Pith. sign in

REVIEW 3 major objections 5 minor 48 references

Value of History in Social Learning: Applications to Markets for History

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The value of history is highest when private signals are all-or-nothing, this paper argues, and a monopolist selling the past prefers noisy signals.

desk verdict The value-of-history concept is nice and the market application is well built, but the proof of Theorem 1 has an admitted coupling gap and Proposition 1 contains a wrong formula—salvageable, but not there yet. read the letter →

arxiv 2507.11029 v1 pith:MSYJJP7V submitted 2025-07-15 econ.TH

Hiroto Sato , Konan Shimizu This is my paper · ORCID
classification econ.TH MSC 91A2691B44
keywords sociallearningvalueofhistoryinformationstructuredesignmarketsfordynamicpricingmean-preservingspreadsequential
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 defines the value of history as the added expected payoff from observing predecessors' actions on top of one's own private signal, in a classic binary-state sequential social-learning model. It tries to establish that over all information structures, this value is maximized by a ternary signal that sometimes reveals the state perfectly and is otherwise uninformative. Because the result holds for every agent and for the discounted social value, it gives an upper bound on how much any history-recording system can be worth. It then applies that bound to a monopolist selling access to history, showing equilibrium prices equal each buyer's value of history, and the seller-optimal signal is noisier than the buyer-optimal or socially optimal one.

What carries the argument

The split information structure: given private posterior beliefs, any belief in (0, 1/2) is replaced by a draw that yields 0 with some probability and 1/2 otherwise, and any belief in (1/2, 1) is replaced by a draw that yields 1/2 or 1. This preserves the first-period expected payoff because the decision cutoff equals the prior 1/2, and a lemma from Sato and Shimizu (2025) shows the i-fold product of split posteriors is a mean-preserving spread of the original product; Jensen's inequality applied to the convex gain function g(x) = max{x - 1/2, 0} then gives the payoff comparison. It carries the entire argument: it is the object whose existence Theorem 1 asserts and whose explicit payoff formula yields the closed-form maximizers in Propositions 1-3.

What would settle it

Compute the value of history for a concrete two-signal structure with inconclusive posteriors, say 0.25 and 0.75, for agent 2, and compare it with the split structure; if V_2(π*) < V_2(π) for any such example, Theorem 1 fails. More directly, exhibit any pair of posterior distributions where the coupling equation in footnote 13 cannot hold and show the Jensen step reverses.

Watch

Extended reading notes

Core claim

The central claim is Theorem 1: for any information structure π and equilibrium σ, there exists a split structure π* whose posterior beliefs are only 0, 1/2, and 1 under which every agent's value of history is weakly larger. The split replaces each inconclusive posterior below 1/2 with a raffle between 0 and 1/2 and each inconclusive posterior above 1/2 with a raffle between 1/2 and 1, preserving the expected payoff from private signals alone while weakly increasing the payoff from also using history. If the original structure has informative but inconclusive signals on both sides of 1/2, later agents are strictly better off; otherwise the structures are equivalent. Consequently, the extreme mixture of full and no information is the design that maximizes the value of history.

Load-bearing premise

The proof assumes that the split signal can be coupled with the original signal so that the conditional expectation of the split posterior given the original posterior equals that original posterior; the paper's footnote 13 admits this is 'not correct' in general.

Editorial extensions

If this is right

  • An upper bound on the value of history: no information structure can make any agent's value of history exceed what the optimal ternary signal delivers, so any history-recording system whose cost exceeds the bound cannot be justified by any information environment.
  • In the market application, a monopolist can charge each buyer exactly the buyer's value of history, so seller surplus equals the social value of history; the optimal design for a patient seller approaches no information at all.
  • Buyers prefer full information, the seller prefers an intermediate mixture, and the social optimum lies between the two; under sticky prices, the same ternary characterization survives.
  • The maximized social value of history tends to 1/4 as discounting vanishes, pinning down the maximum possible return to history infrastructure at one quarter of the per-period payoff scale.
  • When the seller can adjust prices only every t periods, the seller-optimal noise level has a closed form, and full information remains buyer-optimal and socially optimal for sufficiently high buyer weight.

