Most existing work on secure multi-party computation (MPC) ignores a key idiosyncrasy of modern communication networks, that there are a limited number of communication paths between any two nodes, many of which might even be corrupted. The problem becomes particularly acute in the information-theoretic setting, where the lack of trusted setups (and the cryptographic primitives they enable) makes communication over sparse networks more challenging. The work by Garay and Ostrovsky [EUROCRYPT’08] on almost-everywhere MPC (AE-MPC), introduced “best-possible security” properties for MPC over such incomplete networks, where necessarily some of the honest parties may be excluded from the computation. In this work, we provide a universally composable definition of almost-everywhere security , which allows us to automatically and accurately capture the guarantees of AE-MPC (as well as AE-communication, the analogous “best-possible security” version of secure communication) in the Universal Composability (UC) framework of Canetti. Our results offer the first simulation-based treatment of this important but under-investigated problem, along with the first simulation-based proof of AE-MPC. To achieve that goal, we state and prove a general composition theorem, which makes precise the level or “quality” of AE-security that is obtained when a protocol’s hybrids are replaced with almost-everywhere components.

Electronic voting has occupied a large part of the cryptographic protocols literature. The recent reality of blockchains - in particular, their need for online governance mechanisms - has brought new parameters and requirements to the problem. We identify the key requirements of a blockchain governance mechanism, namely correctness (including eliminative double votes), voter anonymity, and traceability, and investigate mechanisms that can achieve them with minimal interaction and under assumptions that fit the blockchain setting. First, we define a signature-like primitive, which we term sharp anonymous multisignatures (in short, ♯AMS) that tightly meets the needs of blockchain governance. In a nutshell, ♯AMSs allow any set of parties to generate a signature, e.g., on a proposal to be voted upon, which, if posted on the blockchain, hides the identities of the signers/voters but reveals their number. This can be seen as a (strict) generalization of threshold ring signatures (TRS). We next turn to constructing such ♯AMSs and using them in various governance scenarios - e.g., single vote vs. multiple votes per voter. In this direction, although the definition of TRS does not imply ♯AMS, one can compile some existing TRS constructions into ♯AMS. This raises the question: What is the TRS structure that allows such a compilation? To answer the above, we devise templates for TRSs. Our templates encapsulate and abstract the structure that allows for the above compilation - most of the TRS schemes that can be compiled into ♯AMS are, in fact, instantiations of our template. This abstraction makes our template generic for instantiating TRSs and ♯AMSs from different cryptographic assumptions (e.g., DDH, LWE, etc.). One of our templates is based on chameleon hashes, and we explore a framework of lossy chameleon hashes to understand their nature fully. Finally, we turn to how ♯AMS schemes can be used in our applications. We provide fast (in some cases non-interactive) ♯AMS-based blockchain governance mechanisms for a wide spectrum of assumptions on the honesty (semi-honest vs malicious) and availability of voters and proposers.

Common blockchain protocols are monolithic, i.e., their security relies on a single assumption, e.g., honest majority of hashing power (Bitcoin) or stake (Cardano, Algorand, Ethereum). In contrast, so-called optimistic approaches (Thunderella, Meshcash) rely on a combination of assumptions to achieve faster transaction liveness. We revisit, redesign, and augment the optimistic paradigm to a tiered approach. Our design assumes a primary (Tier 1) and a secondary (Tier 2, also referred to as fallback) blockchain, and achieves full security also in a tiered fashion: If the assumption underpinning the primary chain holds, then we guarantee safety, liveness and censorship resistance, irrespectively of the status of the fallback chain. And even if the primary assumption fails, all security properties are still satisfied (albeit with a temporary slow down) provided the fallback assumption holds. To our knowledge, no existing optimistic or tiered approach preserves both safety and liveness when any one of its underlying blockchain (assumptions) fails. The above is achieved by a new detection-and-recovery mechanism that links the two blockchains, so that any violation of safety, liveness, or censorship resistance on the (faster) primary blockchain is temporary - it is swiftly detected and recovered on the secondary chain - and thus cannot result in a persistent fork or halt of the blockchain ledger. We instantiate the above paradigm using a primary chain based on proof of reputation (PoR) and a fallback chain based on proof of stake (PoS). Our construction uses the PoR and PoS blockchains in a mostly black-box manner - where rather than assuming a concrete construction we distil abstract properties on the two blockchains that are sufficient for applying our tiered methodology. In fact, choosing reputation as the resource of the primary chain opens the door to an incentive mechanism - which we devise and analyze - that tokenizes reputation in order to deter cheating and boost participation (on both the primary/PoR and the fallback/PoS blockchain). As we demonstrate, such tokenization in combination with interpreting reputation as a built-in system-wide credit score, allows for embedding in our two-tiered methodology a novel mechanism which provides collateral-free, multi-use payment-channel-like functionality where payments can be instantly confirmed.

