monetization | 5 | 6 | 2 | 2 | Juan analyzes the economics of Bell Labs and corporate monopolies funding multi-decade R&D. Dalton points to modern equivalents in self-driving cars and the Linux kernel, while Juan explains how native token models allow open internet protocols to capture value without venture backing. | |
| Market Cycles, Long-Term Value, and Due Diligence | 5 | 4 | 2 | 4 | Dalton compares current crypto speculation to day trading during the dot-com bubble and questions whether deep technical diligence is realistic when investing in basic research. Juan defends rigorous technical diligence over passive hype-chasing. | |
| Protocol Labs Strategy and the Credit Assignment Problem | 3 | 6 | 1 | 2 | Juan outlines the development of CoinList and SAFT agreements before introducing the economic 'credit assignment problem'—the systemic inability of open source ecosystems to route financial upside back to upstream contributors. | |
| Rethinking Capital, Automation, and Value Distribution | 5 | 5 | 2 | 4 | Dalton suggests algorithmic cake-cutting literature as a potential framework for fair compensation and later challenges whether crypto has actually distributed wealth equitably. Juan acknowledges Bitcoin's wealth concentration while advocating for algorithmic, verifiable value distribution mechanisms. | |
| Rewarding Open Source Contributors Across Organizational Lines | 2 | 5 | 0 | 1 | Juan explains experiments to distribute Filecoin returns to early IPFS open source contributors and researchers across corporate boundaries without corrupting non-monetary intrinsic motivations. | |
| Part 3 Intro and Filecoin Disaster Resiliency | 0 | 6 | 0 | 0 | Craig transitions to audience Twitter questions. Juan explains how IPFS and Filecoin achieve disaster resilience through logical decentralization, local shard operation, and tunable erasure coding. | |
| Filecoin Sale Timeline and CoinList Architecture | 0 | 5 | 0 | 0 | Juan answers questions regarding the timeline for the upcoming Filecoin sale and discusses CoinList's mission to reduce compliance friction for token creators. | |
| Filtering ICO Scams, Accredited Limits, and Crowdfunding | 1 | 5 | 0 | 0 | Craig asks how to filter scams. Juan describes CoinList's diligence process and argues that accredited investor rules unhelpfully exclude technically literate non-wealthy retail contributors. | |
| Cloud Storage Replacement and Decentralized Spot Markets | 1 | 6 | 1 | 1 | Craig reframes a user question asking why users would choose Filecoin over Dropbox. Juan explains that Filecoin targets the underlying cloud layer, creating a global algorithmic spot market for latent storage capacity. | |
| Differentiating from Sia and Storj via Proof of Replication | 0 | 5 | 0 | 0 | Juan addresses competition from Sia and Storj, pointing out Filecoin's pre-existing IPFS developer base and Proof of Replication guarantees against Sybil attacks. | |
| How IPFS Mitigates Distributed Denial-of-Service Attacks | 0 | 6 | 1 | 0 | Craig asks if IPFS rendering DDoS impossible is hyperbole. Juan clarifies that while not universally impossible, content addressing makes distributed denial-of-service attacks computationally and bandwidth-wise intractable. | |
| Local Supercomputing, Central Planning, and Market Efficiency | 0 | 6 | 1 | 0 | Juan critiques modern applications routing local collaboration through distant data centers and explains why decentralized protocol markets outperform central planning in discovering local resource optimizations. |
Statements from this episode (39)
Benet: IPFS would have died in academia's publish-or-perish model
“Really I think, like, the only option was either, ah, university or Google, and in the university case, it would have been killed in, in the, you know, publish or perish world where, like, hey, this is way too ambitious. Focus on this, like, one little thing, …”
Benet: Decentralized internet tech directly opposes Google's data control model
“I think this kind of tech is stuff that Google might be interested in funding from the perspective of Google funds a lot of protocols and funds a lot of research, but it also kind of runs counter to basic Google positions around, you know, data, control of dat…”
Benet: The internet must transition from location addressing to content addressing
“People want to address things by what they are, not where they are. Like, it's really time, it's kind of like, it's time for the internet to move from location addressing to content addressing”
Benet: Computing overuses human-readable naming when cryptographic hashes are superior
“I think we have, like, this massive addiction to human readable naming where it shouldn't be used in a lot of places because it brings in a whole bunch of baggage around, hey, now you need a consensus system. Now you need, like, a network stack. Now you need, …”
Benet: Filecoin consensus weights storage capacity over CPU compute power
“If you want to participate in the Filecoin consensus and maintain the Filecoin blockchain what, what's counted is not your CPU raw power as your influence over the consensus, but rather the amount of storage you are providing to the rest of the network.”
