Limits
What the score cannot tell you today, stated in the same place as what it can.
The published document is a pattern, not today
The score document bundled in the public repository is flagged as not live and describes itself as a representative pattern rather than current conditions. It is there so the viewer has something correct in shape to render when the live object cannot be read. Its figures are not published on this site, because an offline sample presented as a reading is exactly the failure this site is built to avoid.
Two domains have no assigned feed
Ocean acidification is scored as a boundary domain and is the boundary most recently reported as transgressed, which makes it one of the more consequential rows in the table. It has no provider assigned. The domain is open for proposals.
Cryosphere carries a citation for its glacier dataset but no feed wired to the domain, and the planetary trend arrow is derived from cryosphere alone. So the direction the score reports rests on the thinner of its two thinnest domains. Together the two are 0.10 and 0.10 of a 1.18 total.
The method is not ratified, and this site will not say it is
The score document's own disclosure states that the boundary thresholds, the split of the former single ocean domain into warming and acidification at 0.10 each, and the newer 0.10 human weight are hand-set proposals awaiting owner ratification. Other material describes the same version as frozen and ratified. Both cannot be true.
Until that is resolved, no page here calls the method ratified. The weights, the domain basis and the confidence rule are reproduced because they are what the document actually carries; the constants behind them are not, because a constant published as settled is worse than a constant not published at all.
How much the weights matter
The document reports its own sensitivity: changing any single domain weight by 0.05 moves the headline number by roughly a point or two, and the regrounded normalizers moved it by a few points. That is small enough that the score is not an artifact of one weight, and large enough that the weights are a real editorial choice rather than a detail.
What a reader cannot check here
The normalizers, the per-domain safe and high-risk thresholds with their citations, the protected area saturation curve, the humility ceiling and the region definitions all exist in writing. None of them are published at this site, so a reader cannot currently reproduce a score from the inputs. That is a disclosure decision rather than a missing piece of work, and it is the largest single gap in this section.