A two-party fuzzy private set intersection (PSI) protocol between Alice and Bob with input sets A and B allows Alice to learn nothing more than the points of Bob that are ''δ-close'' to its points in some metric space dist . More formally, Alice learns only the set {b | dist (a,b) ≤ δ, a ∈ A, b ∈ B} for a predefined threshold δ and distance metric dist, while Bob learns nothing about Alice's set. Fuzzy PSI is a valuable privacy tool in scenarios where private set intersection needs to be computed over imprecise or measurement-based data, such as GPS coordinates or healthcare data. Previous approaches to fuzzy PSI rely on asymmetric cryptographic primitives, generic two-party computation (2PC) techniques like garbled circuits, or function secret sharing methods, all of which are computationally intensive and lead to poor concrete efficiency. This work introduces a new modular framework for fuzzy PSI, primarily built on efficient symmetric key primitives. Our framework reduces the design of efficient fuzzy PSI to a novel variant of oblivious transfer (OT), which we term distance-aware random OT (da-ROT). This variant enables the sender to obtain two random strings (r0, r1), while the receiver obtains one of these values rb, depending on whether the receiver's input keyword a and the sender's input keyword b are close in some metric space i.e., dist (a,b) ≤ δ. The da-ROT can be viewed as a natural extension of traditional OT, where the condition (choice bit) is known to the receiver. We propose efficient constructions for da-ROT based on standard OT techniques tailored for small domains, supporting distance metrics such as the Chebyshev norm, the Euclidean norm, and the Manhattan norm. By integrating these da-ROT constructions, our fuzzy PSI framework achieves up to a 14× reduction in communication cost and up to a 54× reduction in computation cost compared to previous state-of-the-art protocols, across input set sizes ranging from 28 to 216. Additionally, we extend our framework to compute fuzzy PSI cardinality and fuzzy join from traditional PSI-related functionalities. All proposed protocols are secure in the semi-honest model.

Permissionless distributed ledgers, such as those arising from blockchain protocols, have been touted as the centerpiece of an upcoming security-critical information technology infrastructure. Their basic properties-consistency and liveness-can be guaranteed under specific constraints on the resources available to an adversary relative to the resources of the participants that follow the protocol. Given their permissionless participation convention and their intended long-livedness, a critical open security question is their behavior-and potential resilience-to temporary spikes in adversarial resources. In this work we give the first thorough treatment of the self-healing properties of Nakamoto ledgers, addressing both proof-of-work (PoW) and proof-of-stake (PoS) protocols. First, we present a unified model that allows us to define self-healing for both of these protocol classes. Then we provide a formal analysis establishing self-healing with respect to both consistency and liveness in both classes, quantifying the resulting vulnerability period as a function of the magnitude of the spike. Finally, we provide numerical simulations giving explicit quantitative bounds relevant for practice.

We address the problem of Reliable Broadcast in asynchronous message-passing systems with n nodes, of which up to t are malicious (faulty), in addition to a message adversary that can drop some of the messages sent by correct (non-faulty) nodes. We present a Message-Adversary-Tolerant Byzantine Reliable Broadcast (MBRB) algorithm that communicates O(|m|+nκ) bits per node, where |m| represents the length of the application message and κ = Ω(log n) is a security parameter. This communication complexity is optimal up to the parameter κ. This significantly improves upon the state-of-the-art MBRB solution (Albouy, Frey, Raynal, and Taïani, TCS 2023), which incurs communication of O(n|m|+n²κ) bits per node. Our solution sends at most 4n² messages overall, which is asymptotically optimal. Reduced communication is achieved by employing coding techniques that replace the need for all nodes to (re-)broadcast the entire application message m. Instead, nodes forward authenticated fragments of the encoding of m using an erasure-correcting code. Under the cryptographic assumptions of threshold signatures and vector commitments, and assuming n > 3t+2d, where the adversary drops at most d messages per broadcast, our algorithm allows at least 𝓁 = n - t - (1 + ε)d (for any arbitrarily low ε > 0) correct nodes to reconstruct m, despite missing fragments caused by the malicious nodes and the message adversary.