Caldwell: Usability always trumped the elegance of peer-to-peer architectures
“And so sort of my takeaway during that era was that usability always trumped the elegance of peer to peer models.”
Benet: Web deployment is easy only because of decades of centralized engineering
“And I would argue that it's easy to roll out a website today because you're, We're working on top of decades of centralized engineering, whereas we haven't had the same level, deep level of engineering on the peer-to-peer side.”
Benet: Shipping physical storage to consumers may become cheaper than internet bandwidth
“Eventually we're going to get to a point where it might actually be cheaper to ship around stuff to consumers.”
Benet: 2013-2014 blockchain thought in New York and Europe far surpassed Silicon Valley
“I think in 2013 and 14, a lot of the blockchain tech was being built in New York and Europe and far ahead of Silicon Valley. And I remember having a lot of conversations with people here in New York and Europe and just the level of thought outside of Silicon V…”
Benet: Bitcoin and Ethereum enable cheap financial instruments, not consumer utilities
“What Bitcoin and Ethereum did was allow you to create Any kind of, you know, financial instrument extremely cheaply and with, you know, almost free verification of correct proceeding of this financial instrument, which is not normally a consumer need.”
Benet: Adding latent storage capacity would drastically lower market prices
“Like, there's exabytes of storage that are not in use right now, and that if you were to add them to the market, you would drive the price down significantly.”
Benet: Blockchain enabled software to eat finance and law
“And so it's, what's, what happened with the blockchain stuff is that software began to eat finance and law in a way that had never happened before.”
Benet: Blockchain apps will reach mainstream consumers within two years
“So I think it's gonna take about a year or two before you start seeing these things get mainstream consumer use.”
Benet: Silicon Valley only noticed crypto because of the ICO craze
“I think San Francisco and Silicon Valley got interested when the ICO craze happened.
When you suddenly had projects raising tens, you know, 20, thirty-five, hundred million dollars, suddenly everyone was like, what the hell is going on?
What's happening?
And i…”
Benet: VC cannot fund deep tech needing over two years of development
“VC is not built for long-term investment in things that are
Extremely high risk and building some deep foundational technology.
VC is tuned for 10 year returns, which means that in two, three years you have to, like, show very strong you know, sign of, like, m…”
Benet: Modern private funding would fail to build an open internet today
“My claim is you couldn't build something as free and open, and that works as well as the internet today because no group would fund it.
and what you would end up with is a massively stunted version of something that is highly centralized and controlled by a c…”
Benet: Tesla and SpaceX would have failed without Elon Musk's personal wealth
“If Elon hadn't been personally wealthy, both projects would have probably failed.”
Benet: Breaking up the AT&T monopoly killed Bell Labs
“Breaking Bell apart effectively stifled and killed Bell Labs.”
Benet: Google Brain and X focus on shorter horizons than Bell Labs
“Google Brain and Axe are much more focused on, on shorter term Valuable things than Bell Labs was. I think both Brain and X can't yet afford to innovate on the multi-decade time scale. They're innovating in like single decade time scales on their own.”
