Pricing Pandora's Boxes: Revenue Maximization in Sequential Information Acquisition

Dimitris Christou; Shuchi Chawla; Trung Dang; Zhiyi Huang

REVIEW 3 major objections 3 minor 23 references

Pricing Pandora's Boxes: Revenue Maximization in Sequential Information Acquisition

T0 review · 3 major / 3 minor · reviewed 2026-07-31 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims that for the Pandora's Box Pricing problem, posting prices that equalize the Weitzman indices of all alternatives always yields expected revenue within a factor of 4 of the optimal revenue, and that this simple rule is exa

desk verdict A natural new problem, a clean and sound 4-approximation theorem, and a couple of small fixable blemishes on the periphery. read the letter →

arxiv 2607.23359 v1 pith:3VXA65WX submitted 2026-07-25 cs.DS

Shuchi Chawla , Dimitris Christou , Trung Dang , Zhiyi Huang This is my paper

classification cs.DS MSC 91B2668Q25

keywords WeitzmanindexPandora'sboxposted-pricemechanismrevenuemaximizationsequentialsearchapproximationalgorithmuniform-indexpricingoptionalinspection

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper studies a seller who posts a price for inspecting each of several alternatives with privately observed values, while the buyer inspects them sequentially in Weitzman index order and stops optimally. The main result is that a simple, efficiently computable pricing rule—setting prices so that all alternatives share the same Weitzman index—extracts at least a quarter of the optimal revenue in every instance. For several natural special cases (two alternatives, identical distributions, two-point supports, and monotone hazard rate distributions), this uniform-index rule actually achieves the optimum. The paper also shows that equalizing prices themselves, rather than indices, can be a log n factor worse than optimal, and gives positive results for an optional-inspection variant. If true, this means near-optimal information selling does not require complex mechanisms or adaptive pricing.

What carries the argument

The Weitzman index g_i of an alternative, defined by p_i = E[(X_i − g_i)^+], is the reservation value that orders the buyer's inspections: the buyer inspects in decreasing g_i and stops once the best observed value exceeds the next index. The paper parameterizes prices by indices and studies uniform-index prices p(g) with all g_i = g; its revenue formula Rev(I,g) = Σ E[(X_i−g)^+] ∏_{j<i} F_j(g) is the central expression analysed in Theorem 3. The support lemma pins the optimal uniform index to the set {0} ∪ ∪_i supp(X_i), turning the search for the best uniform index into a one-dimensional enumeration.

What would settle it

Compute, for a concrete instance with say three independent Bernoulli alternatives, the exact optimal revenue and the best uniform-index revenue using the paper's closed-form expressions; if any instance has OPT > 4 max_g Rev(I,g), Theorem 3 is false. Conversely, finding an instance with ratio above 1.14 would merely tighten the known gap, not refute the approximation claim.

Watch

Extended reading notes

Core claim

The central discovery is that the seller's revenue decomposes along the buyer's inspection order: split alternatives at the first index k whose inspection probability drops below 1/2. The revenue from alternatives after k is at most half the optimum, and the revenue from alternatives before k does not shrink by more than a factor of 2 when all their indices are lowered to g_k. Setting the uniform index equal to g_k therefore captures at least a quarter of the optimal revenue, and the proof goes through by monotonicity of the inspection-probability products λ_i = ∏_{j<i} F_j(g_i). The paper further shows the optimal uniform index always lies in the union of the supports of the alternatives, m

Load-bearing premise

The revenue formula assumes the buyer breaks ties among equal-index alternatives in the seller's favor; if ties were broken adversarially, or if the alternatives were not independent, the monotonicity of the inspection probabilities—and with it the 4-approximation proof—would fail.