Differential privacy (DP) is a key tool in privacy-preserving data analysis. Yet it remains challenging for non-privacy-experts to prove the DP of their algorithms. We propose a methodology for domain experts with limited data privacy background to empirically estimate the privacy of an arbitrary mechanism. Our Eureka moment is a new link— which we prove—between the problems of DP parameter-estimation and Bayes optimal classifiers in ML, which we believe can be of independent interest. Our estimator uses this link to achieve two desirable properties: (1) black-box, i.e., it does not require knowledge of the underlying mechanism, and (2) it has a theoretically-proven accuracy, depending on the underlying classifier used, allowing plug-and-play use of different classifiers.More concretely, motivated by the impossibility of the above task for unrestricted input domains (which we prove), we introduce a natural, application-inspired relaxation of DP which we term relative DP. Intuitively, relative DP defines a mechanism's privacy relative to an input set$\mathcal{T}$, circumventing the above impossibility when $\mathcal{T}$ is finite. Importantly, it preserves the key intuitive privacy guarantee of DP while enjoying a number of desirable DP properties—scalability, composition, and robustness to post-processing. We then devise a black-box poly-time (ε, δ)-relative DP estimator for any poly-size $\mathcal{T}$— the first privacy estimator to support mechanisms with large output spaces while having tight accuracy bounds. As a result of independent interest, we generalize our theory to develop the first Distributional Differential Privacy (DDP) estimator.We benchmark our estimator in a proof-of-concept implementation. First, using kNN as the classifier we show that our method (1) produces a tight, analytically computed (ε,δ)-DP trade-off of low-dimensional Laplace and Gaussian mechanisms—the first to do so, (2) accurately estimates the privacy spectrum of DDP mechanisms, and (3) can verify a DP mechanism's implementations, e.g., Sparse Vector Technique, Noisy Histogram, and Noisy max. Our implementation and experiments demonstrate the potential of our framework, and highlight its computational bottlenecks in estimating DP, e.g., in terms of the size of δ and the data dimensionality. Our second, neural-network-based instantiation makes a first step in showing that our method can be extended to mechanisms with high-dimensional outputs.

International audience

Abstract Bitcoin is one of the most prominent examples of a distributed cryptographic protocol that is extensively used in reality. Nonetheless, existing security proofs are property-based, and as such they do not support composition. In this work, we put forth a universally composable treatment of the Bitcoin protocol. We specify the goal that Bitcoin aims to achieve as an instance of a parameterizable ledger functionality and present a UC abstraction of the Bitcoin blockchain protocol. Our ideal functionality is weaker than the first proposed candidate by Kiayias, Zhou, and Zikas [EUROCRYPT’16], but unlike the latter suggestion, which is arguably not implementable by the UC Bitcoin protocol, we prove that the one proposed here is securely UC-realized by the protocol assuming access to a global clock, to model time-based executions, a random oracle, to model hash functions, and an idealized network, to model message dissemination. We further show how known property-based approaches can be cast as special instances of our treatment and how their underlying assumptions can be cast in UC as part of the setup functionalities and without restricting the environment or the adversary.

Recent years have witnessed significant advances in programming technology for multi-party computation (MPC), bringing MPC closer to practice and wider applicability. Typical MPC programming frameworks focus on either front-end language design (e.g., Wysteria, Viaduct, SPDZ), or back-end protocol design and implementation (e.g., ABY, MOTION, MP-SPDZ).

Standardized Ethereum tokens, e.g., ERC-20 tokens, have become the norm in fundraising (through ICOs) and kicking off blockchain-based DeFi applications. However, they require the user's wallet to hold both tokens and ether to pay the gas fee for making a transaction. This makes for a cumbersome user experience, and complicates, from the user perspective, the process of transitioning to a different smart-contract enabled blockchain, or to a newly launched blockchain. We formalize, instantiate, and analyze in a composable manner a system that we call Etherless Ethereum Tokens (in short, EETs), which allows the token users to transact in a closed-economy manner, i.e., having only tokens on their wallet and paying any transaction fees in tokens rather than Ether/Gas. In the process, we devise a methodology for capturing Ethereum token-contracts in the Universal Composability (UC) framework, which can be of independent interest.

