7 ms·

Federal Judge Unseals New York Crime Lab’s Software for Analyzing DNA Evidence

33 points by kensai 9y ago

cratermoon 9y ago
Source on github: https://github.com/propublica/nyc-dna-software https://github.com/propublica/nyc-dna-software I wonder if the username/password listed here is still in production: https://github.com/propublica/nyc-dna-software/blob/master/FST.Web/FST_Production_Service/Connection.txt https://github.com/propublica/nyc-dna-software/blob/master/F...
Alex3917 9y ago
Previous discussion: https://news.ycombinator.com/item?id=15518364 https://news.ycombinator.com/item?id=15518364
coldcode 9y ago
Reading the expert opinion it appears that FST did not use (or that there was no evidence given for one) a source code repository, thus it is impossible to identify how the source code changed behavior between the original analysis and that done with the given source code. I wonder if they did this to avoid keeping a change log useful in discovery.
- slaymaker1907 9y ago
  That's kind of like discovering that a building's blueprints consist of some doodles on a napkin. If I were the plaintiff, I would argue that not using a VCS completely removes all credibility of that software and company.
- pmorici 9y ago
  You are giving them too much credit. The simplest and probably correct explanation is that they weren't competent.
NotSammyHagar 9y ago
This should be a good reminder that the only way we can have safe and just criminal courts is that the evidence, techniques and technology used by experts and prosecution are public and reviewable. Whether it's the software that a dui measuring blood alchohol system uses, how they analyze your dna, it all needs to be public so we can measure things. I keep reading cases about how poor people get accussed of a crime and the "evidence" against them can't be understood or defended against because it's coming from unassailable experts with mystical skills. A recent example I read about is the police trained to detect people high on drugs, where there is no real evidence behind the training and many known cases of false accusation, http://reason.com/blog/2017/05/12/georgia-pd-our-drug-recognition-experts http://reason.com/blog/2017/05/12/georgia-pd-our-drug-recogn.... Most of us fortunate to work in the technology world have money and access to lawyers who can defend against such accusations, but poor people don't have that. I am not a cop but I know it must be incredibly hard to be one, and frustrating to see people get off of on technicalities. But poor l.e. practices must be prevented so we can have trust in the system.
rudedogg 9y ago
I wouldn't bet my life or anyone elses on this code being correct https://github.com/propublica/nyc-dna-software/blob/master/FST.Common/Comparison.cs https://github.com/propublica/nyc-dna-software/blob/master/F...

33 points by kensai 9y ago

Source on github: https://github.com/propublica/nyc-dna-software https://github.com/propublica/nyc-dna-software I wonder if the username/password listed here is still in production: https://github.com/propublica/nyc-dna-software/blob/master/FST.Web/FST_Production_Service/Connection.txt https://github.com/propublica/nyc-dna-software/blob/master/F...

Alex3917 9y ago

Previous discussion: https://news.ycombinator.com/item?id=15518364 https://news.ycombinator.com/item?id=15518364

coldcode 9y ago

Reading the expert opinion it appears that FST did not use (or that there was no evidence given for one) a source code repository, thus it is impossible to identify how the source code changed behavior between the original analysis and that done with the given source code. I wonder if they did this to avoid keeping a change log useful in discovery.

slaymaker1907 9y ago
That's kind of like discovering that a building's blueprints consist of some doodles on a napkin. If I were the plaintiff, I would argue that not using a VCS completely removes all credibility of that software and company.
pmorici 9y ago
You are giving them too much credit. The simplest and probably correct explanation is that they weren't competent.

NotSammyHagar 9y ago

This should be a good reminder that the only way we can have safe and just criminal courts is that the evidence, techniques and technology used by experts and prosecution are public and reviewable. Whether it's the software that a dui measuring blood alchohol system uses, how they analyze your dna, it all needs to be public so we can measure things. I keep reading cases about how poor people get accussed of a crime and the "evidence" against them can't be understood or defended against because it's coming from unassailable experts with mystical skills. A recent example I read about is the police trained to detect people high on drugs, where there is no real evidence behind the training and many known cases of false accusation, http://reason.com/blog/2017/05/12/georgia-pd-our-drug-recognition-experts http://reason.com/blog/2017/05/12/georgia-pd-our-drug-recogn.... Most of us fortunate to work in the technology world have money and access to lawyers who can defend against such accusations, but poor people don't have that. I am not a cop but I know it must be incredibly hard to be one, and frustrating to see people get off of on technicalities. But poor l.e. practices must be prevented so we can have trust in the system.

rudedogg 9y ago

I wouldn't bet my life or anyone elses on this code being correct https://github.com/propublica/nyc-dna-software/blob/master/FST.Common/Comparison.cs https://github.com/propublica/nyc-dna-software/blob/master/F...