48 REVIEWED SWITCH RECORDS / 3 NON-INTERCHANGEABLE SYSTEMSHOW SPECIFICATIONS ARE SOURCED
KEYSWITCH LAB

Dataset and matching methodology

How a switch record enters the database, how candidates are gated and how personal observations remain reproducible.

PUBLISHED BY KEYSWITCH LABLAST REVIEWED 20 AUGUST 2026

Record inclusion

A model enters only when a manufacturer product page or data sheet identifies it and supplies fields that can be mapped without guessing. Model identity, family, mechanism, force term, tolerance, actuation or pre-travel, total travel, fit group, pin field, factory state and life claim are retained only when present.

Force terms do not collapse into one number

Operating force, tactile force, initial force, return force and bottom-out force are not synonyms. MX and low-profile rows generally retain operating force. Hall-effect rows can retain initial force. The field label remains visible in the database, comparison and detail page; missing curve points stay missing.

Fixed and adjustable actuation remain different

A mechanical pre-travel value locates a published operating point. A Hall-effect switch can expose keyboard-adjustable actuation. The latter is stored as adjustable—not as 0 mm, not as a guessed default and not as a directly interchangeable fixed mechanical point.

Source boundaries come before numeric similarity

The finder first requires the same electrical switch system. Gateron KS-33, Kailh Choc V2 and CHERRY MX Low Profile also remain in their separate explicit fit groups; Hall-effect polarity or flux support remains a keyboard check. Ordinary MX records have no retained universal fit group, so their results stay labelled physical-fit unresolved. Candidates are then ordered by force and travel with visible differences for family, pins, lubrication, force terminology and life-claim status.

A tied row is not an identical switch

Two records can share every normalized number while differing in materials, full force curve, damping, sample variance, magnet specification or keyboard result. The interface exposes unresolved fields instead of publishing a zero-distance score that implies equivalence.

Personal test records

The test bench attaches ratings to a keyboard, plate, mount, caps, case or dampening, sample size and procedure. Records stay in local storage. They are observations, not laboratory measurements or public review averages.

Update policy

Sources were checked 13 August 2026. A source change, corrected field, compatibility clarification or model-generation issue should be dated in the corrections record and the affected entry reviewed.