{"id":"a936fe22-87cf-4235-9c6c-44cf044f6467","arxiv_id":"1908.08815","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Mean-square optimal OSPA and UOSPA estimators couple independent distant targets, while GOSPA with alpha=2 yields per-target decisions and avoids this spooky effect.","lead":"Optimal estimators built on the OSPA metric show a 'spooky effect at a distance': changing one far-away target's existence probability can change the reported estimates of unrelated distant targets. The paper proves this and shows that the newer GOSPA metric, with alpha=2, avoids it by making each target's decision depend only on its own probability.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"GOSPA decoupling proof relies on a false subset restriction when pairwise distances are merely in (c,2c); a midpoint estimate can beat all subset estimates, so Eq. (9) is not established in the stated regime.","rationale":"I read the paper as establishing a formal property of optimal estimators under the stated multi-Bernoulli/Dirac-delta/d>c assumptions. The reader's ACCEPT is reasonable for the conceptual 'distant targets' narrative, but the precise mathematical claim has a gap. The most load-bearing issue is not location uncertainty but the subset restriction in Appendix A. That restriction is essential: Lemma 4 and Eq. (9) are only proven for estimates that are subsets of the known Bernoulli locations. The appendix's argument that any non-subset point increases the error is invalid because a point between two target locations can reduce expected localization error for either target at the price of one false/missed target. The concrete counterexample with r=0.49 and d=1.1c shows a midpoint estimate is strictly better than every subset estimate, so Eq. (9) is false as a global statement in the regime d>c. It also shows GOSPA can exhibit coupling between independent components in that regime. The intended phenomenon of the paper concerns genuinely far-away targets, and for d>2c the subset restriction can be repaired with a valid dominance argument, so the central conceptual claim likely survives. The verdict should therefore be conditional rather than unchanged: the authors should add the missing separation condition or explicitly restrict the estimator search space, and provide a corrected proof of Appendix A.","tokens_in":12248,"tokens_out":18898,"duration_ms":204771,"concrete_test":"Implement a brute-force minimization of the MSGOSPA cost for the two-component example c=1, x1=0, x2=1.1, r1=r2=0.49 over one-point estimates y in [0,1.1] and compare with the subset estimates. If min_y E[GOSPA^2({y},X)] < 0.49 c^2 (it is about 0.463), Appendix A's subset restriction fails. Then repeat with x2=3 (d=3c>2c); if the minimizer is again among subsets, the missing assumption is exactly d>2c. Optionally, verify that with x2=1.1 and r2=0 the optimal one-point estimate is empty, demonstrating the coupling.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central comparison depends on Lemma 4 and Eq. (9), whose proof first restricts the minimizer to subsets of the known Bernoulli locations {x_i}. This restriction (Appendix A) is the load-bearing step, and it is false under the stated assumption d(x_i,x_j)>c. Counterexample: c=1, x1=0, x2=1.1, r1=r2=0.49, candidate estimate y=0.5. Evaluating the GOSPA alpha=2 assignment formula gives expected square error about 0.463 c^2 for {y}. The best subset estimate is empty, with expected error 0.49 c^2; {x1,x2} gives 0.51 c^2. Hence the true GOSPA minimizer is not a subset of the known locations, and the threshold rule r_i>0.5 in Eq. (9) does not describe the global optimum in this regime. The same mechanism couples the two independent components: with r2=0.49 the optimal estimate is {y}, while with r2=0 it becomes empty, so GOSPA does not automatically avoid the spooky effect when distances are only slightly larger than c. The proof can be repaired by requiring d(x_i,x_j)>2c, or by explicitly restricting estimates to the known target locations, and Appendix A needs a replacement dominance argument that compares arbitrary points to their projections.