Protocol Labs Research

A farewell to Protocol Labs Research

Wed, 01 May 2024 00:00:00 +0000

Reflecting on my time at Protocol Labs Research over the past five years, I can’t help but feel grateful. What began as a journey into uncharted territories of start-up research evolved into an extended mosaic of happy memories.

Together, we pushed boundaries and explored topics far beyond our initial plans. We contributed to metaresearch. We launched a grants programme. We organised world-class research events. We supported critical COVID research. We amplified outstanding research in our long-running seminar series. We incubated and spun off research labs. We launched and maintained open-source projects. We published tutorials on scientific writing and distributed systems. We generated novel research and disseminated it through talks and papers. Though we didn’t achieve everything we set out to do, every output remains a testament to our collective passion for research.

Yet our biggest accomplishment was bringing together a unique group of brilliant staff, advisors, and interns in an environment that would prove hard to replicate. While Protocol Labs Research may no longer exist as an organisation, its legacy lives on through the contributions of its members and the projects it nurtured. The spirit of innovation and collaboration it fostered continues to thrive within many nucleated teams in the larger Protocol Labs Network.

While this website will remain a repository of our work, this will likely be its last update. My heartfelt thanks to everyone who has been a part of this incredible journey.

F3 and GossipBFT: Fast finality on longest-chain protocols

Tue, 27 Feb 2024 00:00:00 +0000

A finality calculator for Filecoin’s Expected Consensus

Thu, 01 Feb 2024 00:00:00 +0000

We propose a finality calculator for Filecoin’s Expected consensus that considers what takes place during epochs and can attain, under normal operating conditions, an error probability of 2^(−30) in 30 epochs (15 minutes) - a 30x improvement over the current 900-epoch threshold. It depends only on a node’s local view and can be implemented without protocol changes.

Filecoin Proof of Useful Space

Wed, 30 Aug 2023 00:00:00 +0000

This document provides a simple formal definition of Proof of Space (taken from the academic literature) and an informal definition of persistent and useful space (needed for Filecoin). It describes construction details and a security proof for the Stacked-DRGs proof of space (SDR), and goes into how SDR is used in Filecoin. In particular, it includes a description and analysis for Filecoin’s PoRep, WindowPoSt and WinningPoSt.

Security analysis of Filecoin's Expected Consensus in the Byzantine vs honest model

Mon, 14 Aug 2023 00:00:00 +0000

Filecoin is the largest storage-based open-source blockchain, both by storage capacity (>11EiB) and market capitalization. This paper provides the first formal security analysis of Filecoin’s consensus (ordering) protocol, Expected Consensus (EC). Specifically, we show that EC is secure against an arbitrary adversary that controls a fraction β of the total storage for βm<1−e−(1−β)m, where m is a parameter that corresponds to the expected number of blocks per round, currently m=5 in Filecoin. We then present an attack, the n-split attack, where an adversary splits the honest miners between multiple chains, and show that it is successful for βm≥1−e−(1−β)m, thus proving that βm=1−e−(1−β)m is the tight security threshold of EC. This corresponds roughly to an adversary with 20% of the total storage pledged to the chain. Finally, we propose two improvements to EC security that would increase this threshold. One of these two fixes is being implemented as a Filecoin Improvement Proposal (FIP).

Janus and Granite

Mon, 17 Jul 2023 00:00:00 +0000

ConsensusDays 23 recordings now available

Fri, 16 Jun 2023 00:00:00 +0000

ConsensusDays 23 took place 5-6 June and we have another successful edition to celebrate! Beyond the exciting programme, here are a few stats from this year:

22 talks
35 submissions
231 registrations
347 members of the #consensus channel
612 members of the ConsensusDays mailing list
614 YouTube views of the raw streams

Today, we bring you the final news of the year: the edited talks are now available on YouTube. The individual videos are also linked from the programme page.

We’d again like to thank our keynote speakers, Aggelos Kiayias and Zarko Milosevic, as well as all authors and speakers. Our gratitude also goes to Alberto Sonnino, Alejandro Ranchal-Pedrosa, Alexander Spiegelman, Alysson Bessani, Arthur Gervais, Chrysoula Stathakopoulou, Duc Le, Ertem Nusret Tas, Giuliano Losa, Guy Goren, Henrique Moniz, Jiangshan Yu, Joachim Neu, Josef Widder, Lei Yang, Matej Pavlovic, Nenad Milosevic, Qiang Tang, Rati Gelashvili, Rodrigo Rodrigues, Sergio Mena, and Srivatsan Sridhar for their hard work reviewing the submissions, and Solaris for their support organising the event.

We’ll start thinking about ConsensusDays 24 soon! To stay in the loop, consider subscribing to the ConsensusLab YouTube channel and Twitter account.

Private retrieval grant 2023 roundup

Tue, 13 Jun 2023 00:00:00 +0000

We have concluded the initial round of funding decisions for RFP014, a research initiative in collaboration with Arcological, focused on exploring mechanisms for private data retrieval. The objective of RFP014 is to foster the advancement of private communication methods. We are excited to support the researchers who are discovering and defining novel techniques to improve both scale and efficiency of solutions to this problem.

In our first cohort, we extended a total of $750,000 in funding to 6 teams or individuals.

3 are investigating the scalability of “private information retrieval” techniques
3 are investigating applications of threshold or multi-party computation
3 are investigating mechanism designs involving multiple hops - across DHTs or mixnets.
3 are investigating the use of homomorphic encryption.

Many of the projects we are supporting advance multiple categories.

State of the Field

There are many ways to divide up the distributed systems problems that are referred to collectively as private data retrieval. We tend to segment the problem by the cryptographic building blocks that are used.

Homomorphic Computational Private Information Retrieval

Spiral recently demonstrated a single server PIR server holding millions of database items that can reach a throughput of 2GB/s. This system has great demos both for ENS resolution and wikipedia. This work is being commercialized as a hosted service, Blyss. There are also active investigations further scaling and improving the performance of this technique.

Multiparty Computation

Medusa provides a building block for a threshold encryption networks. There are system designs using the MPC primitive for unlinking queries, and for re-encryption. Distributed key generation algorithms and threshold cryptography are continuing to increased scale, with threshold key generations for web 3 projects now reaching thousands of participants.

Mixnets

Nym is an active experiment in the effectivness of incentivization of a mixnet system. The ability to reimburse infrastructure operators without deanonymizing traffic adds a new dimension. We are interested to see if incentivization can help overcome one of the major stumbling blocks for this construction: that lots of cover traffic is needed relative to any message being anonymized.

Zero-Knowledge Systems

Zero Knowledge Proof systems have been advancing rapidly across hardware acceleration, compilation primitives, and efficiency. There are direct applications for advances to ZKPs in the blockchain industry, which is already effective in incentivizing development here. Increased efficiency of ZKPs may become sufficiently cheap to support validation of messages within private retrieval constructions. We already see increasing use of the construct in PIR systems.

Oblivious Pseudorandom Functions

Oblivious techniques describe a structure for access or interaction that protects a piece of information. OPRFs allow masking functions and have been applied in PIR and PSI contexts. The sizes and efficiency of these constructions are being scaled through their usage in PAKE’s - password authenticated key exchanges.

What’s next?

We will be iterating on our application process, with an aim to review applications faster. We expect to launch the next iteration of RFP014 in the summer of 2023.
We are invested in this problem ourselves, and will be following the work done by grantees. We will be providing a yearly summary of the maturity, performance, scalability, and opportunities for metadata-private data retrieval.