Reading between the lines

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

  • A structural incentive the paper leaves implicit: a monopolist selling historical data wants the information environment to be noisy, because noise raises willingness to pay; this suggests data-quality regulation may be needed, not just price regulation.
  • The split argument relies on the decision cutoff coinciding with the 1/2 prior; extending the same value-of-history maximization to non-binary actions or asymmetric payoffs would require a different convexity argument, likely producing a different optimal structure.
  • A testable experimental implication: in sequential social-learning games, the measured willingness to pay for past actions should be nonmonotone in signal precision, peaking at an intermediate precision that shifts with the discount factor.
  • The paper's upper bound offers a practical auditing rule for data platforms: if the cost of storing and serving history exceeds one quarter of the relevant payoff scale, no information design can make the history system profitable.
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.

Referee Report

3 major / 5 minor

Summary. This paper studies sequential social learning with a binary state, binary actions, and conditionally i.i.d. private signals, and defines the value of history for agent i as the difference between the agent's equilibrium expected payoff when both the private signal and predecessors' actions are observed and the payoff when only the private signal is used. The social value of history is the discounted sum of these values. The main theoretical result (Theorem 1) claims that for any information structure there is an equivalent-or-better structure whose posterior belief support is contained in {0,1/2,1}, i.e., a mixture of full and no information. Proposition 1 computes the optimal probability ε of the uninformative signal maximizing each agent's value and the social value. The rest of the paper applies this characterization to a model in which a monopolist sells access to history: the seller's dynamic prices equal the values of history, so seller surplus equals the social value of history, and the paper characterizes buyer-, seller-, and socially optimal information structures, including a sticky-price extension.

Significance. If Theorem 1 were established, this would be a useful and clean result: it reduces an infinite-dimensional information-design problem in social learning to a one-parameter family and gives explicit upper bounds on the value of history that are robust to the information structure. The market-for-history application is a natural translation of value-of-history into willingness to pay, and the finding that the monopolist optimally chooses a noisier structure than is socially optimal is economically interesting. The paper's strengths are its clear statement of the main dominance question and the explicit formulas for the optimal noise levels. However, the main theorem currently rests on an admitted gap in the mean-preserving-spread step and on black-box lemmas from the authors' companion preprint; Proposition 1 also contains an incorrect maximizer formula. These are load-bearing and need to be fixed before the claims can be accepted. No code or machine-checked proofs are provided.

major comments (3)
  1. [Proof of Theorem 1, footnote 13] In the proof of Theorem 1, the key inequality V_i(π*) ≥ V_i(π) is obtained by applying Jensen's inequality to a display that requires E[μ^{*⊗i}|μ^{⊗i}=x] = x. Footnote 13 concedes that 'Precisely, this is not correct' for the natural independent coupling and then assumes the equation 'for simplicity.' This is not a proof: a mean-preserving spread relationship guarantees the existence of a martingale coupling, but the paper does not exhibit it. Without such a coupling, the displayed application of Jensen is unjustified, and Theorem 1—and therefore Propositions 2 and 3, which invoke it—is unproven as written. The authors should either construct the coupling explicitly or prove the needed version of Lemma 4 of Sato and Shimizu (2025) inside this paper.
  2. [Proposition 1] The displayed maximizer ε*_i = (1/i)^{i-1} in Proposition 1 is algebraically wrong. The first-order condition for (1/4)(ε - ε^i) is 1 - i ε^{i-1} = 0, hence ε*_i = (1/i)^{1/(i-1)}. For i=3, the paper's formula gives 1/9, while the correct value is 1/√3 ≈ 0.577. Since the objective is concave, the corrected value is the unique maximizer. The formula should be corrected; the expression for ε*_S involves a separate calculation and is not affected.
  3. [Theorem 1 and Appendix A] The proof of Theorem 1 uses Lemmas 4, 5, and 6 of Sato and Shimizu (2025) as black boxes, and Lemma 2 in the appendix also invokes Lemma 5. These lemmas do the substantive work: Lemma 4 supplies exactly the mean-preserving-spread property that is in dispute, and Lemmas 5–6 are used to move from equilibrium payoffs to V_i(π). A referee cannot verify the central claim without access to a proof of these results. Please state and prove the needed lemmas in this paper, or provide a detailed appendix reproducing the arguments.
minor comments (5)
  1. [Section 1] There are several typos in the Introduction: 'Understanding the the value' should be 'Understanding the value'; 'depending on the information structure' should be 'depends on'; and 'revelation principal' should be 'revelation principle.'
  2. [Proposition 2] The formula for ε* is typeset ambiguously; write it as (1 − sqrt((1−α)(1−δ)/(1−2α)))/δ, consistent with the proof.
  3. [Proof of Lemma 1] In the proof of Lemma 1, the line π*_2(s*_1|L,s_1)=0 appears to contain a typo; it should presumably refer to the signal s*_2(1) under π*_2 and to conditioning on s_2.
  4. [Proof of Theorem 1] The notation V_i(π*) is used in the display 'V_i(π*,σ*) = V_i(π*)' without definition; please define V_i(π*) as the ex-ante payoff under π* before using it.
  5. [Section 4.1] The sentence 'the seller makes take-it-or-leave-it offers at this level' and the clause 'it does not produce agents do not purchase the history' are garbled and should be rewritten.

