Skip to content

Keeping the Score Readable

How this page is sourced

Sourced from vendor documentation. Checked against its sources. Not yet reviewed by a person.

A scenario you built is a measuring instrument. Every random element in it is noise on the reading, and noise you added yourself is the kind you can remove.

  • Randomness hides what you did. A score that moves on its own teaches nothing.
  • Weapons are the usual source. Spread and recoil have patterned forms and looser ones.1
  • The score model is a second instrument. What it counts is what you will practice.
  • Strip first, then judge. Remove the noise before concluding anything about your aim.

Noise you added by accident

A random cone is noise, not difficulty. Per bullet spread uses a pre-determined pattern, where plain spread does not.1

The difference decides whether a miss was yours. With a cone, two identical inputs give two different outcomes, so a bad run may be the weapon rather than the aim.

Two panels, each firing two volleys at the same target. With spread, the second volley lands in a different scatter from the first. With per bullet spread, the second volley repeats the first exactly.Bassel Bakrhttps://github.com/Bassel-Bakr/aimSpreada new pattern every volleyPer bullet spreadthe same pattern, always
Two volleys each. Spread lands somewhere new. Per bullet spread repeats itself exactly.

Recoil has the same split. Recoil sets vertical and horizontal kick. Per shot recoil gives a consistent, learnable pattern.1

A learnable pattern is a skill. A random one adds an uneven cost to every run.

Two panels firing two bursts each, with the shots joined into a climbing line. Under recoil, the two bursts climb along different lines. Under per shot recoil, both bursts climb along exactly the same line.Bassel Bakrhttps://github.com/Bassel-Bakr/aimRecoila new climb every burstPer shot recoilthe same climb, learnable
Two bursts each. Only the per-shot pattern climbs the same way twice, which is what makes it learnable.

What to strip and what to keep

Anything not under test should be as free of noise as it can be. That is not the same as making the drill easy: the demand you built the scenario for stays exactly where it was.

Source Keep it when Strip it when
Weapon spread The drill is about spray control The drill is about where you pointed
Random recoil Never, for a drill you score Always, for a drill you score
Reactive bots You are training reading You are training a repeatable motion
Wide spawn scatter Distance variation is the point You are isolating one distance
Several dodge profiles You want mixed behavior You want comparable runs

Strip in one pass, then leave the scenario alone. A scenario you keep adjusting has no history, and history is the only thing that gives a score meaning.

The score model counts too

Whatever the scoring rewards is what you will practice. A model that pays for hits and never charges for misses trains volume. One that subtracts for misses trains restraint.

Neither is wrong. Inheriting one by accident from the scenario you copied is, because then the drill is teaching something nobody chose.

The same nine shots, five hits and four misses, scored twice. A model paying only for hits reaches fifty. A model subtracting for misses reaches ten. The run was identical in both cases.Bassel Bakrhttps://github.com/Bassel-Bakr/aimHits onlyrewards volume+100+10+100+100+1005 hits at +10total 50Hits minus missesrewards restraint+10-10+10+10-10+10-10+10-105 at +10 · 4 at -10total 10
The same nine shots, scored two ways. The run did not change. What it was worth did.

Scoring puts a price on the trade between speed and accuracy. Push the pace and accuracy tends to drop.2 The score model decides whether that trade pays.

Common mistakes

  • Reading a score from a scenario with random spread as a measure of aim.
  • Removing randomness and also removing the demand, leaving a drill that is merely easy.
  • Keeping the copied scenario's scoring without checking what it pays for.
  • Changing the noise and the difficulty in the same edit, so neither result is readable.

In practice

Ask what else could have moved that number. If the answer is anything other than you, the scenario is measuring two things.

Do this next. Open the weapon profile of a scenario you built and check for spread. If it has any, make a copy without it and compare a session on each.

Resources

References

  1. KovaaK's, Weapon Profiles, KovaaK's Wiki · REF-86 ↩↩↩

  2. Aimlabs, The speed-accuracy tradeoff and what it means for your aim training · REF-17 ↩