A scoring instrument for buying a car that doesn't watch you and that you can actually fix. Built from the argument in Old Cars & The Analog Resistance, then extended to cover what that argument leaves out. Everything resolves from a model year, a spec sheet, or a recall database, from your desk.
The reference set, and the fastest way to calibrate. It spans the full range, covers every marque the video names (Bronco, Land Cruiser, Tacoma, Volvo, Land Rover, Porsche, Mercedes), and includes ordinary appliances as a control group. Switch views, sort any column, load a car, then change whatever you disagree with.
MSRP and market are reference only. Neither one feeds the score, and neither can veto a car. An expensive car is not a worse car; it is a budget question, and yours to answer. Figures are approximate 2026 values, rounded, and move constantly.
The cliff survived an attempt to disprove it. Seven of these cars were added specifically to land in the empty 45–64 band and break the finding. Every one missed. The 2018 Civic came closest at 65, then the 2012 F-150 at 72, the 2013 4Runner at 78, the 2015 Wrangler at 80. Twenty-seven cars, and the Transitional band still holds exactly one: the 2014 E350 at 63.
The reason turns out to be a single component. Every car here without an embedded cellular modem scores 65 or above. Every car with one scores 43 or below. Nothing lands between. The modem does not arrive alone: it brings over-the-air updates, a remote path into the car, subscription gating, and a readable event data recorder, and those five criteria are worth about forty points together. That is the cliff. It is not a decade or a design philosophy, it is one part, and you can date it per model.
| Score | Band | What it means in practice |
|---|---|---|
| 85–100 | True analog | Almost certainly pre-1996. Passes on principle, not on technicality. Realistically a second vehicle. |
| 65–84 | Analog-leaning | Daily-drivable with real compromises. Usually private by accident of age rather than by design. |
| 45–64 | Transitional | The computer arrived but the leash hadn't. Fixable if you buy the right tool. |
| 0–44 | Connected | A modern car. It may wear the costume convincingly; it still reports home. |
Fourteen criteria, scored one at a time. Each carries a pointer to where the answer lives online, so a candidate can be fully scored before you contact a seller. Your answers are saved in this browser and in the page URL, so you can bookmark a scored car or send it to someone.
The video doesn't raise any of these, which is its biggest omission. A car can score 100 and still be the wrong car. Tick anything true; one tick outranks the whole score. Price is deliberately not here.
Safety. Deliberately kept out of the score and put in the vetoes instead, because averaging it in would let a strong analog score paper over a car with no airbags. It shouldn't be a tradeable point. It should stop the purchase or not.
Price. Shown in the table, outside both the score and the vetoes. A car is not worse for being expensive, so price never costs a point and never disqualifies anything. It is there because the numbers are interesting on their own: a 1966 Bronco went from $2.4k to $85k while a 1991 LS400 went from $42k to $7k, and that divergence has nothing to do with how analog either one is.
Electric drivetrains. The Longevity section is written around combustion engines, so an EV scores oddly there. A Tesla's motor genuinely is the mechanically simplest drivetrain in the set, and it earns full marks for that, which flatters it. What that section can't see is a sealed structural battery, and Repairability is where that shows up instead.
The specific example. This rubric scores a model. It cannot score the one car in the listing, which is what a pre-purchase inspection is for. Research picks the model; the PPI picks the car.