Circularity Check

2 steps flagged · score 8.0 of 10

Theorem 1's dominance proof reduces to a self-cited companion lemma plus an assumed coupling; Footnote 13 admits the key equation is 'not correct'.

  1. self citation load bearing [Appendix A, Proof of Theorem 1, and Footnote 13]
    "By Lemma 4 in Sato and Shimizu (2025), μ∗⊗i is a mean preserving spread of μ⊗i. Thus, ∑_{y∈supp(μ∗⊗i)} y·μ∗⊗i(y|μ⊗i =x)=x, for each x. ... 13 Precisely, this is not correct because μ∗⊗i and μ⊗i may be independent. ... Here, for simplicity, we assume that μ⊗i and μ∗⊗i itself satisfy this equation."

    The central inequality V_i(π*) ≥ V_i(π) is obtained by applying Jensen's inequality to the conditional expectation E[μ^{*⊗i} | μ^{⊗i}=x] = x. The paper's own footnote concedes this equation is 'not correct' on the natural independent coupling and assumes it 'for simplicity.' The only support for the mean-preserving-spread property is Lemma 4 of the authors' own companion preprint, which is not proved or verified in this paper. Thus the key dominance step is neither derived internally nor backed by an external, machine-checked or otherwise independent proof; it is imported from a self-citation plus an ad hoc assumption. Since Theorem 1 is the basis for Proposition 2 and Proposition 3, the market applications inherit this load-bearing dependence.

  2. self citation load bearing [Introduction, page 2]
    "Technically, our result relies on the analysis of Sato and Shimizu (2025)."

    The paper states at the outset that its main technical result depends on the authors' own companion paper. That companion paper is listed only as an arXiv preprint by the same authors and is not independently verified here. The dependence is therefore load-bearing self-citation: the paper's central characterization of value-maximizing information structures is not self-contained, and the current paper supplies no independent derivation of the cited Lemma 4, Lemma 5, or Lemma 6.

full rationale

The paper contains original modeling of markets for history and closed-form surplus calculations in Propositions 1-3, and the authors are transparent about their reliance on the companion paper. However, the derivation chain for the main theorem is not self-contained: Theorem 1's proof uses Lemma 4 of Sato and Shimizu (2025) to assert a mean-preserving spread, then requires a coupling identity for the Jensen step, which Footnote 13 explicitly says is 'not correct' in general and assumes 'for simplicity.' This means the decisive inequality V_i(π*) ≥ V_i(π) is assumed rather than proved within the paper, and the same is true for the equilibrium-payoff comparisons via Lemmas 5 and 6 of the companion paper. Since Theorem 1 underpins Proposition 2 and Proposition 3, the central results of the market application inherit this dependence. This is not a case of a fitted parameter renamed as a prediction or a definitional equivalence, but it is a clear instance of a load-bearing self-citation chain: the main theorem's force comes from the authors' own unverified prior result plus an admitted simplifying assumption. The score reflects that the central derivation reduces to that self-citation chain, while the market model and algebra are independent contributions.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The model is built on standard social learning assumptions (binary state, binary actions, uniform prior, myopic agents) and on three lemmas imported from the authors' companion paper (Sato and Shimizu 2025). The imported lemmas are load-bearing: Theorem 1's proof relies on them, and the paper explicitly patches one application with an extra assumption. Exogenous parameters δ and α are not fitted to data.

