Anchoring for Truthfulness: The Random-Anchor Volume Mechanism for Multi-Facility Location

Abstract: We study the strategyproof placement of (k) facilities on the real line for (n) agents who privately report their locations, without monetary transfers. For two facilities, the Proportional Mechanism of Lu, Sun, Wang, and Zhu (2010) is strategyproof in expectation and achieves a constant-factor approximation to the optimal social cost. Whether such a guarantee is possible for three facilities in the standard model, where each agent is served by her nearest open facility, has remained open. We resolve this question affirmatively by introducing the \emph{Random-Anchor Volume} mechanism. The mechanism first opens a facility at the report of a uniformly random agent, called the \emph{anchor}, and then jointly selects two additional reports, assigning each pair probability proportional to the product of the two consecutive gaps formed by the pair and the anchor. We prove that the mechanism is strategyproof in expectation and has expected social cost at most (8 OPT_3), where (OPT_k) denotes the minimum social cost achievable using at most (k) facilities. The mechanism naturally extends to every (k\geq 2) by selecting (k-1) additional reports with probability proportional to the product of the consecutive gaps among them and the anchor. Under truthful reporting, this generalization has expected social cost at most (4(k-1)OPT_k). Its incentive guarantee, however, has a sharp boundary: the mechanism is strategyproof in expectation for (k\in{1,2,3}), but is manipulable for every (k\geq 4).

• The paper introduces the Random-Anchor Volume mechanism, which is strategyproof in expectation for three facilities and achieves an 8-approximation to optimal social cost on the real line.
• The mechanism generalizes proportional selection by choosing a random anchor and sampling facility locations according to products of consecutive report gaps, yielding a 4(k−1)-approximation for every k≥2.
• The paper proves the family is manipulable for every k≥4, establishing a sharp boundary for this construction while leaving open whether other truthful constant-approximation mechanisms exist beyond three facilities.

Problem and context

The paper studies strategyproof placement of $k$ facilities on the real line for $n$ agents who privately report their locations, without monetary transfers. An agent's cost is her distance to the nearest open facility, and the objective is a randomized mechanism that is strategyproof in expectation while achieving a small worst-case ratio between expected social cost and $OPT_k$, the optimal $k$-facility social cost. This is the standard setting of approximate mechanism design without money (2608.16550).

The background is well known: for one facility, the median rule is both optimal and group-strategyproof (Black; Moulin's generalized median characterization). For two facilities, every deterministic strategyproof mechanism has approximation ratio $\Omega(n)$, but Lu, Sun, Wang, and Zhu's Proportional Mechanism—choose an anchor uniformly at random, then choose the second facility-hosting agent with probability proportional to her distance from the anchor—is strategyproof in expectation with factor $4$ in every metric space. For three facilities, Lu et al.'s sequential extension is manipulable, and they left open whether any constant-factor truthful mechanism exists for $k \ge 3$; subsequent surveys recorded only population-dependent guarantees.

The mechanism

The Random-Anchor Volume ($\mathrm{RAV}_k$) family generalizes the Proportional Mechanism. The mechanism selects an anchor agent uniformly at random and opens a facility at her report. It then assigns to each set $S$ of $k-1$ non-anchor agents the adjacent-gap volume $n$0: if the anchor location $n$1 and the reports in $n$2, sorted as $n$3, are distinct, this is the product of consecutive gaps $n$4. A set is selected with probability proportional to its volume, and facilities open at the anchor and the selected reports; if fewer than $n$5 distinct reports exist, a fallback opens at every distinct report. The name reflects that the gap product equals the Gram determinant of the segments from the anchor, connecting the rule to fixed-size volume sampling (2608.16550). For $n$6 the rule coincides exactly with the Proportional Mechanism.

Strategyproofness for three facilities

The central positive result is that $n$7 is strategyproof in expectation—the first constant-approximation truthful mechanism for three facilities in the unrestricted model. The proof conditions on each possible anchor identity (valid because the anchor draw is uniform and independent of reports) and reduces to an inequality over the second-stage sampling. The key device is a "cyclic compensation" lemma: expanding the cost difference into triples of non-anchor reports yields symmetric blocks $n$8, each shown nonnegative via three geometric facts about $n$9—a piecewise concave shape, a four-point exchange inequality, and a one-centre product bound, the latter two proved through a normalized separation metric on the punctured line. When the misreport is closer to the deviator than all unchanged reports, the cyclic contribution becomes a piecewise cubic whose derivative is concave on each branch, so no negative interior minimum exists.

This resolves affirmatively the open question posed by Lu et al. and highlighted by Procaccia and Tennenholtz, who observed that the intuition behind the two-facility results "already collapses" at three facilities.

Approximation guarantee

For every $OPT_k$0, $OPT_k$1 achieves expected social cost at most $OPT_k$2; in particular $OPT_k$3 achieves factor 8. The proof conditions on the anchor, groups outcomes by which agent is left out of a selected $OPT_k$4-set, and bounds each group against an optimal clustering using contiguity of clusters on the line: since $OPT_k$5 agents map to at most $OPT_k$6 optimal facilities, some adjacent pair shares a facility, yielding per-group bounds of $OPT_k$7 or $OPT_k$8 depending on cluster structure. Averaging over anchors and applying pairwise triangle-inequality bounds within optimal clusters produces the final factor. Notably, the welfare analysis works directly with gaps rather than linear-algebraic volume-sampling machinery.

Sharp failure beyond three facilities

The incentive guarantee has an exact boundary: $OPT_k$9 is manipulable for every $k$0. The construction fixes an anchor at $k$1 and a deviator at $k$2 supported by 100 other agents at $k$3, one agent at $k$4, and 100 negative-side groups at $k$5 with multiplicities $k$6, so each group contributes equal multiplicity-weighted distance $k$7 from the anchor. Reporting $k$8 creates many new zero-cost positive-volume selections containing both $k$9 and a facility at $\Omega(n)$0, diluting costly outcomes; explicit computation shows the deviation strictly profitable conditional on the anchor. Two lemmas then lift this conditional deviation to the full mechanism: adding many agents at the anchor coordinate amplifies the negative conditional difference while contributions from other anchors stay bounded, and appending distant reports preserves the deviation when extra facilities are added. The paper does not claim a lower bound ruling out some truthful constant-approximation mechanism for $\Omega(n)$1—only that this natural family fails.

Limitations and open questions

Several questions remain open. Whether any strategyproof-in-expectation mechanism achieves a constant approximation for $\Omega(n)$2 is unresolved; the authors suggest searching over generalized weights $\Omega(n)$3 replacing the adjacent-gap product. The paper also introduces a Global $\Omega(n)$4-tuple ($\Omega(n)$5) mechanism, inspired by Ma and Peng's Global Pair mechanism (2608.16550), achieving a $\Omega(n)$6-approximation; whether $\Omega(n)$7 is strategyproof in expectation is left open, as is whether randomizing between $\Omega(n)$8 and $\Omega(n)$9 improves either bound. Finally, the authors disclose that the mechanism was discovered and the results proved by ChatGPT during an extended research interaction, with human verification of all statements—a provenance caveat readers should weigh alongside the mathematical content.

Conclusion

The paper resolves a long-standing open problem by exhibiting a strategyproof-in-expectation, 8-approximate randomized mechanism for three-facility location on the line, extends it to arbitrary $4$0 with a $4$1-approximation, and delineates the exact truthfulness boundary of the family at $4$2. The combination of a clean geometric proof technique and a sharp impossibility boundary makes the Random-Anchor Volume family a useful reference point for further work on multi-facility mechanism design without money.