Collusion-free (CF) and collusion-preserving (CP) protocols enrich the standard security offered by multi-party computation (MPC), to tackle settings where subliminal communication is undesirable. However, all existing solutions make arguably unrealistic assumptions on setups, such as physical presence of the parties, access to physical envelopes, or extreme isolation, where the only means of communication is a star-topology network. The above state of affairs remained a limitation of such protocols, which was even reinforced by impossibility results. Thus, for years, it has been unclear if and how the above setup assumptions could be relaxed towards more realistic scenarios. Motivated also by the increasing interest in using hardware tokens for cryptographic applications, in this work we provide the first solution to collusion preserving computation which uses weaker and more common assumptions than the state of the art, i.e., an authenticated broadcast functionality and access to honestly generated trusted hardware tokens. We prove that our protocol is collusion-preserving (in short, CP) secure as long as no parties abort. In the case of an aborting adversary, our protocol still achieves standard (G)UC security with identifiable (and unanimous) abort. Leveraging the above identifiability property, we augment our protocol with a penalization scheme which ensures that it is not profitable to abort, thereby obtaining CP security against incentive-driven attackers. To define (and prove) this latter result, we combine the Rational Protocol Design (RPD) methodology by Garay et al. [FOCS 2013] with the CP framework of Alwen et al. [CRYPTO 2012] to derive a definition of security in the presence of incentive-driven local adversaries which can be of independent interest. Similar to existing CP/CF solutions, our protocol preserves, as a fallback, security against monolithic adversaries, even when the setup (i.e., the hardware tokens) is compromised or corrupted. In addition, our fallback solution achieves identifiable and unanimous abort, which we prove are impossible in previous CP solutions.

Differential privacy has been widely applied to provide privacy guarantees by adding random noise to the function output. However, it inevitably fails in many high-stakes voting scenarios, where voting rules are required to be deterministic. In this work, we present the first framework for answering the question:“How private are commonly-used voting rules?" Our answers are two-fold. First, we show that deterministic voting rules provide sufficient privacy in the sense of distributional differential privacy (DDP). We show that assuming the adversarial observer has uncertainty about individual votes, even publishing the histogram of votes achieves good DDP. Second, we introduce the notion of exact privacy to compare the privacy preserved in various commonly-studied voting rules, and obtain dichotomy theorems of exact DDP within a large subset of voting rules called generalized scoring rules.

Multi-party computation (MPC) protocols have been extensively optimized in an effort to bring this technology to practice, which has already started bearing fruits.The choice of which MPC protocol to use depends on the computation we are trying to perform.Protocol mixing is an effective black-box -with respect to the MPC protocols-approach to optimize performance.Despite, however, considerable progress in the recent years existing works are heuristic and either give no guarantee or require an exponential (bruteforce) search to find the optimal assignment, a problem which was conjectured to be NP hard.We provide a theoretically founded approach to optimal (MPC) protocol assignment, i.e., optimal mixing, and prove that under mild and natural assumptions, the problem is tractable both in theory and in practice for computing best two-out-of-three combinations.Concretely, for the case of two protocols, we utilize program analysis techniques-which we tailor to MPC-to define a new integer program, which we term the Optimal Protocol Assignment (in short, OPA) problem whose solution is the optimal (mixed) protocol assignment for these two protocols.Most importantly, we prove that the solution to the linear program corresponding to the relaxation of OPA is integral, and hence is also a solution to OPA.Since linear programming can be efficiently solved, this yields the first efficient protocol mixer.We showcase the quality of our OPA solver by applying it to standard benchmarks from the mixing literature.Our OPA solver can be applied on any two-out-of-three protocol combinations to obtain a best two-out-of-three protocol assignment.

We present Ouroboros Crypsinous, the first formally analyzed privacy-preserving proof-of-stake blockchain protocol. To model its security we give a thorough treatment of private ledgers in the (G)UC setting that might be of independent interest. To prove our protocol secure against adaptive attacks, we introduce a new coin evolution technique relying on SNARKs and key-private forward secure encryption. The latter primitive-and the associated construction-can be of independent interest. We stress that existing approaches to private blockchain, such as the proof-of-work-based Zerocash are analyzed only against static corruptions.

