Update README.md (for personal project not related to CDC25)

README.md

@@ -12,6 +12,35 @@ Modern Earth is trending toward reduced habitability due to rising heat, sea-lev
 
 We operationalize this by extracting five key planet/star characteristics from NASA’s Exoplanet Archive, normalizing them to **Earth-relative units**, and building a transparent scoring function on \[0, 1\] where **1 ≈ Earth-like**.
 
+---
+## Limitations & Assumptions
+
+- **Observational incompleteness & bias.** The confirmed exoplanet sample is biased toward short-period and/or larger planets around bright stars (transit/RV selection effects). Many rows lack one or more features, so filtering can skew the sample further.
+
+- **Proxy features, not surface conditions.**  
+  - `pl_eqt` (equilibrium temperature) assumes zero/constant albedo and full heat redistribution; true surface temps can differ by tens of K or more.  
+  - “Orbital distance” and “insolation flux” are correlated; we reduce double-counting by down-weighting distance when `pl_insol` is present.
+
+- **Planet composition uncertainty.** Similar mass–radius pairs can be rocky, water-rich, or mini-Neptunes. We do not infer bulk composition or interior structure beyond simple mass/radius preferences.
+
+- **Stellar environment not fully modeled.** We don’t explicitly account for stellar activity (flares, UV/X-ray), magnetospheres, atmospheric escape, tidal locking climate effects, or long-term orbital stability—each can be decisive for habitability.
+
+- **Anthropocentric scoring.** We target human-centric constraints (Earth-like gravity, moderate temperatures/light cycles) rather than biosignature likelihood. A high score ≠ “habitable”—it’s a **prioritization signal for follow-up**.
+
+- **Point estimates, limited uncertainty.** Measurement errors and asymmetric posteriors are not propagated through the score; rankings can shift as the Archive updates values.
+
+- **Heuristic weights.** Weights are literature-informed but ultimately chosen by us; rankings are sensitive to these choices (the UI exposes sliders to make this explicit).
+
+- **Catalog drift & reproducibility.** The NASA Exoplanet Archive is updated frequently. Results reflect the snapshot time of our TAP query; see `data/README_sources.md` for the exact query and date.
+
+### Mitigations & Future Work
+
+- Add **uncertainty-aware scoring** (Monte Carlo draws over reported intervals) and show rank stability bands.  
+- Incorporate **stellar activity proxies** (e.g., flare rates, `log R'_{HK}`), metallicity priors, and simple **atmospheric escape** checks for small M-dwarf planets.  
+- Use **conservative/optimistic habitable-zone edges** (e.g., Kopparapu et al.) to derive a more physical insolation window.  
+- Explore **composition classification** from mass–radius relations to down-weight likely mini-Neptunes.  
+- Log data snapshots (DOI or hash) to make results **reproducible** across Archive updates.
+
 ---
 ## Data