These news stories from Australia continues a growing realisation that ‘forensic’ science has little science but a lot of forensic; a trend which started over a decade ago; some would say even longer ago than that.  (Forensic generally meaning ‘solving’ crime, which is interpreted as a successful prosecution).  There is a widespread belief that DNA, once termed the ‘gold standard’ of forensic science because of its roots in actual science, is somehow immune from the problems caused by pushing science beyond what can and should be inferred from the evidence.  To suggest that critics of forensic science have an agenda is correct but probably gives no credit to what that agenda is; better science as used in ‘forensics’ and a need for proper scientific and legal scrutiny of every piece of science used in courts.  With the introduction of low template DNA mixture interpretation and probabilistic genotyping, even DNA profiling has fallen far from the claimed ‘Gold Standard’ it claimed over two decades ago. 

There has been a constant desire by some to extract more and more ‘evidence’ from smaller and smaller samples of DNA.  In a relatively short time the ‘science’ has gone from dealing with unambiguous results which few would argue with to samples which show variable results and ‘probabilistic genotyping’ programmes which show massively different results from the same sample treated in ostensibly identical fashion.

It is to be hoped that those tasked with investigating the problems with forensic ‘scientific’ evidence will take the opportunity to thoroughly, sceptically and objectively assess whether the current output from laboratories provides the reliability and objectivity that the courts should surely require.

We recently submitted a letter to a well-known journal, as well as proposing the same experiment at a public meeting, suggesting that the consistency of statistical results from probabilistic genotyping should be tested by sending the same sample to different laboratories to see if they produced the same result.  It was rejected.

Isn’t this exactly the sort of assessment that states or groups of laboratories within the same jurisdiction, should be doing to convince the courts that the statistic they are producing is reproducible and reliable?  Would we not expect the same result from each lab if we wanted them to measure the amount of alcohol in a blood sample?

For years we have been challenging the validity and reliability (and comprehensibility) of the statistics being used in DNA casework using programmes like STRmix, LikeLTD, LiRa, and TrueAllele.  The default position for the vast majority of courts is a demonstrable ignorance of the issues but an acceptance of the results as being somehow admissible despite the demonstrable problems with reproducibility and comprehensibility.  As the NAS report stated in 2009, “the courts have been utterly ineffective”.

We hope that this is something that will feature in investigations into actual and potential forensic science failures.  Perhaps another inquiry should examine whether anyone in the legal profession or lay juries have any idea what the statistics actually mean!