Benet: Protocol Labs uses token models to fund protocol-layer development
“That kind of funding model allows people to remain very close to the actual protocol layer and to think deeply about the protocol itself and its concerns without having to think about a product or a service on top. So this is precisely what protocol apps' busi…”
Benet: Ethereum could reach Google or Apple valuations in 5-10 years
“So if Ethereum continues to be useful five, 10 years from now, you're going to have, and continues to accrue continues to grow, right? So if Ethereum becomes more and more successful, continues to solve a whole bunch of problems, then an ether is going to be w…”
Caldwell: Holding a dot-com peak portfolio long-term still yielded decent returns
“If you would have held and you had a decent portfolio, then you actually, even if you were the dumbest money at the top of the market, you would have done okay.”
Benet: Protocol Labs does not plan to monetize IPFS directly
“IPFS is not something that we plan to monetize directly. Like, IPFS is a large scale infrastructure project That happens to be at a layer where you should not put money in.”
Benet: Distributed labs with long-term funding will replace Bell Labs model
“I think this is, you probably won't see the creation of a Bell Labs like the one before. It's possible that someone like Google and so on creates it, but I think what we'll have instead is, is a much more distributed version of it, where you will have smaller …”
Benet: Tokenized networks are emerging as alternatives to traditional companies
“The company is fading away, or not fading away, but rather a new thing has come in, which is the network or a market that, that is not a company, but it functions kind of like a company. And so you can think of Ethereum not as a company, but rather as a, as th…”
Benet: Crypto speculators receive far greater rewards than builders
“I'm kind of bothered by the fact that in crypto, right now, you're seeing the normal issues with capital flood in, which is that if you're a speculator that has a lot of capital, you can afford to get much greater rewards than the people that actually build th…”
Benet: Careless monetization can corrupt open source and destroy projects
“I'm very, very wary and careful about anywhere where money and open source gets mixed because it can get, like, really screwed up and it can kind of, like, kill projects.”
Benet: Crypto networks reward knowledge work better than salaried employment
“You can't measure knowledge work in hours. Somebody's investment over a decade can put them into the right perspective to make the right contribution at the right time, that creates something like Ethereum or Bitcoin or whatever. And, you know, how do you corr…”
Benet: IPFS nodes can communicate even if disconnected from the network
“The nodes in the IPFS network can continue talking to each other, even if the rest of the network disappears.”
Benet: CoinList's main value will be finding high-signal token projects
“The main selling point of Coinless will be, ah, decreasing the amount of work for token sale creators and creating a network that focuses on finding really high signal projects, right?”
Benet: Accredited investor wealth does not mean crypto technical judgment
“Having a million dollars does not mean that you know what cryptocurrency network is gonna be better, or what cryptocurrency network actually will work.”
Benet: Filecoin is an enterprise cloud storage replacement, not consumer Dropbox
“Filecoin is not to be seen as a Dropbox replacement, although there will be Dropbox-like things that build on top of Filecoin. I think you should think about Filecoin as replacing cloud storage. So it's something that Dropbox would use.”
Benet: Decentralized storage nodes will beat AWS on unit economics
“Our bet there is that yes, that there's a whole bunch of places and cases where Certain individuals or groups in the world have access to either really cheap storage or storage that's positioned well in the network that is kind of like, you know, somewhere bet…”
Benet: Storage miners will multi-home across networks like Uber and Lyft drivers
“I actually bet a lot of people will be mining on both networks, or like all, all, all of the networks, and we'll see how that actually plays out and then we'll see, kind of like right now, there's a lot of drivers driving for either Uber or Lyft”
Benet: Proof of Replication prevents Sybil attacks in decentralized storage
“When you have proof of replication-backed storage, that, that ensures that it is unique to this particular request, and that's a very important thing from a game theory perspective. Like, you don't want people, you don't want, like, networks of civils, basical…”
Benet: IPFS makes DDoS attacks impossible or computationally intractable
“The other case is, hey, like, if there's some really popular piece of content, and suddenly it gets replicated to tons of people, now the DOS attack gets way harder, right? Like, now, now you have to DOS, Thousands of people. And so it's not fundamentally, in,…”
Benet: Routing local collaboration through remote data centers is flawed architecture
“You could have a room full of people with laptops open on Slack or on a Google Doc, and they can't work together because their supercomputers, which again, like, let's be clear here, like, these computers are more powerful than, like, all of the computers on t…”