Supported Projects

The full list of grantees of our first round of funding is:

Private Information Retrieval for IPFS - Miti Mazmudar (University of Waterloo)
Private information retrieval with access control - Sebastian Angel (University of Pennsylvania)
Scalable Private Information Retrieval Protocols using Lattices - David Wu (Univsersity of Texas at Austin)
Private retrieval of data (single-server TreePIR) - Arthur Lazzaretti (Yale University)
PIR on the mixnet - masala (Katzenpost)
Secret Recovery Service - Kevin Wellenzohn, Stefan Hechenberger (Blockshake GmBh)

ConsensusDays 23 programme and registration

Wed, 10 May 2023 00:00:00 +0000

We’re happy to announce the publication of the ConsensusDays 23 programme. The workshop will take place 5-6 June in the 14:00-18:30 UTC period.

We were again overwhelmed by the community interest, and many decisions ended up coming down to program limitations and session planning rather than quality alone. In a repeat of the 2021 edition, the number and quality of talk proposals led us to extend the event with a second day in order to accommodate more discussions. In the end, 20 submissions were accepted out of a total of 35.

We are grateful to all authors for their submissions and to Alberto Sonnino, Alejandro Ranchal-Pedrosa, Alexander Spiegelman, Alysson Bessani, Arthur Gervais, Chrysoula Stathakopoulou, Duc Le, Ertem Nusret Tas, Giuliano Losa, Guy Goren, Henrique Moniz, Jiangshan Yu, Joachim Neu, Josef Widder, Lei Yang, Matej Pavlovic, Nenad Milosevic, Qiang Tang, Rati Gelashvili, Rodrigo Rodrigues, Sergio Mena, and Srivatsan Sridhar for their hard work reviewing the submissions.

In addition to the accepted submissions, the event will also feature invited talks from Aggelos Kiayias (IOHK) and Zarko Milosevic (Informal Systems).

If you haven’t yet, please register to attend the workshop. We’ll be sending videoconferencing links to registered participants shortly before the workshop. In the meantime, we also invite you to join us in Slack. Please join our workspace and head on over to the #consensus channel.

Base fee manipulation in Ethereum's EIP-1559 transaction fee mechanism

Sat, 22 Apr 2023 00:00:00 +0000

In 2021 Ethereum adjusted the transaction pricing mechanism by implementing EIP-1559, which introduces the base fee - a fixed network fee per block that is burned and adjusted dynamically in accordance with network demand. The authors of the Ethereum Improvement Proposal (EIP) noted that a miner with more than 50% of the mining power might have an incentive to deviate from the honest mining strategy. Instead, such a miner could propose a series of empty blocks to increase its future rewards. In this paper, we generalize this attack and show that under rational player behavior, deviating from the honest strategy can be profitable for a miner with less than 50% of the mining power. Further, even when miners do not collaborate, it is rational for smaller mining power miners to join the attack.

IPC subnets land on Filecoin Spacenet

Thu, 20 Apr 2023 00:00:00 +0000

The Interplanetary Consensus framework (IPC), formerly known as Hierarchical Consensus, addresses two challenges of blockchain networks, transaction volume and application heterogeneity. In doing so, it boosts the capabilities of the Filecoin network. Today, after 18 months of development, we’re happy to announce the first public IPC deployment as part of the Spacenet testnet.

We first invite you to ramp up quickly by watching this two-minute introduction to IPC, which explains the high-level aspects in just two minutes. Then keep reading for more detail!

New use cases

The Filecoin Virtual Machine (FVM) launch unlocked programmability across the Filecoin network. Now, the blockchain can be used for more than storage deals, storage proofs, and simple transactions, and EVM compatibility makes it possible to quickly deploy existing Ethereum smart contracts.

Programmability represents a seismic shift for the Filecoin blockchain, which can now play host to DeFi solutions, Data DAOs, metaverse gaming, and much more. Think of all traditional blockchain applications that can be augmented by native storage — or think of adding processing units to the data ocean already on Filecoin.

The forthcoming surge in activity is not without challenges. Filecoin blocks are seldom full nowadays, making for easy transactions with low fees and delays. However, as FVM activity ramps up, there will be competition for block space, increasing the cost of running applications on Filecoin. Moreover, not all applications can operate in a network with 30-second block times and the resulting latency.

This is where IPC comes in. By allowing end users and application developers to dynamically spawn hierarchical subnets that execute transactions in parallel, IPC removes excess load from the root network — the Filecoin mainnet — while providing operating conditions better suited to the needs of applications. In addition, IPC provides cross-subnet transactions and mechanisms to secure subnets by leveraging stronger guarantees provided by their parent.

Under the hood of IPC

IPC is, at its core, a framework that defines the rules for interaction within and between subnets, accompanied by a suite of smart contracts and client software implementing said behaviours. When deployed to the Filecoin mainnet, it will run entirely in userspace, anchoring to the root network via a user-defined actor.

The basic functionality of IPC involves creating and operating subnets, which are organised in a tree. Spawning is permissionless, and users can decide where to anchor their subnet in the tree. Each subnet is associated with an independent set of validators running a separate consensus instance. Subnets submit frequent state checkpoints to their parent network, from which they, in turn, derive some security guarantees.

Subnets may be deployed for many reasons, for instance, to save on transaction fees or to group interactions along geographical, topical, or application criteria. They can be long-lived or ephemeral, established to coordinate computation and discarded shortly after.

Subnet consensus is both swappable and configurable. A subnet can run any consensus algorithm: proof-of-stake, proof-of-work, proof-of-storage, classical BFT, or anything else. Along with the IPC framework, we are releasing the implementation of Trantor, a high-performance BFT consensus algorithm that is a good fit for a broad range of applications. While IPC users can implement other consensus algorithms — and we intend to look at out-of-the-box alternatives in the future, many application requirements can be satisfied by merely changing the default parameters of Trantor, for instance, its block frequency.

A key feature of IPC is the transparent cross-subnet communication mechanism provided out of the box. These mechanisms allow for easy transfers of funds in and out of subnets and for interaction with users and state located elsewhere in the IPC tree.

But how does it all work in practice? The details of how to interact with IPC depend on whether you’re a user, developer, or validator. However, at a high level, a user interacting with IPC locally will rely on the IPC Agent application and a number of blockchain nodes — one for each subnet. The local applications are complemented by two on-chain smart contracts, which are instantiated for each subnet: the Interplanetary Subnet Actor and the Interplanetary Gateway Actor.

If you’d like to learn more about IPC or its components, proceed to our fresh-off-the-boat docs, including the user documentation for the IPC Agent and a technical summary of the framework. Please remember that the software, tooling, and documentation are all works in progress — if you hit any roadblocks, a message in Slack or a GitHub issue will get you the fastest support and help us improve.

What’s next for IPC?

Today’s M1 milestone brings IPC to the final stretch of its road to production. It marks the first time a long-running IPC testnet is available to the public, enabling user, developer, and SP experimentation with subnets and consensus models other than the one the root Filecoin network provides. It is, however, a test deployment, and shouldn’t be used for production applications with value at stake.

Over the coming quarters, we will keep enriching IPC with additional features and increased robustness. As we speak, our partners at Limechain are hard at work translating our actors into user-defined FEVM Solidity actors, which we will use for the first mainnet deployment. We will work on a regular release cadence, with new features and improvements every few weeks.

We are also working with launch customers, particularly the Saturn team, to ensure we enable their use cases in milestone M2, scheduled to land in June 2023. M2 will bring IPC to the Filecoin mainnet and allow early users to deploy their applications in production. It will be followed by a full release in Q3, complete with developer tooling and documentation.

Throughout this process, we’re very keen on collecting user feedback. If you have a use case for IPC, would like to work with us to deploy your application or operate as a subnet validator, or would just like to continue the discussion with this nascent community, please join us in #ipc-help on the Filecoin Slack workspace.