free parameters (2)
  • discount factor δ
    Exogenous discount factor in (0,1); not fitted to data. The optimal information design and social value depend on it.
  • welfare weight α
    Exogenous weight on buyer surplus in the social welfare function, α∈(0,1); not fitted to data.
assumptions (5)
  • domain assumption Sequential social learning model: binary state with uniform prior, binary actions, payoff u(a,H)=1/2 if a=1, -1/2 if a=0, etc. (Section 2).
    Standard framework from Banerjee (1992), Bikhchandani et al. (1992), Smith and Sørensen (2000).
  • domain assumption Lemma 4 of Sato and Shimizu (2025): the split information structure makes μ^{*⊗i} a mean-preserving spread of μ^{⊗i}.
    Used in the proof of Theorem 1; the present paper's footnote 13 notes the precise formulation is not correct as stated and assumes the martingale property.
  • domain assumption Lemmas 5 and 6 of Sato and Shimizu (2025) on upper bounds for agent payoffs under BNE.
    Used in Lemma 2 and the proof of Theorem 1.
  • domain assumption In the market application, the monopolist can commit to a dynamic price path, sells access to history, and non-purchasers' actions are not added to the history (Section 4.1).
    Modeling choice for the market for history.
  • ad hoc to paper Equilibrium selection: when multiple equilibria exist, the one maximizing the social value of history is selected (Definition 1).
    The authors state 'The equilibrium selection rules do not matter for our main results' (footnote after Definition 2), so this is benign but non-standard.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Value of History in Social Learning: Applications to Markets for History." pith.science (2026). https://pith.science/paper/MSYJJP7V

@misc{pith2026250711029,
  author       = {Pith},
  title        = {Pith review of: Value of History in Social Learning: Applications to Markets for History},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MSYJJP7V}},
  note         = {Machine review of arXiv:2507.11029}
}
read the original abstract

In social learning environments, agents acquire information from both private signals and the observed actions of predecessors, referred to as history. We define the value of history as the gain in expected payoff from accessing both the private signal and history, compared to relying on the signal alone. We first characterize the information structures that maximize this value, showing that it is highest under a mixture of full information and no information. We then apply these insights to a model of markets for history, where a monopolistic data seller collects and sells access to history. In equilibrium, the seller's dynamic pricing becomes the value of history for each agent. This gives the seller incentives to increase the value of history by designing the information structure. The seller optimal information discloses less information than the socially optimal level.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 30 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  2. [2]

    The Review of Economic Studies, 78, 1201--1236

    Acemoglu, Daron, Munther A Dahleh, Ilan Lobel, and Asuman Ozdaglar (2011), Bayesian learning in social networks. The Review of Economic Studies, 78, 1201--1236. acemoglu2011bayesian

    2011

  3. [3]

    Journal of Economic Theory, 39, 400--438

    Admati, Anat R and Paul Pfleiderer (1986), A monopolistic market for information. Journal of Economic Theory, 39, 400--438. admati1986monopolistic

    work page 1986

  4. [4]

    The American Economic Review, 78, 96--103

    Admati, Anat R and Paul Pfleiderer (1988), Selling and trading on information in financial markets. The American Economic Review, 78, 96--103. admati1988selling

    work page 1988

  5. [5]

    Econometrica: Journal of the Econometric Society, 901--928

    Admati, Anat R and Paul Pfleiderer (1990), Direct and indirect sale of information. Econometrica: Journal of the Econometric Society, 901--928. admati1990direct

    work page 1990

  6. [6]

    Journal of Economic Theory, 175, 713--729

    Ali, S Nageeb (2018), Herding with costly information. Journal of Economic Theory, 175, 713--729. ali2018herding

    work page 2018

  7. [7]

    arielipositive

    Arieli, Itai, Yakov Babichenko, and Wade Hann-Caruthers (2024), The pisitive effect of garbling on social learning. arielipositive

    work page 2024

  8. [8]

    American Economic Review: Insights, 5, 460--476

    Arieli, Itai, Ronen Gradwohl, and Rann Smorodinsky (2023), Herd design. American Economic Review: Insights, 5, 460--476. arieli2023herd

    2023

Show all 48 references
  1. [9]

    The Review of Economic Studies, 86, 913--940

    Arieli, Itai and Manuel Mueller-Frank (2019), Multidimensional social learning. The Review of Economic Studies, 86, 913--940. arieli2019multidimensional

    2019

  2. [10]

    Mathematics of Operations Research, 46, 1235--1249

    Arieli, Itai and Manuel Mueller-Frank (2021), A general analysis of sequential social learning. Mathematics of Operations Research, 46, 1235--1249. arieli2021general

    2021

  3. [11]

    Research in Economics, 72, 101--116

    Athey, Susan and Jonathan Levin (2018), The value of information in monotone decision problems. Research in Economics, 72, 101--116. athey2018value

    2018

  4. [12]