Editorial extensions

If this is right

A seller can compute near-optimal posted prices for arbitrary independent distributions in polynomial time when the distributions are discrete, without solving the buyer's full adaptive problem.
The 4-factor guarantee extends to settings with seller costs, outside options, and uniform-matroid selection constraints, so the uniform-index heuristic is robust beyond the baseline model.
Revenue-optimal pricing in common special cases (identical, two-point, MHR, n=2) coincides with uniform-index pricing, giving exact characterizations of the optimum.
Because uniform prices can be log n worse than optimal, the paper identifies index equalization—not price equalization—as the right simplicity benchmark.
The optional-inspection results (n/(n−1) and 2 approximations) indicate that the hard buyer-side problem can still yield simple near-optimal seller mechanisms in structured cases.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The gap between the known lower bound (1.14) and the proven upper bound (4) suggests the true worst-case simplicity gap is likely much smaller than 4; a natural next step is to compute the optimal factor by interpolation between uniform-index level sets.
The support lemma implies the optimal uniform index is structurally tied to the alternatives' supports, which may let practitioners fine-tune prices by rounding to observed support points rather than searching over all reals.
For the general optional-inspection model, the paper leaves the arbitrary-distribution case open; its reveal-mechanism technique might extend to distributions with more than two points if the buyer's value ceiling E[max X_i] is used as an upper bound.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The central result is solid and worth your time. The paper introduces Pandora's Box Pricing—a single seller prices access to box inspections, the buyer follows Weitzman, and the seller maximizes expected revenue. The main theorem, uniform-index prices give a 4-approximation, is clean, and I checked the proof: the splitting at the first k where the inspection probability drops below 1/2 is correct, the tail bound uses the optimal revenue itself, and the head bound goes through with the right monotonicity. Nothing circular or fitted. The support lemma (Lemma 4) and the efficient computation corollary are nice, and the extensions to seller costs and uniform matroids are sensible. The special-case characterizations for i.i.d., two-point, and MHR distributions are also meaningful, and the unbounded gap for uniform prices is a good contrast. Credit where due: the model is new relative to Armstrong and Gamp, and the main proof is genuinely elegant.

The soft spots are real but minor, and the reader's report has them in the right proportion. First, Theorem 1(4) asserts that MHR variables are continuous, which is false—geometric distributions are MHR and discrete. The proof uses E[X|X>0]=E[X], which depends on continuity at zero. That's a genuine gap in a theorem statement, but it doesn't touch Theorem 3, and I suspect it can be patched with an approximation argument or a slightly different proof. Second, Theorem 2's lower bound (uniform-index not optimal, ratio >1.14) rests on a Python grid search with no code or detailed parameter tables. That's frustrating for reproducibility, but it's a peripheral result. The example itself seems well-formed; it just needs verification. The tie-breaking convention is seller-favorable, which is standard and justified by the epsilon-perturbation footnote; I don't see a load-bearing issue there. The optional inspection section is more exploratory, and the factor n/(n-1) for identical alternatives is clever, but the reduction relies on a fixed outside option and the claims about revenue domination are plausible though not fully airtight.

This is a paper for people working in algorithmic mechanism design, data pricing, and sequential search. It deserves a serious referee: the main result is strong, the model is natural, and the presentation is clear. I would engage with it and would accept a review invitation. The fixes are small: correct the MHR statement, release the code or convert the grid search into an analytic bound, and tighten the optional-inspection reduction if possible. After that, it should be publishable in a good theory venue.

Recommendation: send it to peer review, and expect minor revision rather than major surgery.

Referee Report

3 major / 3 minor

Summary. The paper studies revenue maximization by a seller who posts non-adaptive prices for inspecting stochastic alternatives, with a buyer who follows an optimal Pandora's Box search strategy. The central claim is that prices that equalize all Weitzman indices achieve a 4-approximation to the optimal revenue (Theorem 3), and that the best uniform-index price can be computed efficiently for discrete alternatives (Corollary 6). The paper also characterizes exact optimality for several special cases (Theorem 1), shows that uniform prices can be an unbounded factor worse than the optimum (Theorem 8), and develops extensions with seller costs, outside options, matroid constraints, and optional inspection. The main proof of Theorem 3 is clean and internally sound, but I find a significant gap in the proof of Theorem 2, an unsupported continuity assumption in Theorem 1(4), and a missing justification in Claim 18.

