Methodology · Version 1.0

How every verdict on this site gets decided

No single test settles a tool category — so we don't pretend one does. Every guide scores the consensus across named, linked, independent sources with published weights, then anchors it with gear we bought with our own money. This page is the contract; when it changes, the changelog below says so.

Version: 1.0 Effective: [LAUNCH DATE] Changelog: initial version

The sources, and what each is worth

Each guide's consensus score blends four source classes. The default weights:

Source class	Weight	What it contributes	Known limits (always disclosed inline)
Measured independent tests	40%	Project Farm's torture tests, Torque Test Channel's dyno runs, magazine bench tests (Machinist's Workshop, Practical Sailor). Numbers, not vibes.	One rig, one rust profile, one day. Project Farm's own disclaimer calls his videos "only for entertainment purposes" — we cite his numbers because he's honest about that.
Our own testing & ownership	30%	Products we bought and used — measured where we can (breakaway torque, weight, runtime), long-term wear notes where we can't.	A one-person garage, not a lab. Every measurement states its method next to the number.
Field consensus	20%	Two decades of GarageJournal, TractorByNet, Yesterday's Tractors, and OPE Forum threads — the closest thing to longitudinal data this industry has.	Anecdote at scale. Loud minorities exist; we weight recurring independent reports, not single posts.
Spec & availability reality	10%	Is it actually sold, in what formats, safe on what materials, from a maker that answers the phone.	Manufacturer claims verified against user reports where possible.

When a guide departs from these weights (some categories have no dyno data; some have rich forum history and thin video coverage), the guide says so at the top of its scoring section.

The rules we hold ourselves to

Every unit is self-purchased. No manufacturer samples, no seeded review units, no sponsored placements — the same rule that makes the testers we cite worth citing.
We never claim hands-on experience we don't have. Every guide carries a "What we own / what we researched" box. Researched-only picks cite their sources inline.
The seal means something. The circular Owned & Tested stamp appears only beside products we bought and used ourselves. A pick without the seal is a researched recommendation — sourced, linked, and labeled as exactly that. If every pick carried the badge, it would mean nothing.
Conflicts between testers get reported, not resolved by hand-waving. When the forums crown a product that a measured test ranked last, that disagreement is the most useful fact in the category — it gets its own section.
Numbers are paraphrased facts with attribution and links. We never reproduce another tester's charts or tables; we build our own from multiple sources and link every one.
No prices on this site. Prices drift daily and Amazon's rules restrict quoting them; buttons say "check price" and mean it. Where a dollar figure appears, it's a manufacturer list price or a historical figure from a cited test, labeled as such.
Updates are real. The date on a guide moves only when its content materially changes; the update log on each page says what changed.

What the affiliate links pay for

Buying through our links earns us a commission at no cost to you. That revenue buys the next test batch — five cans of penetrating oil cost real money, and nobody sends us free ones. Full details on the disclosure page.

Placeholder — fill before launch Photo of the test bench / garage where the testing happens, with a caption naming the measuring tools used (digital torque adapter model, kitchen scale model, decibel app).