Cryptonet launches new open source SNARK system

Thu, 06 Apr 2023 00:00:00 +0000

Testudo is a new open source SNARK system developed by Cryptonet that offers efficient proofs with smaller setups. It uses polynomial commitments and sumchecks to prove the satisfiability of an R1CS system, and applies several optimizations to reduce the trusted setup size, improve proving times, and achieve fast verification and small proof size.

The team is looking for engineers to help push the effort forward. Read more about the details here!

New directions in Network Research

Fri, 31 Mar 2023 00:00:00 +0000

As Protocol Lab’s footprint in the public goods space has grown, the Network Research team has taken the lead on integrating and adapting PL’s work to the needs and interests of the scientific community, with the goal of designing and building public goods to move science forward faster.

At Funding the Commons Lisbon, Network Research described how our public goods primitives – hypercerts, impact evaluators, discourse graphs, research roadmaps, and network funds – can be composed into an integrated coordination system.

We’ve been working with our fellow researchers and developers to build new infrastructure for discourse graphs, and to make publicly-shared graphs discoverable and citable. The Atlas of Synthesized Knowledge is the home for the community of practice around graph-based communication and synthesis protocols.

We’ve also been working with collaborators across various fields to refine, implement, and promulgate our research roadmapping protocol, creating new tools and methods to align researchers and research funders and plot new courses through the scientific problem space.

We’re excited to announce that these efforts have led to a new initiative combining our work on graph synthesis and research roadmapping: the Network Goods Research Roadmapping project integrates the knowledge synthesis and communication capabilities provided by discourse graphs with the coordination and alignment function of research roadmaps, and adds new tools for the discovery and incentivization of scientific public goods.

We’re bidding a fond farewell to the Network Research team designation but keeping our laser focus on building metaresearch initatives enabling revolutionary coordination systems under the Network Goods mandate. Join us at Metascience 2023 to learn more about our portfolio of projects, and watch this space for further news!

Scalable asynchronous randomized byzantine agreement

Thu, 30 Mar 2023 00:00:00 +0000

A metaverse of walled gardens?

Wed, 29 Mar 2023 00:00:00 +0000

tlock: Practical timelock encryption based on threshold BLS

Wed, 29 Mar 2023 00:00:00 +0000

The road to Lurk

Fri, 24 Mar 2023 00:00:00 +0000

(begin

In early 2018, I joined the Filecoin team at Protocol Labs as a contractor. I was fascinated by the promise of decentralized storage and the possibility of consensus based on storage power. I eagerly devoured the Filecoin whitepaper and puzzled over how the proofs were supposed to actually work. Sufficiently persistent enthusiasm paid off, and I ended up joining Protocol Labs full-time to lead the Filecoin Proofs implementation.

Today is my last day at Protocol Labs, after which I am moving on to be the CTO of Lurk Lab. The last five years have gone by in the blink of an eye. As is common in transition, I find myself reflecting on how the past became the present and on the future to come. This post is a whirlwind summary of some of the highlights of my research/development work at Protocol Labs. Because this has been a period of intense and fruitful focus, a narrative thread also emerges.

This post is several things:

A retrospective tour of some things I worked on while at PL, providing some insight into how the sausage is made.
A summary of the latest state of the VDF project and a first look at some related research outputs.
The (eponymous) road to Lurk.

Filecoin

Much has been written about Filecoin and the Filecoin Proofs. Here are a few notes about lesser corners of the work that went into arriving where we did.

One of the earliest abstractions we adopted was embodied in the CompoundProof trait. I am both proud and horrified that I broke ground on what would become a central part of our proofs implementation while simultaneously learning Rust and the underlying cryptography: research development at its ~~finest~~ fastest. More seriously, though, CompoundProof formalized a relationship between ‘vanilla’ proofs and their zk-SNARK counterparts, initializing a thread of investigation eventually leading to Lurk.

On the way to mainnet, we needed a way to model the many parameters associated with the network and especially candidate proof constructions. Hence was born Orient (a relational language for expressing systems of bi-directional arithmetic constraints, implemented in Common Lisp).

From this, we built the Filecoin Ubercalc. Though crude, it proved powerful and flexible enough to interactively model the characteristics of proposed candidates and their parameterizations in order to make difficult time-critical decisions about what we could deploy on time, with acceptable performance, and without sacrificing network security.

One thing that became clear in the Ubercalc analyses was that none of our then-current candidate hashes struck the right balance of performance inside and outside arithmetic circuits. Whereas several years earlier, we had ruled out Poseidon as being ’too new’, we decided making a (hedged) bet on SNARK-friendly hashing was our best option. To that end, I created Neptune, one of the first production implementations of Poseidon, and as far as I know, the first to tackle the sparse matrix optimizations enabling best non-circuit performance for higher arity hashes.

Because we needed to build Merkle trees as fast as possible, that included a GPU implementation (Triton).
- Don’t get me wrong, the initial GPU implementation was nowhere near as performant as possible – though Supranational has more recently done tremendous work to squeeze out the best performance. Rather, it was ‘fast to build’ for someone who didn’t know GPU programming. I used Futhark – a purely-functional GPU programming language (Retrospective on an implementation of the Poseidon hash function in Futhark).
- Sometimes research engineering demands a dose of self-medicated rabbit-holing, and the requisite implementation of Finite Fields for Futhark helped me through those dark days.

Along the way, a number of long-running SNARK engineering research projects allowed us to scout optimal paths,

In the end, we managed to launch in 2020 with what soon became the largest and most performant SNARK deployment ever (at least at the time we published zk-SNARKs for the world!).

VDF

After mainnet launch, I transitioned away from maintenance of the Filecoin proofs to focus on accelerating usability of verifiable delay functions. There were several potential use cases for VDFs in Filecoin, and my aim was to help make a viable implementation available to such downstream applications. As long as it remained a 'someday' technology, practical research and development could not assume it.

As importantly, in targeting this project as a means to drive SNARK research, I identified VDFs as an optimal test case for recursive SNARK technology generally. The extreme simplicity of an iterative algorithm like a VDF would allow us to push proving system design and implementation forward as rapidly as possible – if only by providing demand and supporting early adoption. By focusing on an almost trivial computational problem, we incidentally ensured that success in proving would accrue almost entirely to the underlying proving system.

At the same time, this focus on ultimate simplicity in the circuit to be proved exerted a profound influence on the work (see below) which would become Lurk.

Computational VDF

The dominant approach to VDFs involves functions that are slow to compute but fast to verify for computational reasons.

To that end, in November 2021, I organized a collaboration to build a Practical SNARK-based VDF.
I built and maintained an initial VDF prototype, allowing iteration on the design and coordination between the hardware, algorithmic, and proof design – including explorations of optimal parallelism for min-root informing the eventual evaluator ASIC design (see below).
- In Open VDF: Accelerating the Nova SNARK-based VDF, Supranational writes:
  
  ‘Approximately one year ago, we announced a collaboration with the Ethereum Foundation, Filecoin Foundation, Supranational, Microsoft, and the Electric Coin Company to improve the performance of SNARKs, and to make SNARK-based Verifiable Delay Functions (VDFs) practical. VDFs have the potential to improve the security, privacy, and scalability of blockchain platforms and are being evaluated by a number of blockchain platforms, including Ethereum, Filecoin, Tezos, and Zcash. Over the past year, we have collectively made a huge amount of progress towards that goal, including an ASIC implementation of the MinRoot algorithm, implementation of a recursion-enabled proof system (Nova), and support for GPU accelerated proof generation. While we have laid the groundwork to make SNARK-based VDFs practical, these developments will have much wider-ranging implications. For instance, through the development of a performance approach to incrementally verifiable computation, we have made practical novel and powerful use cases such as true zero-knowledge VMs like Lurk.
  
