Background & Biography
Before his current academic appointment, Fisch pursued doctoral research at Stanford with Dan Boneh. His research biography describes a recurring concern: checking work done by an outside service while limiting what that service learns. That includes calculations, remotely held information and the fairness of selection procedures. He approaches these questions through several disciplines, bringing the study of incentives alongside methods for protecting information and coordinating computers.
Espresso Systems lists him as its chief executive. His faculty position has a listed New Haven office, which supplies this profile’s country association without implying citizenship or residence.
Notable Contributions
In 2017, Fisch, Rafael Pass and abhi shelat studied how participants should divide the proceeds of pooled mining. Their analysis asks a more precise question than whether combining resources makes earnings smoother: which payment rule best serves participants when they dislike uncertainty and value earlier payments differently from later ones? The authors model those preferences explicitly and restrict their central analysis to arrangements distributing earned rewards, without an outside financier absorbing the risk. For a particular mathematical description of preferences, they show that a geometric payment schedule can maximize participants’ welfare after the pool has settled into its continuing pattern, with carefully selected settings. The result therefore holds within that model and its settings. It does not identify an unconditional best pool for every miner. The paper also explains why treating each contributed share separately is an approximation suited to small contributors within a large pool, making that boundary important when interpreting the result.
A 2018 survey with Dan Boneh and Benedikt Bünz examined constructions developed separately by Pietrzak and by Wesolowski. The collaborators compared how the two approaches establish that a deliberately lengthy calculation was completed correctly. Their contribution included an additional computational argument supporting the security of one approach and an examination of assumptions underlying both. The comparison distinguishes the work of making evidence from the work of checking it: a smaller certificate or quicker check can come with different demands on the party producing it. The analysis also considers how additional memory or processors affect that preparation. Security remains tied to particular mathematical assumptions, whose relationship the authors investigate. This is collaborative analysis of constructions credited to other researchers. It should be read as a guide to their differences and justification, without assigning their invention to the survey authors.
His March 2024 proposal distinguished an assurance given by a block proposer from one backed by a committee. It examined how rapid commitments about transactions could coexist with interactions involving Ethereum. A central condition was the earlier base-chain history on which a transaction depended: an assurance could survive some later reversals while still failing if its own required history disappeared. That distinction makes the dependency of each promise important, beyond a headline confirmation time. The proposal also acknowledged an availability problem. Requiring the committee’s approval could stop progress when too many participants were offline, even if the underlying chain continued working. A timeout and fallback were discussed as possible choices for rollups. These are properties and trade-offs of a dated design, rather than measurements of a currently operating service.
A separate market-design article, coauthored with Benedikt Bünz and Ellie Davidson, examined the sale of sequencing opportunities across groups of rollups. Its allocation problem concerns which combinations should be assigned together and how proceeds should be divided. A rollup could specify a minimum acceptable price, with payments related to what its own opportunity could attract individually. To reduce repeated wins by one bidder, the authors developed a lottery variant in which ticket purchases determine chances of receiving an assignment. This choice involved a possible revenue trade-off in exchange for a wider spread of winners. Their analysis explicitly left strategic behavior incompletely resolved: it did not claim that the arrangement always makes honest bidding each participant’s best strategy. The proposed pricing adjustments and equilibrium arguments remained subjects for further examination.
A 2024 manuscript written with Wenhao Wang, Lulu Zhou, Aviv Yaish, Fan Zhang and Benjamin Livshits studied compensation for the computing needed to generate cryptographic evidence. Here the suppliers perform costly work, so the design connects their payments to fees paid by customers. A shared rate per unit of work is based on the accepted transactions’ average fee. The mechanism also imposes a ceiling on each batch, intended to limit opportunities for customers and suppliers to arrange profitable side deals. Setting that ceiling is consequential: too little permitted work wastes available resources, while too much may exceed what eligible suppliers can deliver. The paper identifies unresolved vulnerabilities involving coordinated suppliers and the creation of extra identities. It presents a preliminary mechanism with assumptions and unresolved questions, rather than a complete operating market.
A university report in July 2026 described funding for his research on outsourcing calculations while protecting the inputs and checking the answer. The project aims to make encrypted processing work more efficiently with mathematical evidence of correct execution. Its research directions include handling different numerical representations, integrating the two capabilities throughout a processing pipeline, and tailoring methods to particular demanding tasks. Those are prospective objectives for funded work. The report does not establish that a generally deployable solution has already been delivered.
Why Katharos.ai Follows Ben
Katharos.ai follows the connection in his research agenda between confidentiality and accountability. It offers a useful perspective on digital infrastructure: protecting information and enabling scrutiny are design problems that must be addressed together.
The infrastructure proposals connect two distinct questions: what assurances a system can support and how opportunities to operate it should be allocated.
Areas to Follow
- Transaction assurances: follow how later designs specify dependencies, committee availability and fallback behavior, and what independent operating evidence establishes about those conditions.
- Supplier incentives: watch research on adjusting batch limits and resisting coordinated bids or extra identities without undermining the economics of evidence production.
- Confidential outsourcing: track published methods, evaluations and task-specific implementations from the funded research, with attention to the costs of combining secrecy and checking.
Reporting Scope
Coverage focuses on documented professional work, research arguments and the assumptions attached to proposed systems. Collaborative findings retain their shared attribution. A design’s stated goal is assessed separately from evidence of delivery. This profile offers no recommendation about purchasing tokens, selecting investments or relying on a particular service.
Areas of Influence
His comparative cryptography work contributes to evaluating which assumptions support a verification method and where the burden of producing evidence falls.
The market research likewise places payment rules and supplier behavior within the technical design question.
Photo: Yale University
Katharos.ai publishes this profile for information, based on the sources listed. Inclusion is not an endorsement of the individual, their organizations, or any investment product.
Source notice: Information and images are attributed to the external sources listed. Katharos.ai does not endorse the subjects or claims described and does not independently certify the authenticity of every source statement or image.