Significance. If the results are correct, the paper makes a valuable contribution to algorithmic mechanism design and data pricing: it gives a simple, efficiently computable, parameter-free 4-approximation for revenue in a natural sequential-information-acquisition model, and it provides a sharp separation between index-based and price-based uniform pricing. The proof of Theorem 3 is short and relies only on Weitzman's characterization, which makes the result accessible and likely to be used as a benchmark. The special-case characterizations and the extensions to costs, matroids, and optional inspection broaden the applicability. The main theorem is not affected by the issues I raise, but the paper currently contains an unverifiable numerical proof and an incorrect generality claim in a secondary theorem, so the manuscript needs revision before it meets the standard of a rigorous theory paper.

major comments (3)

[Appendix A.2, Theorem 2] The proof of the constant-factor gap is not a formal proof as written. The text admits that the lower bound expression L(n,z,α,β,γ,δ) cannot be optimized analytically, and instead states that a Python greedy grid search was run and that L>1.14 for the listed floating-point parameters. No code, no exact arithmetic/interval verification, and no reproducible description of the search are provided. A numerical grid search is not a mathematical certificate, and the reader cannot verify the claimed inequality for the stated parameters. This is a load-bearing gap for the theorem that uniform-index prices are not optimal. Please replace this with an analytic construction or provide a machine-checked interval/rational certificate and the corresponding code.
[Section 3, Theorem 1(4) proof] The proof says 'as MHR variables are continuous, E[X|X>0] = E[X]'. This is not a consequence of the MHR definition given in the footnote: discrete distributions (e.g., geometric or two-point distributions) can have monotone hazard rate under the standard discrete definition, and the paper uses discrete distributions elsewhere. The proof also needs the assumption that there is no atom at 0 for E[X|X>0] = E[X]. Please either restrict the MHR case to continuous distributions with support not containing an atom at 0, or adapt the argument to discrete MHR distributions. As written, the statement of Theorem 1(4) is broader than what the proof supports.
[Appendix C.3, Claim 18] The proof asserts that ordering alternatives 2..n by decreasing r_i 'ensures that whenever the buyer selects an alternative j≥2, they will select the one of maximum realization.' This is not true for arbitrary distributions with equal indices: once an observed value exceeds μ1, the buyer halts and may never see a later, larger realization. The claim is salvageable for two-point distributions because μ_j ≤ μ1 implies that any value above μ1 is necessarily the high value r_j, and ordering by decreasing r_j then yields the global maximum. But this argument is absent, and the assertion as stated is false in general. Since Claim 18 is used in the proof of Theorem 16, this needs a proper justification in the two-point setting.

minor comments (3)

[Section 2, Lemma 5] In the exchange argument, if F_i(g) = 1 for some alternative, the conditional expectation E[X_i | X_i > g] is undefined. Such alternatives have price 0 and can be handled separately (they never block later alternatives). Please add a sentence to cover this edge case.
[Section 3, footnote 6] The discussion of existence of a maximizer for unbounded supports is informal ('epsilon-approximate the supremum'). This is acceptable for the main result, but the notation could be tightened by stating all theorems with an explicit ε-optimal price vector for unbounded distributions.
[Appendix C.1, proof of Theorem 15] The proof invokes an extension of Theorem 1 to mandatory inspection with an outside option y ≥ 0, saying 'the details are omitted.' Since Theorem 1 as proven in the main text is for y = 0, please include the (short) argument or a reference, so the special-case optimality claim is self-contained.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central 4-approximation proof is self-contained and does not reduce to fitted inputs or self-citations.

full rationale

The central claim (Theorem 3) is derived directly from Definition 1: the proof fixes an arbitrary optimal index vector g, splits the revenue at the first k with lambda_k >= 1/2, bounds the tail by 1/2 OPT using the feasible price vector that removes alternatives 1..k, and bounds the head by 2 times the uniform-index revenue at g_k. Each step is an algebraic inequality; the uniform-index revenue is an upper bound on the constructed sum, not an assumed target, and no parameter is fitted to data to obtain the factor 4. Lemma 4 and Corollary 6 only restrict optimization to support points using CDF identities, and the special cases of Theorem 1 use either elementary exchange arguments or the external monotone-mean-residual-life theorem of Barlow et al.; no self-citation carries the proof. The numerical grid-search in Theorem 2 is a verification/rigor concern about a lower-bound example rather than a circular derivation, since it exhibits a concrete instance instead of renaming an input as a prediction. Extensions (Theorems 9, 10, 15, 16) reuse the same inequalities or external buyer-characterization theorems (Weitzman, Singla, Doval, Fu et al.) and do not depend on the authors' prior work as load-bearing evidence. Citations to Chawla et al. appear only in related-work context, not as the basis for the main results. Therefore the derivation chain is self-contained and no circular step is identifiable.