  Furthermore, in addition to these SW developments, earlier this year we received initial prototypes of the MinRoot ASIC design, manufactured on 12 nm technology. Over the coming months, we will integrate the MinRoot ASIC with the Nova proof system and produce an end-to-end implementation of SNARK-based VDFs. Once this is complete, we will be sharing this implementation with the community and are excited to see how this new cryptographic primitive can improve the security and scalability of blockchain networks.’

Fundamental Latency VDF

Although computational VDFs are marvelous, they suffer one significant problem. Their performance on the key metric of VDFs ($A_{max}$) depends on engineering attempts to implement an evaluator which is as fast as possible. This fails to whatever extent an attacker can accelerate the computation. In practice, this means computational VDFs with an $A_{max}$ very close to 1 are not realistic.

What this means is that a single VDF computation cannot be used as fine-grained timer for applications that require high assurance that no attacker can obtain the result with less delay than an honest party. Some of the motivating Filecoin applications required this property in order to satisfy their economic requirements (essentially the $A_{max}$ acted as a cost multiplier).

In the spirit of stimulating the otherwise impossible (it was part of a post-mainnet vibe that possessed me), it occurred to me that communication latency does indeed allow for uncheatable minimum times (with the speed of light famously providing a universal speed limit).

I wrote up my initial, partially-baked thoughts in a starry-eyed document titled SpaceVDF.
The ideas were promising enough that I followed up with Cryptosat, and after some discussion, we commissioned a system-level trade study.
I’m excited to announce the publication of the SpaceVDF White Paper.

‘In this document we aim to evaluate how VDF algorithms based on physical limits can be implemented in satellites and which physical properties / or roles of physics we can utilize to guarantee the passage of time. The goal of this study is to perform principal system analysis, identify main issues and risks, propose a path for derisking and come up with a budget and timeline for a suitable satellite (or satellite constellation).’
Just as my seed ideas were only suggestive of the actual engineering that would be required for any real-world deployment of a SpaceVDF, the cryptographic constructs imagined were also vague and incomplete. To that end, I connected with Bernardo David, who agreed the core ideas were interesting enough to develop further.
We sponsored Bernardo and his co-authors to put meat on the bone, and the result has just been released as an eprint: CaSCaDE: (Time-Based) Cryptography from Space Communications DElay

‘In this work, we investigate how to construct time-based cryptographic primitives from communication delay, which has a known lower bound given the physical distance between devices: the speed of light. In order to obtain high delays, we explore the sequential communication delay that arises when sending a message through a constellation of satellites. This has the advantage that distances between protocol participants are guaranteed as positions of satellites are observable, so delay lower bounds can be easily computed. At the same time, building cryptographic primitives for this setting is challenging due to the constrained resources of satellites and possible corruption of parties within the constellation.

We address these challenges by constructing efficient proofs of sequential communication delay to convince a verifier that a message has accrued delay by traversing a path among satellites. As part of this construction, we propose the first ordered multisignature scheme with security under a version of the discrete logarithm assumption, which enjoys constant-size signatures and, modulo preprocessing, computational complexity independent of the number of signers. Building on our proofs of sequential communication delay, we show new constructions of Publicly Verifiable TLPs and VDFs whose delay guarantees are rooted in physical communication delay lower bounds. Our protocols, as well as the ordered multisignature, are analyzed in the Universal Composability framework using novel models for sequential communication delays and (ordered) multisignatures. A direct application of our results is a randomness beacon that only accesses expensive communication resources in case of cheating.’

Lurk

Of the potential applications in which VDFs might play a critical role, one of the most interesting was Hierarchical Consensus (which has since been re-branded¹ as Interplanetary Consensus). However, following a worrisome trend, the more I pondered it, the less obvious it became that the application would be straightforward. Whereas the VDF (see above) represented the simplest possible proof statement, it seemed more elaborate statements would need to be proved in order to serve the needs of substantial applications like hierarchical consensus.

I wrote Hierarchical Consensus with VDFs and/or Turing-Complete SNARKs before Chat GPT came along, so the concept of LLM hallucination was not yet widespread. Nevertheless, it would not be too far off to view this document and the prompt that is its epigraph to be a fever dream. Please note that it was written in mid-2021 and predated current work on IPC, so nothing about that project should be inferred from its first-principle extrapolations. As it turns out, the Turing-Complete SNARKs part may indeed provide high value to IPC (or a system like it), even if the VDFs prove dispensable.

Gripped by further self-medication of research development fever, I saw fit to focus on what seemed another missing piece to be forced into existence ahead of its time, leading to the MetaProof project – described as follows:

‘The first step of the MetaProof project follows the shortest path to Turing complete SNARKs via recursive computation. We design a minimal Lisp-family language using SNARK-friendly hashing for memory access — and implement an arithmetic circuit proving one reduction step of this language’s expression evaluator. Combined with any IVC-like construction (including SnarkPack + input aggregation), this will allow succinct proofs of arbitrary computation from a single circuit. In addition to the operational benefits of avoiding the research/development lifecycle associated with new circuits (potentially including trusted setup), this will allow entirely new categories of computations to be proved.’

One thing led to another, and Lurk was born.

In house style, implementation led articulation, but now we have an articulate explanation as well: LURK: Lambda, the Ultimate Recursive Knowledge.

‘We introduce Lurk, a new LISP-based programming language for zk-SNARKs. Traditional approaches to programming over zero-knowledge proofs require compiling the desired computation into a flat circuit, imposing serious constraints on the size and complexity of computations that can be achieved in practice. Lurk programs are instead provided as data to the universal Lurk interpreter circuit, allowing the resulting language to be Turing-complete without compromising the size of the resulting proof artifacts. Our work describes the design and theory behind Lurk, along with detailing how its implementation of content addressing can be used to sidestep many of the usual concerns of programming zero-knowledge proofs. '

And that brings us up to the edge of the present.

Lurk Lab calls.

PL has partnered with Lurk Lab to bring general-purpose zero-knowledge proofs to Filecoin, and is our largest investor.

Here’s a brief (if imperfect) Lurk demo.

This post focused mostly on the intertwined narratives of ‘my time at PL’ and ‘how Lurk came to be’. Frequently, the accomplishments I cite as ‘mine’ were the result of collaboration. Astute readers will detect that some recurring stylistic mishaps cannot be blamed on others, however.

) ‘And death i think is no parenthesis’

-- Chhi'mèd Künzang, 24 March 2023

Technically, “brought into the light of the solar system,” per @jsoares. ↩︎

SpaceVDF: Verifiable delay functions using cryptographic satellites

Tue, 21 Mar 2023 00:00:00 +0000

In this document we aim to evaluate how VDF algorithms based on physical limits can be implemented in satellites and which physical properties / or roles of physics we can utilize to guarantee the passage of time. The goal of this study is to perform principal system analysis, identify main issues and risks, propose a path for derisking and come up with a budget and timeline for a suitable satellite (or satellite constellation).

LURK: Lambda, the ultimate recursive knowledge

Thu, 16 Mar 2023 00:00:00 +0000

We introduce Lurk, a new LISP-based programming language for zk-SNARKs. Traditional approaches to programming over zero-knowledge proofs require compiling the desired computation into a flat circuit, imposing serious constraints on the size and complexity of computations that can be achieved in practice. Lurk programs are instead provided as data to the universal Lurk interpreter circuit, allowing the resulting language to be Turing-complete without compromising the size of the resulting proof artifacts. Our work describes the design and theory behind Lurk, along with detailing how its implementation of content addressing can be used to sidestep many of the usual concerns of programming zero-knowledge proofs.