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies minimum mean-square estimation of a multi-target set when the posterior is a multi-Bernoulli density with N independent components whose single-target densities are Dirac deltas at known locations x_i. It claims that, provided the locations are separated by more than c, the optimal estimates under the OSPA and UOSPA metrics exhibit a 'spooky effect at a distance': the optimal report for one independent component can abruptly change when another far-away component changes its existence probability. In particular, for equal existence probabilities the OSPA estimator reports either no targets or all N targets according to (1-r)^N < r (Eq. 11). In contrast, it claims that the optimal mean-square GOSPA estimator (alpha=2) detects each target independently iff r_i > 0.5 (Eq. 9), so GOSPA avoids the spooky effect. The paper provides closed-form MSE formulas (5)-(7) under the assumption that the optimal estimate is a subset of {x_i}, proves the all-or-nothing OSPA result in Appendix C, and illustrates the decision regions in Figures 1-3.","tokens_in":12434,"tokens_out":18564,"duration_ms":171764,"significance":"The paper addresses an important conceptual question in multi-target tracking: whether the choice of metric induces unwanted coupling in optimal estimates. Its analytical setup is clean, and the closed-form expressions (5)-(7) and the all-or-nothing condition (11) are useful and clearly presented. The OSPA/UOSPA spooky-effect phenomenon is demonstrated concretely, and the contrast with GOSPA's decoupling property is a valuable message for the tracking community. However, the central GOSPA claim is established only under a subset restriction that is not valid for all separations d(x_i,x_j)>c; the paper needs a corrected proof or a strengthened assumption. With that repair, the paper would be a solid contribution.","major_comments":[{"comment":"The assertion that the optimal estimate must be a subset of {x_1,...,x_N} is false for GOSPA when the separation is only d(x_i,x_j)>c. For c=1, x_1=0, x_2=1.1, and r_1=r_2=0.49, the singleton estimate y=0.5 has mean square GOSPA error c^2[0.2601*0.5 + 0.2499*0.25 + 0.2499*0.36 + 0.2401*0.75] = 0.4626 c^2, whereas the best subset estimate is empty with error 0.49 c^2 (and {x_1,x_2} has error 0.51 c^2). Thus the true minimizer of (2) is not a subset, contradicting the premise on which Lemma 4 and Eq. (9) rest. The same example shows that GOSPA can exhibit the spooky effect in this regime: with r_2=0 the optimal estimate for the remaining Bernoulli component is empty, while with r_2=0.49 the optimal estimate is no longer empty (the singleton {0.5} beats every subset estimate). The proof in Appendix A only lower-bounds non-subset distances by the cardinality term, which is 0 for equal-cardinality comparisons and does not rule out a non-subset estimate that is much closer to one target than c/sqrt(2). The results can be repaired by strengthening the separation assumption to d(x_i,x_j)>2c (or explicitly restricting estimates to subsets of {x_i}), but as written the theorem overstates the regime.","section":"Appendix A / Section III-A"},{"comment":"The paper presents the GOSPA decoupling result and the comparison with OSPA/UOSPA as valid for all d(x_i,x_j)>c. Since Eq. (9) is derived from Lemma 4, which assumes the invalid subset restriction, the GOSPA panel of Figure 1 and the GOSPA curves in Figure 3 do not describe the unconstrained minimizer of (2) when c<d(x_i,x_j)<2c. The paper should either prove the subset restriction under a clearly stated stronger condition (for example d(x_i,x_j)>2c) or explicitly frame the results as optimal estimates within the class of subsets of known locations. This is a load-bearing issue because the central claim that GOSPA 'avoids the spooky effect' is precisely what the counterexample above calls into question.","section":"Section III-D and Figures 1-3"}],"minor_comments":[{"comment":"The phrase 'sufficiently far' is used informally; since the formal assumption d(x_i,x_j)>c is not sufficient for the proof, the separation condition should be stated explicitly wherever it is used.","section":"Section III-A"},{"comment":"The inequality for non-subset estimates compares only against a cardinality-only lower bound; even after strengthening the separation, the proof would be clearer as a dominance argument, for example by projecting any candidate point onto the nearest known location and comparing the resulting GOSPA distances for every ground-truth set.","section":"Appendix A"},{"comment":"Example 5 does not specify the distances between the two Bernoulli components; since the claimed OSPA behaviour relies on the subset restriction, the example should either state d much larger than c or be updated to the regime in which the formal assumptions hold.","section":"Example 5"}],"recommendation":"major_revision","confidential_remarks":"No editor-only concerns; the reliance on Proposition 3 of [21] is appropriate and properly cited."