REVIEW 5 major objections 6 minor 4 references
Ordinal Tax To Sustain a Digital Economy
T0 review · 5 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read With zero transportation costs, a 1% cost advantage becomes a 33% profit gap, and an ordinal distance tax can prevent the resulting monopoly.
desk verdict New ordinal tax idea worth thinking about, but the numerical evidence is not reproducible because the model leaves the rationing rule unspecified; treat the percentages as illustrative, not exact. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The key machinery is a two-stage quantity-then-price game between two firms located with buyers on a one-dimensional ring, where each buyer faces a transaction cost that depends on distance to the firm. In the cardinal-tax version the cost is $\lambda \cdot d \cdot p_i$, with $d$ the physical distance; in the ordinal-tax version each buyer ranks firms by physical distance and pays a tax based on rank $D$, so $D=0$ for the nearest firm, $D=1$ for the second nearest, and so on. Firms choose production quantities first and prices second, and the reported prices, quantities, and profits are Nash equilibria of that game, found numerically because location-dependent taxes prevent a closed-form solution. The mechanism works by making the ordinal tax a centralisation threshold: marginal cost advantages are absorbed by the tax, while advantages large enough to outweigh the tax still flow to the more efficient firm.
What would settle it
Recompute the two-firm equilibria on the same ring with an explicit rationing rule, such as proportional allocation when a firm's capacity is insufficient, and compare the 5%, 7%, and 33% profit differences and the ordinal-tax revenue ratio; if these figures shift substantially, the reported magnitudes are not determined by the model as stated.
Extended reading notes
Core claim
The central claim is that zero-cost transportation is not merely an efficiency gain; it is a structural force that amplifies small, arbitrary differences between competitors. In the model, two identical firms facing a small cardinal distance tax have nearly equal profits, with a 5% profit gap, and a 1% cost difference widens that gap only to 7%. Removing the tax entirely turns the same 1% difference into a 33% profit gap. The paper argues that excess profit can be reinvested to widen the cost gap, so the market tips toward monopoly. The proposed ordinal tax charges each buyer a tax based on the rank of distance to each firm, with the closest firm at rank zero, the second closest at rank one, and so on. In the numerical example, this keeps total revenue close to the untaxed level, leaves small quality differences unable to overcome the tax, and still allows a large 20% cost advantage to produce substantially higher revenue for the more efficient firm.
Load-bearing premise
The numerical equilibrium results depend on how a sold-out firm's product is rationed among remaining buyers, but the paper does not state that rationing rule, and in quantity-price games the rationing rule can materially change equilibrium prices and profits.
Editorial extensions
If this is right
- If transport costs keep falling, untaxed digital markets should be expected to tip toward a single dominant firm even from random 1% cost differences, because the resulting 33% profit gap can be reinvested to widen the advantage.
- An ordinal distance tax removes the physical-clustering incentive created by cardinal transport costs, since only the rank-order of distance matters, not the actual distance.
- Total market revenue under the ordinal tax is predicted to stay near the no-tax level, unlike the cardinal tax, which reduces revenue, making the ordinal tax a less costly intervention.
- The ordinal tax acts as a tunable threshold: setting the scaling factor $\lambda$ to zero gives free digital trade, while very large $\lambda$ forces local monopolies, and intermediate values can target a desired level of resistance to centralisation.
- Large genuine competitive advantages still earn higher revenue under the ordinal tax, so the policy blocks random or minor advantages rather than all rewards for superiority.
Reading between the lines
- The same rank-based principle could extend beyond physical distance: any platform that ranks alternatives by closeness, such as language, community, or recommendation similarity, could tax by rank and create a similar stabilising threshold.
- The dynamic tipping story is only sketched; a formal model with reinvestment of excess profits would likely show the 33% gap compounding, but the paper does not prove that explicitly.
- The reported percentages come from a fixed arrangement of buyers and firms; averaging equilibria over many random ring configurations would reveal whether 5%, 7%, and 33% are typical or specific to the pictured geography.
- The welfare effects of redistributing the ordinal tax are left unexplored, so the policy argument would require a distributional analysis to be complete.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper argues that as transportation costs vanish, digital markets become prone to monopolization: in a two-firm, two-stage (capacity-then-price) game on a ring, a 1% cost difference yields a 33% profit difference if no distance tax is present, versus 7% with a cardinal distance tax. The authors propose an 'ordinal' tax that charges a buyer's second-closest firm (rank D=1) rather than taxing by physical distance, and report numerically that this tax equalizes profits under equal costs, keeps total revenue close to the no-tax level, and still rewards major cost advantages. The analysis is entirely numerical (using the Gambit library), and the model is only sketched rather than fully specified.