ConsensusDay 23: deadline extended to 19 Mar

Wed, 15 Mar 2023 00:00:00 +0000

We have already received a number of excellent submissions for ConsensusDay 23 but have decided to extend the submission deadline in order to accommodate several pending contributions. We will therefore be accepting submissions until the end of Sunday, 19 March, anywhere on earth.

If you have already completed your submission but would like to update it in the extra time available, the system remains open. If you have any questions, send us an email or drop by #consensus on Filecoin slack.

Discourse graphs and the future of science

Mon, 27 Feb 2023 00:00:00 +0000

Interview between Tom Kalil, Chief Innovation Officer of Schmidt Futures, Dr. Evan Miyazono, Research Team Lead at Protocol Labs, and Dr. Matt Akamatsu, Assistant Professor of Biology at the University of Washington.

Tom Kalil: I wanted to talk to Evan and Matt because of their pioneering work on discourse graphs – a potentially new way of communicating and sharing scientific arguments. I think more scientists and funders of science should be aware of these ideas. I’m also interested in the use of graphs for fostering innovation and commercialization of research, such as work by Deep Science Ventures on outcomes graphs.

Below is a copy of the Q&A conducted over email between me, Evan, and Matt.

Individual answers are from Evan Miyazono (EM) and Matt Akamatsu (MA).

What is a discourse graph, and how are scientists beginning to use them?

EM: A discourse graph is a way of structuring and sharing scientific arguments. Each brief note is labeled as a question, claim, or evidence. These notes (nodes) are connected to each other (via edges) in a graph based on their relationships.

Discourse graph schema. Scientific arguments are decomposed into their constituent parts – questions, claims, evidence – and connected into a graph. Evidence can support or oppose a claim.

MA: Discourse graphs currently have two use cases:

Knowledge synthesis and sharing. A modular, reusable alternative to systematic reviews, discourse graphs capture the current lines of evidence that make up the state of knowledge in a given research subfield. Each evidence page contains enough context for the user to assess its validity and relevance, and are used to build novel claims that propose to answer a given research question. Multiple, opposing claims can coexist; since each claim is linked to the underlying evidence, each reader can assess the validity of each claim on its own merit (figure below). Discourse graphs give researchers the means to structure and reuse their literature searches, and importantly distribute the effort in generating and maintaining the database.

Example discourse graph for the question: Are bans an effective way to mitigate antisocial behavior in online forums? (Source: Joel Chan)

Open-source, modular research communication. Adapting the question/claim/evidence schema to ongoing research has allowed us to discretize the scientific research process for modular, collaborative research. At the end of the literature review (above), researchers pose hypotheses (testable claims) which motivate the collection of new data, leading to original results (new evidence) that support evolving claims (figure below). Each unit of new research lives in the discourse graph for context and reuse by other researchers, potential collaborators, science communicators, and funders. This enables gap analysis and facilitates discoverability of interesting problems within the field.

(Left) Schematic of discourse graph usage for original research. (Right) Illustrative historical example.

MA: The Akamatsu Lab shares a discourse graph to pose new research questions, get up to speed on the current state of knowledge for that research question, and find an entry point for useful contributions. Lab members share evidence from articles, working hypotheses, requests for new experiments (called issues), and original results such that we can build on each others’ findings. Contributing to discourse graphs has accelerated students’ onboarding to our research field and given them a structured, discretized means of contributing to research.

(Left) Interactive bulletin board for new contributors to the Akamatsu lab discourse graph. Each element on the right column links to a live query within the discourse graph. (Right) Flow diagram of modular research sharing within the Akamatsu lab discourse graph.

Researchers are generating and sharing¹ their discourse graphs (most with permissioned write, permissionless read access).

How did you get interested in discourse graphs?

EM: My team had invited Prof. Joel Chan to present his research on synthesis of academic literature. He started describing how he’s changed the way he structured his notes based on his research, and when he said “I don’t give new students a stack of papers when they join my lab, I give them access to my notes graph” I formed the immediate conviction that this was the future of scientific collaboration. It was easy to see this as a tool to achieve the paradigm of collaboration presented in Michael Nielsen’s Reinventing Discovery.

MA: Our project is based on the hypothesis that new tools for collaboration can facilitate cultural and organizational change within scientific research communities. I connected with Joel Chan after trying networked note-taking tools to connect my notes for a collaborative project on SARS-CoV-2 cell biology during the pandemic. Our conversations led to the insight that discourse graphs could be applied to our ongoing research, which might facilitate the types of modular, rapid research contributions that I wished were possible for the COVID researcher community.

What do you think are the potential advantages of discourse graphs, relative to the journal article or preprint?

EM: A journal article is usually the culmination of 6-24 months of work. The initial idea or result, which would be usable to others in the field, often is generated in the initial months of the project. The promise of Discourse Graphs (or any robust, graph-based, notes schema) is that useful progress can be shared sooner, with sufficient context for collaborators or with the research community at large. Preprints increase the accessibility, but usually do not increase the availability of research to the extent that it could be said to accelerate science².

DGs provide the opportunity for modular peer review, with clear affordances to ask “does this evidence really support that claim?” or “is this evidence reproducible?” without needing to pass judgment on an entire body of work.

Journal articles and preprints are also required to demonstrate a complete scientific narrative, starting from a research question, generating a hypothesis, designing and running an experiment to test the hypothesis, and generating conclusions. Researchers must perform all steps to receive credit; discourse graphs could enable specialization of labor within the scientific system. While the complete narrative still has clear value, discourse & results graphs enable us to reward efforts both in whole and in part³. Eventually, discourse graphs could also make scientific data more computable (discussed more in later questions).

MA: In my field, journal articles have ballooned to include dozens of results and authors, taking several years from idea to publication. Collaboration is disincentivized because a single first author claims the majority of credit. Individual researchers collect individual results - why not attribute credit for each research contribution? Generating a discourse graph cites atomic units of work – a single result or hypothesis – so that researchers have an ongoing, accurate representation of their work and its impact. Journals’ “author contributions” sections have provided limited value because they are added manually post facto and do not track individual results or intellectual contributions.

While journal articles and preprints center around one or more claims, discourse graphs link to the underlying evidence, which makes it easier for other researchers to make their own claims from the aggregated evidence.

Journal articles tend to select submissions based on perceived novelty of claims, rather than rigor of results. With discourse graphs, we envision that results from a single experiment will gain strength over time by being reproduced by other researchers. Overlapping work will be mutually beneficial rather than adversarial. Hypotheses can be updated as new evidence becomes available. Research output, rather than resembling single-use consumables, will instead be reused in a sustainable scientific ecosystem. These global changes result from the following local changes:

Modular peer review – individual hypotheses can be challenged by others who present additional evidence from the literature or their lab notebooks. Peers can corroborate results by reproducing or flagging for irreproducibility, supporting constructive addition to the work rather than gatekeeping.
Decentralized coordination – within groups, the transparency facilitates useful exchange of research between peers, rather than relying on PIs as the primary sources of plans and ideas. This decentralization raises the agency and responsibility of individual researchers, and facilitates the generation of novel organizational structures for science (e.g. DAOs).
Constraints liberate – an open, structured conversation between researchers from idea to hypothesis to result to claim more closely reflects the intrinsic discovery process (have an idea, get excited about it, share with people who want to help figure out the answer, figure it out together and celebrate the finding) than the current first-author incentive system (have an idea, guard it, toil alone for years in order to claim sole credit for contributing but a piece to a broader collective endeavor).