Assumptions & free parameters 1 free parameters · 6 assumptions · 0 invented entities

The central 4-approximation rests on Weitzman's index theorem, seller-favorable tie-breaking, and the revenue formula in Definition 1. The MHR optimality theorem additionally assumes continuity of MHR distributions, which is not justified. The Theorem 2 lower bound introduces a numerically fitted instance but no new physical or conceptual entity.

free parameters (1)

lower-bound instance parameters (z, alpha, beta, gamma, delta, n) for Theorem 2 = z=6.82e-7, alpha=7.5e-11, beta=1.2e4, gamma=2.27e-6, delta=7.625e-1, n=10^6
Chosen via greedy grid search to force L>1.14. This is an existence-proof device rather than a model constant, but the claimed lower bound depends on this hand-fitted parameter set.

assumptions (6)

domain assumption The buyer's optimal search is characterized by Weitzman indices and Fact 1: buyer utility equals E[(max_i min{X_i,g_i})_+].
Section 2; the entire revenue formula and all proofs depend on this characterization.
domain assumption The n alternatives are independent, nonnegative, with known priors; the seller posts non-adaptive prices.
Model setup in Section 2; correlated or adaptive settings would invalidate the index ordering and the revenue formula.
domain assumption Ties in the buyer's inspection order are broken in the seller's favor.
Footnote in Section 2; the revenue formula uses max over consistent permutations, and the approximation proof relies on the seller-chosen order.
standard math Mean residual life of an MHR variable is non-increasing in g (Barlow et al., 1963).
Used in the MHR case of Theorem 1 to bound E[X-g | X>g] by E[X | X>0].
ad hoc to paper MHR distributions are continuous, so E[X | X>0] = E[X].
Theorem 1(4) proof states 'as MHR variables are continuous'; this is not implied by the paper's own hazard-rate definition and is false for discrete MHR distributions such as geometric or Bernoulli.
domain assumption Singla's index-based characterization of the optimal buyer policy under matroid constraints.
Section 4.2; the matroid extensions inherit this external characterization.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Pricing Pandora's Boxes: Revenue Maximization in Sequential Information Acquisition." pith.science (2026). https://pith.science/paper/3VXA65WX

@misc{pith2026260723359,
  author       = {Pith},
  title        = {Pith review of: Pricing Pandora's Boxes: Revenue Maximization in Sequential Information Acquisition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3VXA65WX}},
  note         = {Machine review of arXiv:2607.23359}
}

read the original abstract

We study a mechanism design problem in which a seller controls access to information about a set of stochastic alternatives, and a buyer sequentially acquires information in order to choose a single alternative with high value. The value distributions of the alternatives are known to both parties. The seller posts non adaptive prices for revealing each alternative's realized value, and the buyer responds optimally by following a Pandora's Box strategy: deciding which alternatives to inspect and when to stop by accepting the best inspected alternative. The seller's goal is to maximize his expected revenue, i.e. the total payment collected from all inspections. We study the revenue objective through the lens of simplicity versus optimality. Our main result is that a simple and efficiently computable pricing scheme obtains a 4 approximation in the worst case to the optimal revenue. This pricing rule equalizes the Weitzman indices across all alternatives. In contrast, we show that equalizing the prices themselves can be an unbounded factor worse than the optimum. Furthermore, for several natural special cases, including identically distributed alternatives and monotone hazard rate distributions, we fully characterize the optimal pricing. Finally, we also study a variant of our model under optional inspection, where the buyer may select an alternative without observing its realization. In this setting, we obtain an n/(n-1) approximation for the special case of n identically distributed alternatives, as well as a 2 approximation for the special case where each alternative's value distribution has support size two. Overall, our results highlight both the computational challenges and the power of simple pricing schemes in selling information to a sequential searcher.

Figures

Figures reproduced from arXiv: 2607.23359 by the authors.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

23 extracted references · 1 canonical work pages

[8]

Comput.50, 3 (2019), STOC16–160

A duality-based unified approach to bayesian mechanism design.SIAM J. Comput.50, 3 (2019), STOC16–160. Yang Cai and Grigoris Velegkas

2019
[10]

In Proceedings of the 49th Annual ACM SIGACT Symposium on Theory of Computing(Montreal, Canada)(STOC 2017)

Simple mechanisms for subadditive buyers via duality. In Proceedings of the 49th Annual ACM SIGACT Symposium on Theory of Computing(Montreal, Canada)(STOC 2017). Association for Computing Machinery, 170–183. Shuchi Chawla, Dimitris Christou, and Trung Dang. 2024a. Commitment Gap via Correlation Gap. InWeb and Internet Economics: 21st International Confere...