Significance. The policy idea—taxing ordinal distance rather than cardinal distance—is original and, if rigorously established, would be relevant to the taxation of digital services. The paper also offers a falsifiable comparative-statics claim: without transportation costs, small quality differences translate into large profit imbalances in a static duopoly. However, the manuscript does not currently contain the formal development needed to support these claims: the two-stage game is under-specified, the numerical results are not reproducible, and the step from a one-shot profit imbalance to monopolization is not modeled. These weaknesses are central rather than cosmetic, so the contribution is best viewed as a promising research proposal in need of substantial completion.
major comments (5)
- [Section 'The two firms choose capacity and price...' (paragraph 2) and Figure 2] The two-stage game is not fully specified because the rationing rule is never stated. When a firm's capacity q_i is less than the number of buyers who, at the posted price (including the per-buyer tax), prefer that firm, the paper does not say how the scarce units are allocated among those buyers. In Kreps-Scheinkman games the equilibrium is defined only after choosing a rationing rule (e.g., efficient/parallel, proportional, or random), and the residual demand functions—and hence equilibrium prices and profits—differ across rules. Because the tax is buyer-specific, it changes which buyers prefer each firm at a given price configuration, so the missing rule is not an innocuous detail. Consequently, the profit differences (5%, 7%, 33%) and the revenue comparisons in Figure 2 are not uniquely determined by the model as written.
- [Section 'Because the tax depends... Gambit' and Figure 2] The numerical results are not reproducible. The paper does not give the extensive-form representation used in Gambit, the payoff functions for every terminal history, the algorithm for finding Nash equilibria, or the code and parameter files. The reported numbers on the left and right axes of Figure 2 therefore cannot be checked by an independent reader. The footnote that 'changing these numbers does not alter the qualitative statements' is not a substitute for providing the exact computations, especially because the central qualitative claim about being 'prone to monopolization' is quantitatively anchored in the 33% figure.
- [Section 'If there is no cost of transportation...' and footnote 2] The paper's central claim that a zero-transportation-cost market is 'prone to monopolization' is supported only by a static one-shot profit imbalance. The text asserts that the 33% excess profit can be used to further decrease production costs and eventually force the rival out of business, but no dynamic model is presented; footnote 2 merely says the claim 'can be formalised by considering multi-stage games.' Without an explicit dynamic process (e.g., repeated investment, entry/exit, or cost-reducing R&D), the static profit difference is an observation about a single equilibrium, not a proof of a monopoly tendency. This is a load-bearing gap in the paper's main argument.
- [Section 'Therefore, we propose to introduce an ordinal tax instead'] The claim that the ordinal tax 'is not susceptible to (small) changes in buyer or firm locations, and hence does not lead to physical centralisation (profits are exactly equal if c1=c2)' is stated without qualification or proof. In the depicted configuration, each firm is closest to exactly 6 buyers, so symmetry alone could produce equal profits. For a generic arrangement in which one firm is closest to more buyers than the other, the ordinal tax gives the firm with more D=0 buyers a structural demand advantage, and equal profits are not obvious; the paper itself concedes that 'if the distribution of buyers were more irregular, so would be the profits.' The assertion needs a precise statement of the conditions under which it holds (e.g., balanced configurations) and a derivation or at least a numerical exploration outside the symmetric case.
- [Section 'The buyers cannot buy at price p_i...'] The demand side is under-specified. The paper says there are n potential buyers, each with demand curve p = u - q, but it does not explain how individual demands aggregate into a market demand or how a firm's quantity q_i is allocated across the n heterogeneous buyers. This ambiguity interacts with the missing rationing rule and makes the payoff functions used in the numerical computation undefined. The authors should specify the buyer-level demand and the allocation mechanism clearly at the beginning of the model.
minor comments (6)
- [Section 'The buyers cannot buy at price p_i...'] The formula for the ordinal tax is never written down; after defining D, the paper should give the tax explicitly, for example as λ·D·p_i, or state whether it is a per-unit tax.
- [Figure 2] The figure's x-axis labels are difficult to parse (e.g., '0% cost differencecardinal tax' is a run-on); please separate the conditions and add a legend that distinguishes left-axis bars (profit difference) from right-axis bars (revenue difference).
- [Section 'The two firms choose capacity and price...'] The phrase 'similar to Kreps and Scheinkman (1983)' is not sufficient; the paper should state how the two-stage game differs from KS, including whether the tax is levied on buyers (a freight cost) or on firms, and who receives the tax revenue.
