Modelling e-target accuracy requirements

I would like to propose a method I have in mind to determine how much accuracy we need for fullbore target systems.  It is an attempt to derive something independent of the ISSF model; because I honestly don't know if their model is sound or arbitrary, I thought it safest to try a clean-sheet approach.  Please let me know if you think it is a good approach or a flawed one.  I will follow up with a deliberately-separate email in which I suggest numerical values for the model that I propose.  Obviously if the model is wrong or flawed, there is no need to discuss numbers for it.

Proposal for establishing e-target accuracy requirements

for Fullbore TR and F-Class shooting

• Establish a statistical model of the grouping characteristics of a shooter+rifle+ammo, under ideal conditions (i.e. zero wind), for the most demanding scenario(s) of each shooting discipline under consideration.  This model must give group patterns and shot distributions (V-counts vs. 5-counts vs. 4-counts) that experienced shooters will recognize and accept as being a valid model of real groups fired by shooters.
• Choose an "allowable error" that we are willing to accept as the price of the various conveniences of accepting e-targets into our sport.  In making this judgement we should be thinking not in terms of millimetres or MOA but rather by how much error in our scores is acceptable and whether or not this would make a meaningful difference in the outcomes (placings of shooters) of the matches we shoot.
• Choose group size parameters to reflect the ability of a world-class shooter under ideal conditions.  From deciding this, we should see our model generate simulated scores that match our expectations and experience as shooters and coaches, and also gives a reasonable fit to actual observed shooting scores (short range aggs at national matches, observed 300m ISSF scores, etc).  This should be repeated for each discipline of interest (e.g. TR would have a less stringent accuracy level requirement than F-Class would, and offhand shooting would have a lower accuracy requirement than TR).   The idea is that we need to make sure that even when the very best shooters are firing under the very best conditions, we are still able to deliver results that are "fair" (I suggest "fair" be defined something along the lines of there being no meaningful change in scores caused by e-target measurement errors)
• Given the assumed performance of an idealized top-talent shooter (which we express in MOA but we assess in terms of the scores and V-counts we expect), and the amount of error we are willing to accept in his reported scores (expressed as a certain number of points or Vs in an aggregate of a certain number of shots), calculate the amount of shot position measuring error that the e-target system is permitted to introduce in order to produce this "maximum allowable error"
• This "shot position measuring error", expressed in millimetres or MOA, would then be a criterion that e-target systems would have to live within.

Some characteristics of this approach that I can think of:

• We are fitting the accuracy of the system to the capabilities (observed, imputed etc) of the shooters using it
• the scoring ring dimensions of the target system we shoot has no direct link to the level of accuracy that we are looking for
• This approach determines a "good enough" level of shot measurement accuracy that gives us an error in our results that we have decided we are willing to accept.  This is in contrast to an approach in which we try to determine the "best achievable" level of measurement accuracy and then report what that gives us and what it does not give us.

As a comment, I would not be surprised if a longer term result of the fullbore world looking into accuracy requirements for e-target system is a more fundamental understanding of how we mark shooter performance (i.e. scoring ring dimensions).  My suspicion is that there are probably some things about the scoring ring sizes we have chosen that are "wrong" (i.e. are improvable) and if so then perhaps we might reform/improve our target designs over the next ten years or so.

For what it's worth I expect all e-target manufacturers' systems in ordinary use will be able to easily attain a level of shot measurement accuracy well beyond what is needed for the very highest levels of TR and F-Class shooting talent.

Of greater concern to me is how abnormal or fault conditions are dealt with and detected, since it is unlikely that a "challenge" procedure can be designed.  It is vital that e-target systems used for significant competitions incorporate some means of error checking or correction.   In my opinion an e-target system ought to not only measure the position of a shot but also measure the possibility that an error has occurred and if so it should report this so that the shooter may be given the benefit of the doubt.  The details of this will depend on the design characteristics of each e-target system but for example if a particular e-target system design has a low but nonzero probability of a near-simultaneous shot on an adjacent target producing an erroneus but plausible shot position being reported, it would be *highly* desirable for the system to detect this and include some sort of "quality report" noting that an out-of-nominal measurement condition has occurred.  For example

• "I have detected a shot arriving on target but due to a detected measurement error am unable to calculate shot position" (presumably the rules would be written so that the shooter would be entitled to take another shot)
• "Here is a 5 (or 3) that you have just shot, it is likely to be correct but I have detected a possibility that this might be in error" (presumably the rules would be written so that a shooter could elect to take another shot)
• "Here is a 4 that you shot.  It is only 4mm outside the line but all sensors indicate a valid measurement and no fault conditions detected.  Yes I know you are one of the best shooters in the world and that you have just fired 7 Vs in a row but this is a 4 and you are legitimately stuck with it" (no you won't hire me to compose the wording that the e-target display panel shows but it is important to be able to give Match Committees a basis to make this sort of judgement)

Anyhow I think "shot quality" measurement is an important thing but is in a way a distraction from this email; let's deal with that separately.

Back to the topic of measurement errors.

Whatever method we take in establishing an acceptable level of measurement error, I would like to suggest that the following be considered:

• precision.  The figure being discussed above and in most of our correspondence (and in my opinion the single most important parameter), is actually the shot measurement precision, and where necessary should be distinguished from technical meaning of the word accuracy.  If you'll forgive a descent into electrical terms the precision that we are interested in is the ac component of the radial shot measurement error and it is different (probably largely random) from shot to shot.  The dc component is a fixed (same from shot to shot) error that can be incorporated into the calibration offset constants.
• accuracy.  The actual absolute accuracy of the e-target system is much less important than ensuring that the precision is high enough.  A small amount of offset error applied to each and every shot is relatively benign (I expect that at 300 it would be acceptable to allow say a 1/4-MOA calibration error and at 1000 to allow a 1/2 MOA calibration error).  The effect of fixed error offsets is to interfere with the reliability of a shooter's wind zeros and elevation zeroes.  Allowing an e-target system to have on the same order of error as a shooter actually does (note, don't poll shooters for this, poll coaches!! ;-) ought to be "acceptable", in the sense that it won't be making things any worse for them.
• variable precision.  It might be a good idea to consider spec'ing a certain high level of precision for the central (say 5,V) region of the target, and permitting a looser level of precision elsewhere.  The reason I suggest this is in case a manufacturer has a technical design which allows him to favour measurement precision at the centre of the target at the expense of precision elsewhere, he should be encouraged to optimize precision where it actually matters.  My justification for this is that the farther away from the centre of the target a shot lands, the larger an error has been made by the shooter - my argument is that the measurement system's error budget ought to increase accordingly

I think it is important for shooters to know and to understand that while yes there are nonzero measurement errors made by e-target systems that a careful experimenter will be able to measure and show, it is also the case that these errors are small enough that they truly don't affect reported scores.  We should probably articulate and enumerate the effects of us accepting this amount of allowable error.  For example, take actual standings results from various national matches and examine how (or if) the standings would be affected by plausible e-target system errors.