The narrow band
Open any personal rating list, yours or anyone's, and look at the spread. Ten points were available; almost everything sits between 6.5 and 8.5. Game critics have carried this joke for years, the infamous "7 to 9 scale," but private lists compress just the same, and it isn't laziness. Three quiet forces push every rater into the band.
Your sample is rigged
You never rate a random sample of the world. You rate what you chose: the film a trailer sold you, the restaurant a friend vouched for, the book that survived its first chapter. The genuinely bad candidates were screened out before you ever pressed play, so the honest average of what you actually experience really is high. The trouble starts when you score against an imagined universe of everything instead of your own pre-filtered sample. "Pretty good, for something I picked" rounds to 7, every time.
Borrowed anchors
Public aggregates pull your scale toward theirs. IMDb clusters in the mid-7s, so a film you liked "a bit more than average" lands at 7.8 almost by reflex. And even in a list nobody else will read, a 3 feels like an insult to something a person made. The result is a scale borrowed from the crowd and softened by politeness, which is to say, not really yours.
The unused tails
The endpoints sit idle for opposite reasons. The 10 is held in reserve for a perfection that hasn't arrived yet, as if spending it would leave nothing for later. The 1s and 2s go unused because you abandon their rightful owners early; things bad enough to earn them rarely get finished, let alone rated. With both tails empty, everything real gets squeezed into the middle-upper stretch that remains.
What compression costs you
A rating scale is a measuring instrument, and its usable range is its resolution. Work inside two effective points and the differences you actually care about drown in noise: 8.2 versus 8.3 is a coin flip, your tier list collapses into one long stripe of A's, and this year's 7.5 no longer means what last year's did. The list stops answering the one question it exists to answer: which of these did I truly like more?
How to recalibrate
- Redefine 5. Let it mean the median of everything you have tried, not cosmic mediocrity. In principle, half of what you rate should sit below it.
- Anchor the ends with real memories. Your best-ever is the 10; the worst thing you actually finished is the 1. Every new score becomes a placement between things you remember, not a negotiation with an ideal.
- Think in percentiles. A 9 is "top tenth of everything I've eaten," which is a claim you can sanity-check, unlike "excellent."
- Audit your histogram. A healthy distribution has spread and tails. A bell curve centered on 7.5 with nothing below 5 is a calibration warning, not a lucky streak.
- Re-anchor occasionally. Revisit a handful of old ratings each year against your current anchors. Drift is continuous; correction should be too.
Or sidestep the scale entirely
There is also a way around the problem rather than through it. Pairwise ranking never asks "how good?" at all, only "which of these two?", and that question is immune to compression. Your band can be as narrow as it likes; the duels still put the list in true order.
RateTheThings works both ends of this. Every list has a histogram so you can see your spread at a glance, the stats page nudges you when scores bunch, and pairwise duels resolve the neighbors your numbers can't separate. If your last twenty ratings were all 7-point-something, open your list and look at the histogram. It knows.