- [Section 'The two firms choose capacity and price...'] The values λ=0.1, u=120, and γ=1 are taken without sensitivity analysis; given that footnote 1 says some quoted numbers may change, it would be informative to see at least a robustness table over λ and over the buyer configuration.
- [Throughout] The terms 'revenue' and 'profit' are used inconsistently; the left axis of Figure 2 reports relative difference in profit, while the right axis reports relative difference in total revenue. Please standardize the terminology.
- [Section 'An immediate side-effect...'] The statement 'the overall level of revenue is elevated compared to a cardinal tax, as now only remote purchases (D > 0) are taxed' conflates consumer price and producer revenue; a tax on remote purchases reduces quantity purchased and can lower producer revenue depending on the elasticity. The revenue comparison should be stated as a model result, not as a direct implication of the tax's incidence.
Circularity Check
No significant circularity; the paper's numerical equilibrium results are model outputs with no fitted inputs or load-bearing self-citations.
full rationale
The paper defines a two-stage capacity-then-price game and computes Nash equilibria numerically with Gambit. The monopoly-prone result (33% profit difference with a 1% cost difference and no tax) and the revenue comparisons are outputs of the equilibrium computation, not restatements of the model definition. The ordinal tax is defined by distance ranks (D=0 for the closest firm, D=1 for the next, etc.), not by the outcome it is claimed to produce; its stabilizing effect is demonstrated through the same numerical equilibrium procedure. No parameters are fitted to data. No load-bearing self-citations appear: Kreps-Scheinkman, Hotelling, and Isard are external standard references, and the appeal to Kreps-Scheinkman is not used as an unproved uniqueness theorem. The unspecified rationing rule when capacity is insufficient is a genuine modeling ambiguity that affects numerical values, but it is an under-specification, not a circular reduction. Hence no step reduces by construction to its inputs.
Assumptions & free parameters
free parameters (3)
- lambda (tax scaling) =
0.1
- u (demand intercept) =
120
- gamma (distance exponent in cardinal tax) =
1
assumptions (4)
- standard math The two-stage quantity-then-price competition of Kreps-Scheinkman applies, with buyers choosing the firm offering the lower all-in price (price plus tax).
- domain assumption Demand is linear and identical across buyers: p = u - q, with total quantity allocated to the chosen firm.
- ad hoc to paper The ordinal tax is a multiplicative surcharge on the price of the second-closest firm (D=1), independent of physical distance.
- ad hoc to paper The market has exactly two firms and a fixed, arbitrary buyer-firm configuration; results are claimed to be generic.
Cite this review
Pith. "Pith review of Ordinal Tax To Sustain a Digital Economy." pith.science (2026). https://pith.science/paper/X6BIHYMP
@misc{pith2026190803287,
author = {Pith},
title = {Pith review of: Ordinal Tax To Sustain a Digital Economy},
year = {2026},
howpublished = {\url{https://pith.science/paper/X6BIHYMP}},
note = {Machine review of arXiv:1908.03287}
}
read the original abstract
Recently, the French Senate approved a law that imposes a 3% tax on revenue generated from digital services by companies above a certain size. While there is a lot of political debate about economic consequences of this action, it is actually interesting to reverse the question: We consider the long-term implications of an economy with no such digital tax. More generally, we can think of digital services as a special case of products with low or zero cost of transportation. With basic economic models we show that a market with no transportation costs is prone to monopolization as minuscule, random differences in quality are rewarded disproportionally. We then propose a distance-based tax to counter-balance the tendencies of random centralisation. Unlike a tax that scales with physical (cardinal) distance, a ranked (ordinal) distance tax leverages the benefits of digitalization while maintaining a stable economy.
Figures
Reference graph
Works this paper leans on
-
[1]
Hotelling, H. (1929). Stability in Competition. The Economic Journal, 39(153):41–57
work page 1929
-
[2]
Isard, W. (1954). Location theory and trade theory: short-run analysis. The Quarterly Journal of Economics , pages 305–320
work page 1954
-
[3]
Kreps, D. M. and Scheinkman, J. A. (1983). Quantity precommitment and Bertrand competition yield Cournot outcomes. The Bell Jour- nal of Economics, pages 326–337
work page 1983
-
[4]
McKelvey, R. D., M. A. M. and Turocy, T. L. (2016). Gambit: Soft- ware Tools for Game Theory, Version 16.0.1. 3
work page 2016
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.