We present a novel Proof-of-Stake (PoS) protocol, Ouroboros Genesis, that enables parties to safely join (or rejoin) the protocol execution using only the genesis block information. Prior to our work, PoS protocols either required parties to obtain a trusted "checkpoint" block upon joining and, furthermore, to be frequently online or required an accurate estimate of the number of online parties to be hardcoded into the protocol logic. This ability of new parties to "bootstrap from genesis" was a hallmark property of the Bitcoin blockchain and was considered an important advantage of PoW-based blockchains over PoS-based blockchains since it facilitates robust operation in a setting with dynamic availability, i.e., the natural setting---without external trusted objects such as checkpoint blocks---where parties come and go arbitrarily, may join at any moment, or remain offline for prolonged periods of time. We prove the security of Ouroboros Genesis against a fully adaptive adversary controlling less than half of the total stake in a partially synchronous network with unknown message delay and unknown, varying levels of party availability. Our security proof is in the Universally Composable setting assuming the most natural abstraction of a hash function, known as the strict Global Random Oracle (ACM-CCS 2014); this highlights an important advantage of PoS blockchains over their PoW counterparts in terms of composability with respect to the hash function formalisation: rather than a strict GRO, PoW-based protocol security requires a "local" random oracle. Finally, proving the security of our construction against an adaptive adversary requires a novel martingale technique that may be of independent interest in the analysis of blockchain protocols.

Direct Memory Access (DMA) attacks can allow attackers to access memory directly, bypassing OS supervision or software protections. In this work, we put forth and benchmark a cryptographically secure attestation scheme, which detects DMA attacks. In fact, our scheme detects any attack in a more general class of attacks which we call "direct injection". We prove security of our scheme under a realistic machine model which extends in a non-trivial manner a cryptographic model proposed by Lipton, Ostrovsky, and Zikas (ICALP 2016.) Despite the fact that our scheme, in its current form, protects against write-only attacks, both our security model and our scheme can be extended to allow the attacker to have additional read access to memory---thereby capturing leakage---as well as detecting more types of memory corruptions such as bit flips.

Traditional protocols for secure multi-party computation among n parties communicate at least a linear (in n) number of bits, even when computing very simple functions. In this work we investigate the feasibility of protocols with sublinear communication complexity. Concretely, we consider two clients, one of which may be corrupted, who wish to perform some “small” joint computation using n servers but without any trusted setup. We show that enforcing sublinear communication complexity drastically affects the feasibility bounds on the number of corrupted parties that can be tolerated in the setting of information-theoretic security. We provide a complete investigation of security in the presence of semi-honest adversaries—static and adaptive, with and without erasures—and initiate the study of security in the presence of malicious adversaries. For semi-honest static adversaries, our bounds essentially match the corresponding bounds when there is no communication restriction—i.e., we can tolerate up to $$t < (1/2 -\epsilon )n$$ corrupted parties. For the adaptive case, however, the situation is different. We prove that without erasures even a small constant fraction of corruptions is intolerable, and—more surprisingly—when erasures are allowed, we prove that $$t < (1 - \sqrt{0.5} - \epsilon )n$$ corruptions can be tolerated, which we also show to be essentially optimal. The latter optimality proof hinges on a new treatment of probabilistic adversary structures that may be of independent interest. In the case of active corruptions in the sublinear communication setting, we prove that static “security with abort” is feasible when $$t < (1/2 - \epsilon )n$$ , namely, the bound that is tight for semi-honest security. All of our negative results in fact rule out protocols with sublinear message complexity.

An important benchmark for multi-party computation protocols (MPC) is their round complexity. For several important MPC tasks, (tight) lower bounds on the round complexity are known. However, for some of these tasks, such as broadcast, the lower bounds can be circumvented when the termination round of every party is not a priori known, and simultaneous termination is not guaranteed. Protocols with this property are called probabilistic-termination (PT) protocols. Running PT protocols in parallel affects the round complexity of the resulting protocol in somewhat unexpected ways. For instance, an execution of m protocols with constant expected round complexity might take O(log m) rounds to complete. In a seminal work, Ben-Or and El-Yaniv (Distributed Computing '03) developed a technique for parallel execution of arbitrarily many broadcast protocols, while preserving expected round complexity. More recently, Cohen et al. (CRYPTO '16) devised a framework for universal composition of PT protocols, and provided the first composable parallel-broadcast protocol with a simulation-based proof. These constructions crucially rely on the fact that broadcast is ``privacy free,'' and do not generalize to arbitrary protocols in a straightforward way. This raises the question of whether it is possible to execute arbitrary PT protocols in parallel, without increasing the round complexity. In this paper we tackle this question and provide both feasibility and infeasibility results. We construct a round-preserving protocol compiler, secure against a dishonest minority of actively corrupted parties, that compiles arbitrary protocols into a protocol realizing their parallel composition, while having a black-box access to the underlying protocols. Furthermore, we prove that the same cannot be achieved, using known techniques, given only black-box access to the functionalities realized by the protocols, unless merely security against semi-honest corruptions is required, for which case we provide a protocol.