What do you view as the primary obstacles to the adoption of discourse graphs, such as the incentives for scientists to contribute?

EM: One obstacle for researchers is becoming comfortable sharing results and ideas sooner⁴. Hopefully the sharing of progress immediately and continuously is viewed as an extension of the preprint revolution, rather than a new practice. We need the community to respect notes graphs as artifacts in the scientific literature, which means integrating them into existing workflows and allocating credit (i.e. funding, awards, patents, & citations) based on work done in notes graphs. The current culture in many fields is one of secrecy over openness, which needs to be inverted. This secrecy seems to be driven by zero-sum reputation building, rather than focus on cooperative achievement of shared goals.

MA: Academic career incentives are an obstacle to adoption. DGs encourage researchers to make careful contributions that can be built upon, which may not correlate with current “first past the post” publication incentives. In DGs, impact is assessed by the usage and lasting power of individual results, rather than by article citations or journal reputation⁵.

Many academic researchers (particularly established researchers) will resist moving away from journal articles as career currency. That said, I imagine that DGs and journal articles can coexist for the foreseeable future; in our lab, we will publish some traditional journal articles and co-publish companion DGs, or alternately micropublish individual results or conclusions from our DG as connected modular preprint articles⁶.

EM: Journals currently provide a valuable signal around recognizing certain research as important and making the results salient, and researchers will need new habits and tools to do this at least as well in a world where all science ideas are shared immediately and continuously; therefore, we need to smoothly transition to such a world, still leveraging existing incentives and metrics, like citations and publications. The eventual use of tools leveraging NLP over DGs to filter the most relevant contributions and then match and merge relevant results clearly has capacity to surpass the information salience provided by journals.

Another obstacle is adopting the subtle change in how they take notes⁷. This should be close to the change of using Overleaf or Google Docs instead of emailing .doc attachments to each other, but it is a barrier. There are multiple emerging proof-of-concept tools for generating DGs, like Lateral.io, Samepage.network, and tooling for RoamResearch; however, the user experience for teams is still early days and a bit rough around the edges.

Lastly, DGs do not carry universal context. Much of any single working DG will initially only have enough context for close collaborators to use it. That said, the addition of this context is often much of the beneficial aspects of writing up research results in journals, so we see this as an epistemological limitation on scientific communication made explicit by the graph format, rather than a limit of the format itself. Relatedly, DGs are most useful when they are extensions of a researcher’s (or group’s) ongoing efforts at notetaking and sensemaking; in this way the DG is not an external additional layer of formalization to spend time on, but instead serves as an integral tool for shaping and improving the thinking process, from unstructured to structured thought and communication.

What role could different types of individuals and organizations play in accelerating the adoption of discourse graphs?

EM: Funders could give funding for research contingent on open publishing of structured notes for that research, in support of open science and positive-sum collaborations. However, the idea becomes more compelling if it’s recognized that funders may eventually be more interested in the discourse graph as a research output. These discourse graphs could be used to inform future funding allocation decisions (e.g. pursuing a specific question raised or searching for additional evidence to back a claim). I expect that some research funders and teams will realize a competitive advantage by using research roadmaps (an extension of the discourse graph schema that we’re developing to describe yet-to-be-completed research) to guide their research plans, expectations, timelines, and decisions.

MA: Funders could accept structured hypotheses within discourse graphs as substitutes for grant proposals. Funders can also support developers who are making discourse graph tools easier to use, and allow teams to hire “cybrarians” (knowledge graph enthusiasts who assist in generating and maintaining a team’s DG). It might be fruitful to found an organization focused on improving discourse graph tools for teams. Similarly, assembling/funding new teams or organizations and giving them an empty discourse graph (and research assistant) would give teams the chance to assemble a discourse graph together from scratch.

EM: Publishers could partner in the development and usage of citation conventions and infrastructure to be applied to discourse graphs⁸; longer term, publishers could co-publish discourse graphs alongside or embedded within journal articles, and update their citation metrics to leverage the added value of discourse graphs. Publishers could also create high-profile venues exclusively for work shared in an immediately and continuously open way. In this way, examining the discourse graphs behind the science could also reduce the review burden as we move into an age where large language models dramatically accelerate the rate at which researchers can write manuscripts or reformat them for resubmission.

MA: Researchers could pilot the use of team discourse graphs that are open or will be shared upon publication of resultant journal articles and results. They can provide user feedback to the developers improving the usability of discourse graph tools.

EM: Developers could build tooling for interoperability between different note-taking platforms to facilitate adoption; build tools to use discourse graphs to assist in knowledge synthesis, possibly converting parts of the existing scientific literature corpus into discourse graphs, or using discourse graphs as logical back-ends or training data for NLP & LLMs).

What types of pilots should the research community try? How might we evaluate the efficacy of these pilots?

EM: A structured program could be developed for any of the above roles. Examples could include:

A funder funds research collaborations that pilot discourse graphs for coordination & synthesis; the individual teams pilot discourse graph use internally to measure if it noticeably results in more cross-pollination of research elements
A funder, professional organization, or institution makes a publication community (e.g. PubPub) and, invites teams of researchers; this could be tied to funding opportunities

These pilots should seek to show that discourse graphs are both (a) appealing to use and (b) more effective than other methods of collaboration. The following are a few ideas of how these could be measured (not intended to be either exhaustive or particularly robust):

(a) Test the claim: “Discourse graphs are appealing to use” with

net promoter scores, or other reviews
daily or weekly active users
referrals (fraction of users joining on a collaborator’s recommendation)
number of new nodes per public graph
number of public graphs
number of collaborators per graph

b) Test the claim: “Discourse graphs are more effective than other methods of collaboration” with

citations per discourse graph node vs. final paper (and relative time lag)
subjective evaluation of the rate of progress in subfields where shared notes graphs are used vs. other fields

How do discourse graphs relate to NSF’s support for knowledge networks, such as the Convergence Accelerator on Open Knowledge Networks, which has funded projects like the UCSF-led Biomedical Open Knowledge Network?

EM: Knowledge networks provide unifying venues for data; discourse graphs represent the scientific process and its epistemological uncertainty in the format of a graph.

In this way, we should expect discourse graphs to leverage and, in some ways, extend knowledge networks.

I envision the difference as follows: a paper tomorrow might analyze and draw new insights and conclusions from a knowledge network. These insights and conclusions would be structured as a discourse graph, in such a way that a future LLM might be able to generate a research paper from the discourse graph.

Discourse graphs can identify points of contention within nodes of the knowledge graph, and propose hypotheses that explain some of the contents of the domain-specific knowledge graphs. Most likely the DGs will have simpler and more universal ontology, applicable to most empirical research, while knowledge graphs will have boutique and evolving ontology that is unique to each field or subfield.

What role can large language models play in automating the process of building discourse graphs, and how can the availability of discourse graphs increase the performance of large language models for tasks such as question-answering?

MA: We and others are making early progress using user-annotated papers as training data to help GPT-3 classify discourse graph elements of research papers. In this way, discourse graphs will help LLMs to extract relationships between claims from the existing academic literature. LLMs can also summarize the context for a given piece of evidence and tie a sequence of discourse nodes into a narrative, to help translate between discourse graph content and traditional articles for diverse audiences. As a medium-term aspiration, an LLM could generate new research questions or hypotheses from existing discourse graphs by identifying unseen patterns. Eventually these language models would be able to reason over the claims and evidence in the graph.

EM: Discourse graphs should also be able to provide a source of truth to prevent hallucinations of scientific LLMs. Within my team, we claim that discourse graphs could be effective in reducing hallucinations by LLMs because they leverage LLMs’ focus on syntax in a clever way to impart semantic meaning that humans can later extract.

