Methodology
How specs get on this site, and how recommendations are produced.
Where specs come from
Specs come from manufacturer specification pages, Kelley Blue Book, Edmunds, and similar public sources. Every spec field on this site has at least one entry in the model's sources array.
Verified vs unverified
Each source carries a verified: boolean flag. Until a human has reviewed it, every numeric field on that model renders as Needs verification (or a dash in tables) with a visual badge. We don't publish numbers we haven't sourced.
Recommendations are deterministic
The quiz uses a fixed mapping from answers to a weight vector across six persona dimensions, then computes a weighted dot product against each model's persona scores. There is no language-model call anywhere in the recommendation path, and reasons shown to the user are templated from the user's answers.
Editorial fields require humans
Owner complaints, watch-outs, and best-for buckets are written by hand and reviewed. We don't auto-generate editorial copy, and we don't make comparative claims between named models without a sourced basis.
No fabricated numbers
If a number isn't sourced and verified, it's null in the data file and Needs verification in the UI. This is the single most important rule on the site.
Current MVP state
The MVP scaffolds the entire site with all numeric data set to null and persona scores uncalibrated. The verification pass is a manual editorial process that follows MVP launch.