Protecting software from malware injection is one of the biggest challenges of modern computer science. Despite intensive efforts by the scientific and engineering community, the number of successful attacks continues to increase. This work sets first footsteps towards a provably secure investigation of malware detection. We provide a formal model and cryptographic security definitions of attestation for systems with dynamic memory, and suggest novel provably secure attestation schemes. The key idea underlying our schemes is to use the very insertion of the malware itself to allow for the systems to detect it. This is, in our opinion, close in spirit to the quantum Observer Effect. The attackers, no matter how clever, no matter when they insert their malware, change the state of the system they are attacking. This fundamental idea can be a game changer. And our system does not rely on heuristics; instead, our scheme enjoys the unique property that it is proved secure in a formal and precise mathematical sense and with minimal and realistic CPU modification achieves strong provable security guarantees. We envision such systems with a formal mathematical security treatment as a venue for new directions in software protection.

The vast majority of works on secure multi-party computation (MPC) assume a full communication pattern: every party exchanges messages with all the network participants over a complete network of point-to-point channels. This can be problematic in modern large scale networks, where the number of parties can be of the order of millions, as for example when computing on large distributed data.

Security of distributed cryptographic protocols usually requires privacy (inputs of the honest parties remain hidden), correctness (the adversary cannot improperly affect the outcome), and fairness (if the adversary learns the output, all honest parties do also). Cleve's seminal result (STOC '86) implies that satisfying these properties simultaneously is impossible in the presence of dishonest majorities, and led to several proposals for relaxed notions of fairness. While these works also suggest completeness results (i.e., the ability to design protocols which achieve their fairness notion), their assessment is typically of an all-or-nothing nature. In this work we put forth a new approach for defining relaxed fairness guarantees that allows for a quantitative comparison between protocols with regard to the level of fairness they achieve. The basic idea is to use an appropriate utility function to express the preferences of an adversary who wants to violate fairness. We also show optimal protocols with respect to our notion, in both the two-party and multi-party settings.

Universally composable (UC) protocols retain their security properties even when run concurrently alongside arbitrary other protocols. Unfortunately, it is known that UC multiparty computation (for general functionalities, and without assuming honest majority) is impossible without some form of setup. To circumvent this impossibility, various complete setup assumptions have been proposed. With only a few exceptions, past work has viewed these setup assumptions as being implemented by some ideal, incorruptible entity. Any such entity is thus a single point of failure, and security fails catastrophically in case the setup entity is subverted by an adversary. We propose here a clean, general, and generic approach for distributing trust among m arbitrary setups, by modeling potential corruption of setups within the UC framework, where such corruption might be fail-stop, passive, or arbitrary and is in addition to possible corruption of the parties themselves. We show several feasibility and impossibility results in this model, for different specifications of the corruptible sets. For example, we show that given m complete setups, up to t of which might be actively corrupted in an adaptive manner, general multiparty computation with no honest majority is possible if and only if t < m/2.

Existing work on "rational cryptographic protocols" treats each party (or coalition of parties) running the protocol as a selfish agent trying to maximize its utility. In this work we propose a fundamentally different approach that is better suited to modeling a protocol under attack from an external entity. Specifically, we consider a two-party game between an protocol designer and an external attacker. The goal of the attacker is to break security properties such as correctness or privacy, possibly by corrupting protocol participants; the goal of the protocol designer is to prevent the attacker from succeeding. We lay the theoretical groundwork for a study of cryptographic protocol design in this setting by providing a methodology for defining the problem within the traditional simulation paradigm. Our framework provides ways of reasoning about important cryptographic concepts (e.g., adaptive corruptions or attacks on communication resources) not handled by previous game-theoretic treatments of cryptography. We also prove composition theorems that-for the first time-provide a sound way to design rational protocols assuming "ideal communication resources" (such as broadcast or authenticated channels) and then instantiate these resources using standard cryptographic tools. Finally, we investigate the problem of secure function evaluation in our framework, where the attacker has to pay for each party it corrupts. Our results demonstrate how knowledge of the attacker's incentives can be used to circumvent known impossibility results in this setting.