Digital Evidence in Youth Court
You can't do forensic science without measurement. These days most forensic measurements are done using an instrument with a digital display. You can't form an opinion about the measurement result reliability of a digital instrument without considering the stability of the signal.
The instrument in the video above is in its "DVM test" mode. The instrument has nothing but air in its sample chamber. The instrument is "at rest". It is not yet doing a measurement. We are just looking at an "indication" of a numerical value of the electrical signal while it is at rest.
The video depicts "drift". The signal is going up and down - a lot. It is normal for the signal to vary a little bit. This signal is drifting a lot. I suspect that the reason for the drift in this particular example has to do with dirt in the sample chamber or malfunction of the light bulb or light detector. This instrument needs servicing - it hasn't been properly maintained for years.
It is important to realize that "drift" is something that happens "over time". In this case you can see drift over a few seconds. The signal varies from 402 to 411 in this video. This is "short-term drift".
Many electronic measuring instruments (including this one) have self-diagnostics systems that run on start-up, before a measuring sequence, or during a measuring sequence to try to flag short-term drift, render an error signal if such drift (positive or negative) is detected, and then shut down the measurement sequence preventing a measured result. On this particular instrument, shown in this video, the instrument started up and ran its automatic self-diagnostics, but the self-diagnosis passed. In other words, during the particular period "over time" that the instrument ran its self-diagnosis neither the positive nor negative short-term drift was enough to sett off the alarm. During self-diagnostics this drift is sometimes referred to as "positive stability" or "negative stability".
In the following video you will see the same measuring instrument, but this time the DVM test values are in the 200s. Notice at the beginning of the video that the instrument fails a self-diagnostics test (processor error) and then when it goes into DVM test mode, notice how much the values are wandering. However, if you watch the whole video, you will notice that the operator is able to get beyond the error message, start it up again properly passing the automatic diagnostics test. Subsequently the operator runs a stand-alone test and gets a "Diagnostics OK" message and a printed card record showing "Diagnostics Passed" including "Positive Stability" and "Negative Stability".
The point is that during one very short "over time" period you may get enough positive and negative stability to pass the self-diagnostics and even print an evidence card to use in Court to try and prove such stability. However, when a DVM test (or several DVM tests) is run you can see for yourself how much the signal is wandering.
If a police officer is trying to establish guilt using a measuring instrument that has a problem with signal drifting (I'm not suggesting anyone should or would do this), perhaps police should just "turn it off and turn it on again" until they get "proof" of a passed self-diagnostics - until the instrument appears to run normally. If police only use the diagnostics "pass" in Court and are never required to disclose the maintenance records, the diagnostics "fail", or the electronic "audit trail" data in the instrument, then no one will ever know that the stability of the instrument's signal is compromised.
Why is this important? A stable signal is necessary both at the time zero is set and at the time of sampling - when the measurement result is indicated. These are usually two different times that may be several minutes apart. If zero, at the time zero is set, is lower than the relevant zero, at the time of the sampling, then the measurement result will be inflated. On electronic measuring instruments such as this one, zero floats. Zero can get set by the instrument, for example, at 213, 320, 417, or 480. The final measurement result is the value of the signal for the substance in the sample chamber minus 213, 320, 417, or 480. If zero is set at 417 and the measurement of the sample takes place when zero is at 480, then a result where the signal for the substance in the chamber is 550 becomes 550-417=133 rather than 550-480=70. The result is terribly inflated because of a significant zero error.
We can exclude (or minimize the possibility of) errors like this by looking at the track record of the instrument long-term "over time", and excluding the possibility of a history of bad control checks, inconsistent control checks, previous diagnostics errors, and drifting DVM during maintenance tests.