Hacker News Reader

Comment by kevin_thibedeau | Hacker News Reader

Given the fallibility of DNA testing they may as well be using a divining rod to declare someone a witch. This will be abused to falsely accuse innocent people. Lives will be ruined and the authorities will act like their hands are clean. 23andMe is not a forensics lab. Their techs have no incentive to avoid mistakes with cross contamination. This is just plain dumb.

View on HN

In this case, sites like 23andMe were only used to generate leads on suspects. Two independent forensic DNA tests were used for evidence of guilt. I agree that DNA testing has flaws, but that's the purpose of the criminal trial. Defendants have due process rights to cross-examine the DNA testing process, present their own testimony, and provide exculpatory evidence.

I think you're confusing an investigation with a trial. This technique seems like a breakthrough for narrowing a list of suspects. But your claim that this would be used as evidence in a trial is unfounded and a bit silly. They have him. They have the DNA of the killer. They can perform their own testing without relying on any online DNA test results at all.

In this case. But what about the next guy who is a false "match" for a rapist and has life turned upside down because of it.

They took samples (probably from a trash can). If it wasn't a match,I'd hope they wouldn't have 'turned his life upside down's anyway.

If DNA tests are so unreliable, how come they are still accepted in court?

Because the public is ignorant of the problems with it and the judiciary likes tools that can win easy convictions. Breathalyzers don't actually measure blood alcohol and there is no credible way they can estimate it to three significant digits but courts allow them too. Polygraphs are pure bullshit but the examiner's word that someone is "lying" is taken as sacrosanct evidence in court.

Polygraphs are not admissable IIRC.

IIRC The Wire had an accurate description of its use as an interogation tool (note I did not say investigative tool).

I don't know if this is the commenter's intent but isn't there a potential false positive issue with DNA testing?

Something along the lines of if only 0.01 percent of the world's population can match a DNA sequence found that still leaves us with a million matches where I just made up both those numbers.

Fallability which way though? Do you disagree then when death row inmates are exonerated due to DNA testing?

Careful, your bias is showing..

> Careful, your bias is showing..

That isn't bias but rather an understanding of burden of proof. If you think it is, then you are very fundamentally misinformed about how our society functions.

This 'bias' is not necessarily inappropriate because of the inherent asymmetries of DNA evidence - it is statistically more powerful for exoneration than conviction. This is because of the False Positive Problem (similar to the Birthday Problem in probability): if the actual incidence of a phenomenon is lower than the false positive rate of the test, a positive result is likely erroneous even on a highly accurate test.

This is why a strong prior, developed through traditional policework, is necessary to increase the reliability of DNA results - most results obtained through database trawling will be false - and these will be numerous if the database is large enough. It is also part of the reason why exonerations are more robust than identifications.

This particular case may not suffer from these statistical problems, but they are worth keeping in mind, as it sets a precedent.

Reply on news.ycombinator.com