If you’d like to support the improvement or adoption of discourse graphs in any way - you can reach Evan, Matt, and other members of the Discourse Graph community by emailing discoursegraphs@protocol.ai

Claims are made with implicit reference to some context. MA: so far, we’ve found that our discourse graphs provide sufficient context within teams or groups of collaborators, but we’ll need to prepare a subset of the pages so that there will be enough context for a wider community. ↩︎
Peer review has been shown to 1) not filter out false results (shown by the replication crisis), and2) not be strictly necessary for credit & rigor (preprints + eLife change) ↩︎
EM: I have anecdotal evidence that some materials science researchers who generate the highest quality reagents/devices may struggle to get publications because they specialize in synthesis/fabrication, but they’re never listed first on their collaborator’s high-profile publications. Additionally, NASA’s policy change to immediately open data from JWEST has been met with justifiable push-back by researchers who propose the data collection, because their significant value may go unrewarded in the current paradigm of crediting publications. ↩︎
MA: an intermediate option (which we are using) is to share results within your trusted network (e.g. lab) until it is ready to be shared more widely. ↩︎
MA: I am advocating for impact to be assessed using a qualitative contributor page, reminiscent of the “github contributor” page, that displays which subsequent hypotheses and results rely on your research contributions. Rather than replacing journal impact factor with another metric that people will over-optimize for, such a contributor page could serve as a robust, accurate, and interactive 5-minute (but not 5-second) assessment of a researcher’s impact in their field. ↩︎
EM: I am interested in a more incremental approach, wherein we start applying existing metrics to both discourse graphs and scientific journals in the near future. Eventually, I’d like to see journals acting primarily as a system to highlight impactful work being shared in discourse graphs, with most journal papers acting the way that review articles, survey papers, SoK papers, and metastudies do in their respective fields, and at this point, it’s likely possible to host many different quantitative and qualitative automated metrics that tenure committees or funders could use to estimate individual researchers’ potential or impact for a specific problem. ↩︎
MA: Knowledge workers I know have tended to benefit immediately from switching to graph-based note taking; it has taken 1-2 weeks for people in my lab to adjust (and we are all sharing one graph!) ↩︎
EM: It’s worth noting that peer-reviewed publications have a reputation system to reduce gaming the citation metrics by only counting citations from certain venues, and discourse graphs might naively be susceptible to this gaming. Existing metrics like h-index, however, could make a useful distinction between (1) discourse graph nodes directly cited by papers in approved publications, (2) discourse graph nodes transitively cited by such papers, and (3) discourse graph nodes that do not have a citation connection to such papers. I claim that counting only nodes in (1) and possibly also (2) would prevent this form of gaming. ↩︎

ConsensusDay 23: call for contributions

Wed, 15 Feb 2023 00:00:00 +0000

ConsensusDay is back! We’re returning to our roots and organising a virtual event on 5 June 2023, in a format similar to the 2021 edition.

That means we will not be publishing proceedings this year and therefore welcome both novel contributions and those under review or published elsewhere in the last 12 months. Be sure to check out the call for contributions for full details.

We invite you to submit your work for presentation at the workshop by the 15 March 2023 deadline. If you have any questions, send us an email or drop by #consensus on Filecoin slack.

tlock: Practical timelock encryption from threshold BLS

Mon, 13 Feb 2023 00:00:00 +0000

We present a practical construction and implementation of timelock encryption, in which a ciphertext is guaranteed to be decryptable only after some specified time has passed. We employ an existing threshold network, the League of Entropy, implementing threshold BLS [BLS01, B03] in the context of Boneh and Franklin’s identity-based encryption (IBE). At present this threshold network broadcasts BLS signatures over each round number, equivalent to the current time interval, and as such can be considered a decentralised key holder periodically publishing private keys for the IBE where identities are the round numbers. A noticeable advantage of this scheme is that only the encryptors and decryptors are required to perform any additional cryptographic operations; the threshold network can remain unaware of the TLE and does not have to change to support the scheme. We also release an open-source implementation of our scheme and a live web page that can be used in production now relying on the existing League of Entropy network acting as a distributed public randomness beacon service using threshold BLS signatures.

Pausing PL Research Open Research Grants

Wed, 18 Jan 2023 00:00:00 +0000

Over the 2022 calendar year, Protocol Labs Research received a tremendous amount of participation and submissions for our Open Research Grants program. Having awarded more than $1 million USD for researchers around the globe, we could not be more proud of the success of this program and the quality of our award recepients. However, in light of recent changes to our funding roadmap, PL Research is announcing the immediate pause of our Open Research Grants program for 2023.

We recognize the difficulty of the current funding environment and apologize for the unfortunate timing. This change to our grant offerings is the result of a temporary internal shift towards targeting support for clearly scoped research projects critical to our work at Protocol Labs. To that end, we are happy to report that Protocol Labs still has two open Requests for Proposals (RFPs), RFP-013: Cryptonet network grant and RFP-014: Private retrieval of data, in addition to RFP-X.

Given our continuous interest in understanding the scientific process and how we can leverage our resources and expertise towards improving the research landscape, we will take this opportunity to reflect on how we can increase our impact and broaden our reach before reopening our offerings at a future date.

Background

The PL Research grants program launched in April 2018 when PL Research was first announced as an established team within Protocol Labs. While the team has since grown and changed as various Labs have nucleated, PL Research still oversees the current grant offerings. Open Research Grants were added to grants offering in August 2021, and this past year-and-a-half of receiving your proposals has allowed our labs to partner and collaborate with researchers in countries spanning five continents.

Accomplishments

In addition to funding various research proposals we, under Protocol Labs, were a founding sponsor of the Stanford Center for Blockchain Research(CBR). CBR focuses on research closely aligned with some of our research labs and Open Problems. We encourage you to check out CBR’s seminars, conferences, and call for papers here.

Current offerings

While we are scaling back in the total number of awards offered, we are thrilled to see what collaborations and research proposals come our way.

RFP-X will continue to be open for research projects not otherwise covered by open RFPs. The success rate for RFP-X applications is significantly lower and advised to be the path of last resort.
RFP-013: Cryptonet network grant is accepting applications until 23:59 UTC on July 30, 2023, or earlier if allocated funding is exhausted. Applications are reviewed on a quarterly basis. Additional information about Cryptonet grants can be found here. We encourage you to visit #lobby in the Cryptonet Discord if you are considering applying or have any questions.
RFP-014: Private retrieval of data reviews applications on a monthly basis and is awarded in phases. Phase 1 will be open until 23:59 UTC on March 1, 2023 or earlier if allocated funding is exhausted. We encourage you to visit #private-retrieval in the Lodestar Discord if you’re considering applying or have any questions.

Looking forward

We are grateful to have the opportunity to continue supporting talented researchers through our ongoing RFPs. If you are interested in applying for funding, please submit your proposal using our application management system at https://grants.protocol.ai/ and reach out to research-grants@protocol.ai with any questions.

Generalized Impact Evaluators, A year of experiments and theory

Thu, 12 Jan 2023 00:00:00 +0000

TLDR: Protocol Labs’ Network Funding team is releasing a whitepaper on Impact Evaluators, a funding mechanism designed for nontraditional projects with high uncertainty and high upside. Our goal is to add structure to the ongoing dialogue and share practical implementation advice based on a year of experience. The paper is available here.