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, this paper deserves a serious referee. The core observation is real and, as far as I know, new: optimal mean-square OSPA and UOSPA estimates for a multi-Bernoulli posterior couple independent targets across the state space, and one far-away Bernoulli component can flip the optimal report from zero to all targets. The closed-form formulas and the all-or-nothing condition (1-r)^N < r are clean and useful.\n\nThe soft spot is in the GOSPA half. Appendix A claims the optimal estimate must be a subset of the known Bernoulli locations whenever d(x_i,x_j)>c. That claim is false. For c=1, x1=0, x2=1.1, r1=r2=0.49, the midpoint estimate y=0.5 has expected squared GOSPA error about 0.463, while the best subset estimates are empty (0.49) and both targets (0.51). So the true minimizer is not a subset, and Eq. (9)'s threshold r_i>0.5 is not the global optimum in this regime. In fact, with r2=0.49 the optimal estimate is {0.5}, and with r2=0 it becomes empty, so GOSPA also shows the spooky effect when targets are only slightly farther apart than c. The paper says the results hold for d(x_i,x_j)>c, so this is not a minor wording issue.\n\nThe fix is easy: assume pairwise distances greater than 2c, or explicitly restrict candidate estimates to the known locations. Under either of those, the GOSPA decoupling is correct, and the OSPA/UOSPA spooky effect survives in the well-separated regime where the effect is conceptually interesting.\n\nThe self-citation to the GOSPA paper [21] is fine; Proposition 3 is proved there and the derivation here is independent. No code, but the math is explicit.\n\nVerdict: worth peer review. I would ask for a revision that fixes the subset restriction before accepting. The central insight is sound, and people working on multi-target tracking metrics will get real value from this paper.","headline":"The paper's core OSPA/UOSPA coupling observation is real, but the GOSPA avoidance proof is overclaimed: for distances just above c a midpoint estimate beats every subset, so the proof needs a d>2c fix.","tokens_in":13079,"tokens_out":5529,"would_cite":true,"duration_ms":53027,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper shows that the minimum mean-square OSPA estimator couples independent far-away targets, so a change in one existence probability can flip the optimal report for all others, while GOSPA with α=2 decouples the decisions and…","keywords":["spooky effect at a distance","optimal sub-pattern assignment","generalised OSPA","multi-Bernoulli posterior","minimum mean square error estimation","random finite sets","multitarget tracking","cardinality mismatch penalty"],"falsifier":"Take a two-component multi-Bernoulli posterior with known locations separated by more than $c$, $r_1=0.6$, $r_2=0.4$, and enumerate the expected squared GOSPA error for all four subsets $\\emptyset$, $\\{x_1\\}$, $\\{x_2\\}$, $\\{x_1,x_2\\}$; the paper predicts $\\{x_1\\}$ is the unique minimiser. If any other subset wins, the $r>0.5$ rule fails even in the paper's Dirac-delta regime. The same enumeration with $r_1=r_2=r_3=0.4$ under OSPA tests the prediction that the optimal estimate jumps from $\\emptyset$ to all three targets because $(0.6)^3<0.4$.","tokens_in":11958,"feed_emoji":"👻","tokens_out":10037,"duration_ms":89315,"temperature":0.7,"pith_summary":"This paper establishes a property of three multi-target estimation metrics — OSPA, unnormalised OSPA (UOSPA), and generalised OSPA (GOSPA) with $\\alpha=2$ — in a deliberately simple posterior: independent potential targets whose positions are known and pairwise separated by more than the metric cutoff $c$. It proves that the minimum mean-square OSPA and UOSPA estimators couple the decisions for all components, so changing the existence probability of one far-away component can flip the optimal report for every other component; the authors call this the spooky effect at a distance. In the equal-probability case the optimal OSPA estimate jumps from reporting