2017
[12]

Shuchi Chawla, David Malec, and Balasubramanian Sivan

Pandora’s Box with Correlations: Learning and Approximation.2020 IEEE 61st Annual Sym- posium on Foundations of Computer Science (FOCS)(2019). Shuchi Chawla, David Malec, and Balasubramanian Sivan

2020
[13]

Shuchi Chawla, Rojin Rezvan, Yifeng Teng, and Christos Tzamos

The power of randomness in bayesian optimal mechanism design.Games and Economic Behavior91 (2015), 297–317. Shuchi Chawla, Rojin Rezvan, Yifeng Teng, and Christos Tzamos

2015
[17]

Guided Search. (2022). Working paper. 35 Evangelia Gergatsouli and Christos Tzamos

2022
[18]

Johannes H¨ orner and Andrzej Skrzypacz

Selling multiple correlated goods: Revenue maximization and menu-size complexity.Journal of Economic Theory183 (2019), 991–1029. Johannes H¨ orner and Andrzej Skrzypacz

2019
[20]

Ziv Scully and Laura Doval

Simple mechanisms for a subadditive buyer and applications to revenue monotonicity.ACM Transactions on Economics and Computation (TEAC)6, 3-4 (2018), 1–25. Ziv Scully and Laura Doval

2018
[23]

A survey on data markets.arXiv preprint arXiv:2411.07267(2024). 36

arXiv 2024

Show all 23 references

[1963]

Dirk Bergemann, Alessandro Bonatti, and Alex Smolin

Properties of probability distri- butions with monotone hazard rate.The Annals of Mathematical Statistics(1963), 375–389. Dirk Bergemann, Alessandro Bonatti, and Alex Smolin

1963
[1979]

Jiayao Zhang, Yuran Bi, Mengye Cheng, Jinfei Liu, Kui Ren, Qiheng Sun, Yihang Wu, Yang Cao, Raul Castro Fernandez, Haifeng Xu, et al

Optimal Search for the Best Alternative.Econometrica47, 3 (1979), 641–654. Jiayao Zhang, Yuran Bi, Mengye Cheng, Jinfei Liu, Kui Ren, Qiheng Sun, Yihang Wu, Yang Cao, Raul Castro Fernandez, Haifeng Xu, et al

1979
[2007]

Hu Fu, Jiawei Li, and Daogao Liu

The Price of Advice.RAND Journal of Economics38, 4 (2007), 863–880. Hu Fu, Jiawei Li, and Daogao Liu

2007
[2009]

Moshe Babaioff, Nicole Immorlica, Brendan Lucier, and S Matthew Weinberg

Prominence and Consumer Search.The RAND Journal of Economics40 (2009), 209–233. Moshe Babaioff, Nicole Immorlica, Brendan Lucier, and S Matthew Weinberg

2009
[2015]

Yang Cai, Nikhil R Devanur, and S Matthew Weinberg

Pricing Lot- teries.Journal of Economic Theory156 (2015), 144–174. Yang Cai, Nikhil R Devanur, and S Matthew Weinberg

2015
[2016]

Yingkai Li

Selling Information.Journal of Political Economy 124, 6 (2016), 1515–1562. Yingkai Li

2016
[2017]

Mark Armstrong, John Vickers, and Jidong Zhou

Ordered Consumer Search.Journal of the European Economic Association 15, 5 (2017), 989–1024. Mark Armstrong, John Vickers, and Jidong Zhou

2017
[2018]

American Economic Review108, 1 (January 2018), 1–48

The Design and Price of Information. American Economic Review108, 1 (January 2018), 1–48. doi:10.1257/aer.20161079 Dirk Bergemann, Yang Cai, Grigoris Velegkas, and Mingfei Zhao

2018 doi
[2019]

InProceedings of the 2019 ACM Conference on Economics and Computation (EC)

Pandora’s Problem with Nonobligatory Inspection. InProceedings of the 2019 ACM Conference on Economics and Computation (EC). 131–132. Shant Boodaghians, Federico Fusco, Philip Lazos, and Stefano Leonardi

