In looksmaxxing, PSL means the 0–8 facial rating scale inherited from the old PuaHate, SlutHate and Lookism forums — the community's strictest standard for rating faces and tracking improvement. Median 4.0, no inflation, one fixed ruler.
The scale wasn't designed — it accreted. On PuaHate and its successors in the early 2010s, users kept re-calibrating ratings against each other's photos until a consensus ladder emerged: 0–8, harsh anchors, structure over styling. When those forums declined, the vocabulary migrated to Lookism, then Reddit, then exploded on TikTok with the looksmaxxing wave.
The community's own record is unambiguous about the name: PSL = the three forums. Expansions like "proportion, size, lineation" appeared years later on commercial rating sites (we document all three claims here).
Casual 1–10 ratings drift. Today's "8" was a "6" fifteen years ago, and thread ratings swing with lighting and mood. A scale with a fixed median at 4.0 solves the one problem looksmaxxers actually have: "did my face change?" When someone drops body fat, fixes their eyebrows and retakes under identical conditions, a +0.4 on PSL is signal. On a vibes-based 1–10, it's noise.
That's also why the community polices "frauding" (angle/lighting tricks) so aggressively — the scale is only useful if measurements are honest.
The scale came with a dictionary. If you're reading looksmaxxing content, the terms you'll hit most: LTN / MTN / HTN (low/mid/high-tier normie), Chadlite / Chad, mogging (outshining someone facially), sub-5 (below median), hunter eyes (deep-set positive-tilt eyes), canthal tilt. Our looksmaxxing dictionary covers the full set — each entry with what it means and where it actually matters.
🔒 Free measuring runs entirely on your device. When you unlock the AI deep report you can choose to attach your photo — used only to generate that one readout, then discarded, never stored. Scores are geometric measurements, not a judgement of your worth.