Funding vital public goods is hard, and doing so more effectively has massive economic potential. At Protocol Labs, we recognize this, and we’ve dedicated our Network Funding team to pushing forward innovation in how public goods and commons are funded, allocated, and sustained. This has led us to explore mechanisms we call “Impact Evaluators”, a term coined by Juan Benet & Evan Miyazono of Protocol Labs, that builds on concepts from blockchain incentive mechanisms and retroactive funding. Impact Evaluators can provide incentives or market mechanisms for highly effective “impact funding”. Using Impact Evaluators, groups of agents are able to work towards objectives by assessing the impact against those objectives and retrospectively rewarding valuable work. When implemented effectively, an Impact Evaluator guides potential contributors toward particular objectives.

Our team at Protocol Labs believes that this mechanism offers a powerful way to fund important public goods with high upside and high uncertainty. In the past 12 months, we have 1) pushed the theory of Impact Evaluators forward and 2) experimented with several smaller-scale Impact Evaluator tests:

Due to the interest generated by events (Funding the Commons, CryptoEconDay) and the community at large, we’ve seen an uptake in projects experimenting with implementing Impact Evaluators. To share our learnings, we’ve written a whitepaper on Generalized Impact Evaluators. The purpose of this paper is to accelerate the development of Impact Evaluator mechanisms by providing:

Standard language to define and construct Impact Evaluators
A framework to efficiently implement Impact Evaluators

SEE THE WHITEPAPER HERE

As Protocol Labs scales up its deployment of Impact Evaluators, we will build a more robust theoretical basis, which we will publish iteratively. In the meantime, please see the bottom of this article for additional resources related to Impact Evaluators.

This is part of an open source effort - if you would like to contribute, join the Impact Evaluator Discord or email us at network-goods@protocol.ai.

Further materials: Vitalk Buterin, Optimism - Retroactive Public Goods funding Juan Benet - Introduction to Impact Evaluators Juan Benet - Impact Evaluator Design Evan Miyazono - Impact Evaluators Matthew Frehlich - Impact Evaluators: Existential Challenges and Opportunities

Announcing the final award for RFP-011, "Changing the Internet"

Tue, 10 Jan 2023 00:00:00 +0000

PL Research is pleased to announce the final research proposal funded as part of the Changing the Internet Initiative: a project by Maria Apostolaki and Jennifer Rexford of Princeton University’s Electrical and Computer Engineering and Computer Science departments, respectively, to improve the state of the art in routing and edge networking. In the researchers’ own words:

In their unrelenting quest for lower latency, cloud providers are deploying servers closer to their customers and enterprises are adopting paid Network-as-a-Service (NaaS) offerings with performance guarantees.

Unfortunately, these trends contribute to greater industry consolidation, benefiting larger companies and well-served regions while leaving little room for smaller cloud providers and enterprises to flourish. Instead, we argue that the public Internet could offer good enough performance if only edge networks could control wide-area routing.

More concretely, we envision an incrementally deployable system, namely Tango, that allows individual pairs of edge networks (e.g., access, enterprise, and data-center networks) to optimize Internet paths between them without collaboration from the Intenet core. Tango relies on the cooperation between the two edge networks to expose more wide-area paths, and achieve accurate and trustworthy monitoring. While a single cooperative pair of Tango edge networks is already beneficial, it barely scratches the surface of what is possible with a Tango overlay i.e., a set of edge-network PoPs (hardware and/or software) that collaboratively optimize paths between them. If successful our proposal will allow edge networks to make the most out of the public Internet, effectively fighting the industry consolidation and the associated privacy, financial, and political risks.

PL Research is excited to support work on new ideas for building the next generation of the internet. While the Changing the Internet RFP has closed, we encourage interested researchers to check out our open RFPs, particularly RFP-014, “Private Retrieval of Data”, and to reach out to us via email with any questions at research-grants@protocol.ai. Submit your proposal using our application management system at https://grants.protocol.ai/.

Generalized Impact Evaluators

Tue, 10 Jan 2023 00:00:00 +0000

Existing funding systems fail to sufficiently fund public goods and common goods due to insufficient mechanisms for coordinating various agents towards valuable outcomes. Relative to traditional capital systems that scalably organize activity around maximizing financial performance, impact funding remains underdeveloped, especially in the ability to reward high-upside, high-uncertainty work. Here, we propose Impact Evaluators (IEs) as a modular system for coordinating work by measuring, evaluating, and retrospectively rewarding the impact achieved towards specified valuable objectives. We present a structure to define Impact Evaluators as well as design schematics to facilitate their implementation. We then discuss implementation considerations, practical learnings from past experiments, and integration with the broader ecosystem of public goods and commons funding systems.

Enriching Kademlia by partitioning

Tue, 13 Dec 2022 00:00:00 +0000

Decentralizing the Web is becoming an increasingly interesting endeavor that aims at improving user security and privacy as well as providing guaranteed ownership of content. One such endeavor that pushes towards this reality, is Protocol Labs’ Inter-Planetary File System (IPFS) network, that provides a decentralized large scale file system to support the decentralized Web. To achieve this, the IPFS network leverages the Kademlia DHT to route and store pointers to content stored by network members (i.e., peers). However, due to the large number of network peers, content, and accesses, the DHT routing needs to be efficient and quick to enable a decentralized web that is competitive.

In this paper, we present work in progress that aims at improving the Kademlia DHT performance through the manipulation of DHT identifiers by adding prefixes to identifiers. With this, we are able to bias the DHT topological organization towards locality (which can be either geographical or applicational), which creates partitions in the DHT and enables faster and more efficient query resolution on local content. We designed prototypes that implement our proposal, and performed a first evaluation of our work in an emulated network testbed composed of 5000 nodes. Our results show that our proposal can benefit the DHT look up on data with locality with minimal overhead.

Pikachu: Securing PoS blockchains from long-range attacks by checkpointing into Bitcoin PoW using Taproot

Tue, 13 Dec 2022 00:00:00 +0000

Blockchain systems based on a reusable resource, such as proof-of-stake (PoS), provide weaker security guarantees than those based on proof-of-work. Specifically, they are vulnerable to long-range attacks, where an adversary can corrupt prior participants in order to rewrite the full history of the chain. To prevent this attack on a PoS chain, we propose a protocol that checkpoints the state of the PoS chain to a proof-of-work blockchain such as Bitcoin. Our checkpointing protocol hence does not rely on any central authority. Our work uses Schnorr signatures and leverages Bitcoin recent Taproot upgrade, allowing us to create a checkpointing transaction of constant size. We argue for the security of our protocol and present an open-source implementation that was tested on the Bitcoin testnet.

The Filecoin Spacenet goes live

Wed, 30 Nov 2022 00:00:00 +0000

For all of the past year, our team at ConsensusLab has been hard at work pushing the boundaries of Filecoin scalability and devising a framework for horizontal scaling that allows for the seamless spawning of interoperable subnets, as well as a reference implementation of a consensus algorithm suitable for running said subnets.

We are pleased to announce that, as of Monday, the first phase of the Interplanetary Consensus (IPC) project is available for public testing, with the launch of the Filecoin Spacenet. Spacenet is a new test network for IPC, which will initially provide a platform for builders to experiment with Trantor consensus.

FVM support is expected in the coming weeks, unleashing the potential for novel applications and use cases that require lower latency than provided by Filecoin mainnet consensus, and support for subnets will be added next year, enabling elastic capacity and paving the road for deployment in the main network.

We invite you to visit the newly launched Scaling Filecoin website (fil.space) and the Spacenet repository. If you need funds, head to the Spacenet faucet. Please reach out if you’d like to get involved, be it by operating a node, deploying an application, building tooling, or anything else! Development discussions take place on #spacenet (Filecoin Slack).

Oh, one last note! If you’d like to learn more about IPC and Trantor, the recordings from the recent ConsensusLab Summit are now available, and cover these and other projects in detail.

Applications of Interplanetary Consensus