no targets to reporting all $N$ targets exactly when $(1-r)^N < r$, so a single distant, unlikely component can trigger a report of all targets. GOSPA with $\\alpha=2$ avoids the effect because its squared error separates into a sum over components and the optimal estimate reports each target exactly when its existence probability exceeds $0.5$. The paper's point is that the choice of metric changes what a Bayesian tracker reports even when the posterior is identical.","feed_headline":"Optimal OSPA estimation goes all-or-nothing across distant targets","feed_subtitle":"Changing one distant target's existence probability flips all reports; GOSPA decouples them with a 0.5 threshold.","key_machinery":"The load-bearing construction is the multi-Bernoulli posterior of Eqs. (3)–(4): $N$ independent Bernoulli components, where component $i$ has existence probability $r_i$ and a Dirac-delta density at $x_i$, with $d(x_i,x_j)>c$ for $i\\neq j$. Because the positions are exactly known, any candidate estimate that is not a subset of $\\{x_1,\\ldots,x_N\\}$ strictly increases the expected error (Appendix A), so estimates are parameterised by detection flags $\\hat{e}_i$. The metric expressions in Lemma 4 then determine the behaviour: for GOSPA with $\\alpha=2$, the assignment-set representation makes the squared error a per-component sum with threshold $0.5$, whereas UOSPA and OSPA include $\\max(n,\\hat{n})$ cardinality terms that tie every component's contribution to the total number reported. This structural difference — additivity versus cardinality normalisation — is the mechanism that turns a change in one remote $r_i$ into a global flip of the estimate.","core_discovery":"Within a multi-Bernoulli posterior with Dirac-delta single-target densities at known locations $x_i$ and pairwise distances greater than $c$, the paper derives closed-form mean-square errors for OSPA, UOSPA, and GOSPA ($\\alpha=2$) and shows that every optimal estimate is a subset of $\\{x_1,\\ldots,x_N\\}$. For GOSPA the squared error is additive, $\\operatorname{MSGOSPA}=\\frac{c^2}{2}\\sum_{i=1}^N [r_i(1-\\hat{e}_i)+(1-r_i)\\hat{e}_i]$, so the optimal detection flag is $\\hat{e}_i=1$ iff $r_i>0.5$, independent of all other components. For UOSPA and OSPA the errors contain cardinality-normalised terms $\\max(n,\\hat{n})$ that couple components; in the equal-probability case the optimal OSPA estimate is either empty or all $N$ targets, with all $N$ chosen exactly when $(1-r)^N<r$. The authors present this spooky effect at a distance as a reason to prefer GOSPA for conventional tracking, while noting that applications focused on total target count might deliberately want such coupling.","pith_inferences":["The structural cause visible in Lemma 4 — cardinality normalisation couples independent components — implies that any metric whose square error includes a $\\max(n,\\hat{n})$ or cardinality-mismatch term will exhibit spooky behaviour in mean-square estimation, not just OSPA and UOSPA; the paper shows two instances, and the mechanism is general.","A practical consequence the paper leaves implicit is that GOSPA-based Bayesian reporting can be decentralised: each local tracker can threshold its own existence probability at $0.5$ without a global consensus step, whereas OSPA-based reporting requires knowing all components before deciding any single report.","A testable extension, not covered by the proofs, is to replace the Dirac deltas with narrow Gaussians or to allow separations comparable to $c$; whether the $0.5$ threshold and the all-or-nothing jump survive location uncertainty is an open question that simulation could settle.","The equal-probability condition $(1-r)^N<r$ gives a distinctive signature: OSPA-evaluated trackers should produce a bimodal reported-track-count distribution at 0 and $N$ as scene size grows, which could be checked in Monte Carlo runs without rederiving the mathematics."],"forward_implications":["For well-separated independent potential targets, the minimum mean-square GOSPA ($\\alpha=2$) estimate is obtained componentwise: report target $i$ iff $r_i>0.5$, with no global search over subsets.","In the same regime, the minimum mean-square OSPA and UOSPA estimates cannot be obtained componentwise, because the decision for one target depends on the existence probabilities of all other targets even at arbitrarily large separations.","With $N$ equal-probability components, the optimal OSPA estimate is bimodal — either no targets or all $N$ targets — and switches at $(1-r)^N<r$, so adding a distant component can reverse the report for every target.","UOSPA changes the reported number of targets one at a time, but adding or removing a far-away component still adds or removes reports for targets elsewhere, so UOSPA also shows the spooky effect.","The paper concludes that the spooky effect gives a rationale for using GOSPA ($\\alpha=2$) in standard multitarget tracking, since it separates localisation, false-target, and missed-target costs."],"supporting_citations":[{"why":"Defines the OSPA metric that is the paper's main object of study.","marker":"[11]"},{"why":"Supplies the earlier point-process metric that OSPA builds on.","marker":"[12]"},{"why":"Proposes unnormalised OSPA for estimation with unknown target number, the UOSPA object analysed here.","marker":"[19]"},{"why":"Introduces GOSPA and the assignment-set decomposition for alpha=2 that underlies the decoupling result.","marker":"[21]"},{"why":"Documents the original spooky action at a distance in the CPHD filter, which motivated the effect studied here.","marker":"[26]"},{"why":"Provides the multi-Bernoulli cardinality distribution used in the closed-form error expressions of Lemma 4.","marker":"[27]"},{"why":"Defines the Bayesian multi-target estimation and set-integral framework in which the optimal estimators are posed.","marker":"[7]"}],"fun_headline_variants":["GOSPA kills the spooky effect with a 0.5 threshold","OSPA's spooky coupling is gone in GOSPA's additive error","Optimal OSPA flips all targets or none; GOSPA decouples them","Spooky effect vanishes when GOSPA penalizes misses and false alarms","GOSPA's per-target threshold beats OSPA's all-or-nothing jumps"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The key premise is that the posterior is a multi-Bernoulli density with Dirac-delta single-target densities at known locations and pairwise separations larger than the cutoff $c$; the closed-form computations and the $0.5$ threshold are proven only in this idealised, well-separated regime.","fun_headline_variants_meta":{"raw":{"variants":["GOSPA kills the spooky effect with a 0.5 threshold","OSPA's spooky coupling is gone in GOSPA's additive error","Optimal OSPA flips all targets or none; GOSPA decouples them","Spooky effect vanishes when GOSPA penalizes misses and false alarms","GOSPA's per-target threshold beats OSPA's all-or-nothing jumps"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000597,"raw_usage":{"total_tokens":2790,"prompt_tokens":939,"completion_tokens":1851,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":555,"completion_tokens_details":{"reasoning_tokens":1745}},"tokens_in":555,"tokens_out":1851,"duration_ms":13533,"temperature":1.0,"reasoning_tokens":1745,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:29:07.459610+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a two-component multi-Bernoulli posterior with known locations separated by more than $c$, $r_1=0.6$, $r_2=0.4$, and enumerate the expected squared GOSPA error for all four subsets $\\emptyset$, $\\{x_1\\}$, $\\{x_2\\}$, $\\{x_1,x_2\\}$; the paper predicts $\\{x_1\\}$ is the unique minimiser. If any other subset wins, the $r>0.5$ rule fails even in the paper's Dirac-delta regime. The same enumeration with $r_1=r_2=r_3=0.4$ under OSPA tests the prediction that the optimal estimate jumps from $\\emptyset$ to all three targets because $(0.6)^3<0.4$.","supporting_citations":[{"cited_title":"A consistent metric for performance evaluation of multi-object filters,","cited_arxiv_id":null,"evidence_quote":"Defines the OSPA metric that is the paper's main object of study."},{"cited_title":"A new metric between distributions of point processes,","cited_arxiv_id":null,"evidence_quote":"Supplies the earlier point-process metric that OSPA builds on."},{"cited_title":"An efficient, variational approximation of the best fitting multi-Bernoulli filter,","cited_arxiv_id":null,"evidence_quote":"Proposes unnormalised OSPA for estimation with unknown target number, the UOSPA object analysed here."},{"cited_title":"\"Spooky action at a dis- tance","cited_arxiv_id":null,"evidence_quote":"Documents the original spooky action at a distance in the CPHD filter, which motivated the effect studied here."}],"review_version":1}