    Games and Economic Behavior, 46, 1--22

    Banerjee, Abhijit and Drew Fudenberg (2004), Word-of-mouth learning. Games and Economic Behavior, 46, 1--22. banerjee2004word

    2004

  5. [13]

    The Quarterly Journal of Economics, 107, 797--817

    Banerjee, Abhijit V (1992), A simple model of herd behavior. The Quarterly Journal of Economics, 107, 797--817. banerjee1992simple

    1992

  6. [14]

    Economics Letters, 235, 111515

    Ben-Shahar, Danny and Eyal Sulganik (2024), New characterization of Blackwell 's order for subsets of information structures. Economics Letters, 235, 111515. ben2024new

    2024

  7. [15]

    American Economic Journal: Microeconomics, 7, 259--294

    Bergemann, Dirk and Alessandro Bonatti (2015), Selling cookies. American Economic Journal: Microeconomics, 7, 259--294. bergemann2015selling

    2015

  8. [16]

    Annual Review of Economics, 11, 85--107

    Bergemann, Dirk and Alessandro Bonatti (2019), Markets for information: An introduction. Annual Review of Economics, 11, 85--107. bergemann2019markets

    2019

  9. [17]

    American economic review, 108, 1--48

    Bergemann, Dirk, Alessandro Bonatti, and Alex Smolin (2018), The design and price of information. American economic review, 108, 1--48. bergemann2018design

    2018

  10. [18]

    Journal of Political Economy, 100, 992--1026

    Bikhchandani, Sushil, David Hirshleifer, and Ivo Welch (1992), A theory of fads, fashion, custom, and cultural change as informational cascades. Journal of Political Economy, 100, 992--1026. bikhchandani1992theory

    1992

  11. [19]

    In Proceedings of the second Berkeley symposium on mathematical statistics and probability, volume 2, 93--103, University of California Press

    Blackwell, David (1951), Comparison of experiments. In Proceedings of the second Berkeley symposium on mathematical statistics and probability, volume 2, 93--103, University of California Press. blackwell1951comparison

    1951

  12. [20]

    The Annals of Mathematical Statistics, 265--272

    Blackwell, David (1953), Equivalent comparisons of experiments. The Annals of Mathematical Statistics, 265--272. blackwell1953equivalent

    1953

  13. [21]

    Journal of Economic theory, 144, 1421--1439

    Callander, Steven and Johannes H \"o rner (2009), The wisdom of the minority. Journal of Economic theory, 144, 1421--1439. callander2009wisdom

    2009

  14. [22]

    Games and Economic Behavior, 47, 72--86

    C elen, Bo g a c han and Shachar Kariv (2004), Observational learning under imperfect information. Games and Economic Behavior, 47, 72--86. ccelen2004observational

    2004

  15. [23]

    Technical report

    De Lara, Michel and Olivier Gossner (2017), An instrumental approach to the value of information. Technical report. de2017instrumental

    2017

  16. [24]

    The Rand Journal of Economics, 38, 863--880

    Es o , P \'e ter and Bal \'a zs Szentes (2007), The price of advice. The Rand Journal of Economics, 38, 863--880. esHo2007price

    2007

  17. [25]

    American Economic Review, 109, 3650--3680

    Frankel, Alexander and Emir Kamenica (2019), Quantifying information and uncertainty. American Economic Review, 109, 3650--3680. frankel2019quantifying

    2019

  18. [26]

    Games and Economic Behavior, 45, 329--346

    Gale, Douglas and Shachar Kariv (2003), Bayesian learning in social networks. Games and Economic Behavior, 45, 329--346. gale2003bayesian

    2003

  19. [27]

    Journal of Economic Theory, 173, 383--409

    Hann-Caruthers, Wade, Vadim V Martynov, and Omer Tamuz (2018), The speed of sequential asymptotic learning. Journal of Economic Theory, 173, 383--409. hann2018speed

    2018

  20. [28]

    American Economic Journal: Microeconomics, 4, 1--31

    Hendricks, Kenneth, Alan Sorensen, and Thomas Wiseman (2012), Observational learning and demand for search goods. American Economic Journal: Microeconomics, 4, 1--31. hendricks2012observational

    2012

  21. [29]

    Econometrica, 92, 1033--1062

    Kartik, Navin, SangMok Lee, Tianhao Liu, and Daniel Rappoport (2024), Beyond unbounded beliefs: How preferences and information interplay in social learning. Econometrica, 92, 1033--1062. kartik2024beyond

    2024

  22. [30]

