Imagine you're tuning a guitar, but instead of six separate strings each losing energy to friction at the bridge, you route all the damping through a single felt pad under one string. Counterintuitively, that one lossy string makes the entire instrument maximally sensitive to tiny perturbations in tuning. That's the core mechanism here: concentrating dissipation into a single direction doesn't weaken the system — it sharpens its response. The paper studies non-Hermitian systems at full-order exceptional points (EPs) — states where every eigenmode of the system coalesces into one. These are exotic degeneracies that have attracted enormous interest in photonics and sensing because perturbations near an EP produce spectral splittings that scale as the Nth root of the perturbation strength, dramatically amplifying small signals. The open question: given that passive systems necessarily have loss, what arrangement of that loss maximizes the spectral response strength? Wiersig's committed claim: for any passive system of order N at a full-order EP, the spectral response strength is maximized when and only when the anti-Hermitian (loss) part of the Hamiltonian has rank one — meaning all dissipation flows through a single effective channel. He computes the spectral response strength exactly for the rank-one case and derives a sharp upper bound at fixed EP decay rate. This isn't a numerical observation or a conjecture from simulations; it's a closed-form analytical proof valid for arbitrary system order. The ladder context matters. Previous work, notably Wiersig's own 2022 estimates and results from the photonics community on second-order EPs (N=2), established bounds that happened to be tight at order two but became loose for higher orders. This paper tightens the bound for all N≥2 and proves it is saturated only by rank-one loss. The two third-order trimer examples are instructive: the rank-one system hits the bound exactly, while a rank-two passive realization falls strictly below it. This is the kind of clean separation theorists look for. Architecturally, this is pure spectral perturbation theory for finite-dimensional non-Hermitian matrices — the Jordan normal form world. No numerics, no optimization, no variational machinery. The heavy lifting is linear algebra: bounding the norm of the resolvent and connecting it to the structure of the anti-Hermitian part. The method leans on the fact that rank-one perturbations have maximally constrained geometry, which turns out to be exactly what you want for maximizing the leading-order spectral response coefficient. For integrity, this is a mathematical proof paper, not an experimental one. The validation is internal: theorem, proof, illustrative examples. The two trimer models serve as sanity checks rather than independent validation. No experimental realization is presented, and no code or data is involved. This is appropriate for the claim being made — it's a bound, not an empirical observation — but it means the real test comes when experimentalists build rank-one-loss devices at higher orders and measure whether the predicted response enhancement holds up against noise and fabrication imperfections. The practical upshot is a design principle: if you're engineering a photonic sensor or laser at a high-order exceptional point, don't spread loss across multiple channels. Concentrate it. This converts an abstract spectral theory result into an actionable fabrication guideline, which is why the paper frames dissipation as a 'design resource.' The next milestone is experimental demonstration at order N≥3 in a passive photonic or microwave platform, confirming that the rank-one bound is achievable in practice and that the enhanced sensitivity survives realistic noise.