2019
[2020]

Robin Bowers, Elias Lindgren, and Bo Waggoner

Pandora’s Box Problem with Order Constraints.Proceedings of the 21st ACM Conference on Economics and Computation(2020). Robin Bowers, Elias Lindgren, and Bo Waggoner

2020
[2021]

185), James R

(Leibniz Inter- national Proceedings in Informatics (LIPIcs), Vol. 185), James R. Lee (Ed.). Schloss Dagstuhl – Leibniz-Zentrum f¨ ur Informatik, Dagstuhl, Germany, 81:1–81:20. doi:10.4230/LIPIcs.ITCS. 2021.81 34 Yang Cai and Mingfei Zhao

2021 doi
[2022]

Behav.134 (2022), 104–116

Buy-many mechanisms are not much better than item pricing.Games Econ. Behav.134 (2022), 104–116. Xi Chen, Ilias Diakonikolas, Dimitris Paparas, Xiaorui Sun, and Mihalis Yannakakis

2022
[2023]

InWeb and Internet Economics - 19th International Conference, WINE 2023, Vol

Buy-Many Mechanisms for Many Unit-Demand Buyers. InWeb and Internet Economics - 19th International Conference, WINE 2023, Vol. 14413. Springer, 21–38. Shuchi Chawla, Yifeng Teng, and Christos Tzamos

2023
[2024]

Sahil Singla

Local hedging approximately solves Pandora’s box problems with nonobligatory inspection.arXiv preprint arXiv:2410.19011(2024). Sahil Singla

2024 arXiv
[2026]

InProceedings of the 2026 Annual ACM-SIAM Symposium on Discrete Algorithms (SODA)

Combinatorial Selection with Costly Information. InProceedings of the 2026 Annual ACM-SIAM Symposium on Discrete Algorithms (SODA). 664–717. Shuchi Chawla, Evangelia Gergatsouli, Jeremy McMahan, and Christos Tzamos

2026

Pith tools

Reviewed July 31, 2026 · model on record in the stance chip above.

[1] [8]

Comput.50, 3 (2019), STOC16–160

A duality-based unified approach to bayesian mechanism design.SIAM J. Comput.50, 3 (2019), STOC16–160. Yang Cai and Grigoris Velegkas

2019

[2] [10]

In Proceedings of the 49th Annual ACM SIGACT Symposium on Theory of Computing(Montreal, Canada)(STOC 2017)

Simple mechanisms for subadditive buyers via duality. In Proceedings of the 49th Annual ACM SIGACT Symposium on Theory of Computing(Montreal, Canada)(STOC 2017). Association for Computing Machinery, 170–183. Shuchi Chawla, Dimitris Christou, and Trung Dang. 2024a. Commitment Gap via Correlation Gap. InWeb and Internet Economics: 21st International Confere...

2017

[3] [12]

Shuchi Chawla, David Malec, and Balasubramanian Sivan

Pandora’s Box with Correlations: Learning and Approximation.2020 IEEE 61st Annual Sym- posium on Foundations of Computer Science (FOCS)(2019). Shuchi Chawla, David Malec, and Balasubramanian Sivan

2020

[4] [13]

Shuchi Chawla, Rojin Rezvan, Yifeng Teng, and Christos Tzamos

The power of randomness in bayesian optimal mechanism design.Games and Economic Behavior91 (2015), 297–317. Shuchi Chawla, Rojin Rezvan, Yifeng Teng, and Christos Tzamos

2015

[5] [17]

Guided Search. (2022). Working paper. 35 Evangelia Gergatsouli and Christos Tzamos

2022

[6] [18]

Johannes H¨ orner and Andrzej Skrzypacz

Selling multiple correlated goods: Revenue maximization and menu-size complexity.Journal of Economic Theory183 (2019), 991–1029. Johannes H¨ orner and Andrzej Skrzypacz

2019

[7] [20]

Ziv Scully and Laura Doval

Simple mechanisms for a subadditive buyer and applications to revenue monotonicity.ACM Transactions on Economics and Computation (TEAC)6, 3-4 (2018), 1–25. Ziv Scully and Laura Doval

2018

[8] [23]

A survey on data markets.arXiv preprint arXiv:2411.07267(2024). 36

arXiv 2024