    International Game Theory Review, 8, 21--31

    Kultti, Klaus and Paavo Miettinen (2006), Herding with costly information. International Game Theory Review, 8, 21--31. kultti2006herding

    2006

  23. [31]

    The BE Journal of Theoretical Economics, 7, 0000102202193517041320

    Kultti, Klaus K and Paavo A Miettinen (2007), Herding with costly observation. The BE Journal of Theoretical Economics, 7, 0000102202193517041320. kultti2007herding

    2007

  24. [32]

    IEEE Transactions on Signal and Information Processing over Networks, 3, 239--251

    Le, Tho Ngoc, Vijay G Subramanian, and Randall A Berry (2017), Information cascades with noise. IEEE Transactions on Signal and Information Processing over Networks, 3, 239--251. le2017information

    2017

  25. [33]

    Annals of Statistics, 16, 521--533

    Lehmann, EL (1988), Comparing location experiments. Annals of Statistics, 16, 521--533. lehmann1988comparing

    1988

  26. [34]

    The Quarterly Journal of Economics, 135, 389--448

    Liang, Annie and Xiaosheng Mu (2020), Complementary information and learning traps. The Quarterly Journal of Economics, 135, 389--448. liang2020complementary

    2020

  27. [35]

    Theoretical Economics, 10, 807--851

    Lobel, Ilan and Evan Sadler (2015), Information diffusion in networks through social learning. Theoretical Economics, 10, 807--851. lobel2015information

    2015

  28. [36]

    UB Economics--Working Papers, 2022, E22/434

    Lorecchio, Caio Paes Leme (2022), Persuading crowds. UB Economics--Working Papers, 2022, E22/434. lorecchio2022persuading

    2022

  29. [37]

    arXiv preprint arXiv:2303.13409

    Mekonnen, Teddy, Zeky Murra-Anton, and Bobak Pakzad-Hurson (2023), Persuaded search. arXiv preprint arXiv:2303.13409. mekonnen2023persuaded

    2023 arXiv

  30. [38]

    American Economic Journal: Microeconomics, 8, 83--109

    Mueller-Frank, Manuel and Mallesh M Pai (2016), Social learning with costly search. American Economic Journal: Microeconomics, 8, 83--109. mueller2016social

    2016

  31. [39]

    Journal of Economic Theory, 207, 105586

    Parakhonyak, Alexei and Nick Vikander (2023), Information design through scarcity and social learning. Journal of Economic Theory, 207, 105586. parakhonyak2023information

    2023

  32. [40]

    Econometrica, 68, 135--148

    Persico, Nicola (2000), Information acquisition in auctions. Econometrica, 68, 135--148. persico2000information

    2000

  33. [41]

    Econometrica, 87, 2141--2168

    Rosenberg, Dinah and Nicolas Vieille (2019), On the efficiency of social learning. Econometrica, 87, 2141--2168. rosenberg2019efficiency

    2019

  34. [42]

    arXiv preprint arXiv:2503.05015

    Sato, Hiroto and Konan Shimizu (2025), Value of information in social learning. arXiv preprint arXiv:2503.05015. sato2025value

    2025 arXiv

  35. [43]

    Games and Economic Behavior, 39, 137--166

    Sgroi, Daniel (2002), Optimizing information in the herd: Guinea pigs, profits, and welfare. Games and Economic Behavior, 39, 137--166. sgroi2002optimizing

    2002

  36. [44]

    Econometrica, 68, 371--398

    Smith, Lones and Peter S rensen (2000), Pathological outcomes of observational learning. Econometrica, 68, 371--398. smith2000pathological

    2000

  37. [45]

    Available at SSRN 1138095

    Smith, Lones and Peter Norman Sorensen (2013), Rational social learning by random sampling. Available at SSRN 1138095. smith2013rational

    2013

  38. [46]

    The Review of Economic Studies, 88, 2527--2554

    Smith, Lones, Peter Norman S rensen, and Jianrong Tian (2021), Informational herding, optimal experimentation, and contrarianism. The Review of Economic Studies, 88, 2527--2554. smith2021informational

    2021

  39. [47]

    Journal of Economic Theory, 166, 324--333

    Song, Yangbo (2016), Social learning with endogenous observation. Journal of Economic Theory, 166, 324--333. song2016social

    2016

  40. [48]

    Available at SSRN 4498779

    Xu, Wenji (2023), Social learning through action-signals. Available at SSRN 4498779. xu2023social

    2023

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.