It looks like you're using an Ad Blocker.

Please white-list or disable AboveTopSecret.com in your ad-blocking tool.

Thank you.

 

Some features of ATS will be disabled while you continue to use an ad-blocker.

 

Secret source code pronounces you guilty as charged

page: 2
21
<< 1   >>

log in

join
share:

posted on Oct, 18 2015 @ 02:35 PM
link   
a reply to: roadgravel

I'm guessing all y'all missed this:


UPDATE: Sunday, 10:15am PT: Mark Perlin, the chief scientist and executive officer of Cybergenetics, says in an e-mail and telephone interview with Ars that "source code is not used to assess forensic software reliability." (More below.)


You should be noticing the text that says "source code is not used to assess (forensic) software reliability"

And it is absolutely true, n fact, the source code won' to help you to understand the results, only how they are reached in "this instance" (there are many ways to reach the same results).

It seems that a very large misunderstanding about software is at play here.

To verify the output of the software in question; test it...with several samples. Compare the results the software provides with the results of a "manual" analysis...it is really quite easy...and of course try to learn more about software and how it is produced, designed, tested, etc.

By the way...their software isn't 170,000 lines f code. Matlab is...maybe, but not this "software in question."




posted on Oct, 18 2015 @ 02:55 PM
link   

originally posted by: tanka418
And it is absolutely true, n fact, the source code won' to help you to understand the results, only how they are reached in "this instance" (there are many ways to reach the same results).

In my opinion both things are important in this case, as people's lives may be affected by it.

Yes, looking at the source code doesn't say what the result will be, but it can show you if there are any errors or "features" (like in the VW case) that may be looking for specific data to give specific results.

Imagine that someone added some code that identified some specific DNA characteristics and gave always positive (or negative) results. You can only find that if you test all possibilities or if you look at the source code.



posted on Oct, 18 2015 @ 04:42 PM
link   

originally posted by: ArMaP

originally posted by: tanka418
And it is absolutely true, n fact, the source code won' to help you to understand the results, only how they are reached in "this instance" (there are many ways to reach the same results).

In my opinion both things are important in this case, as people's lives may be affected by it.

Yes, looking at the source code doesn't say what the result will be, but it can show you if there are any errors or "features" (like in the VW case) that may be looking for specific data to give specific results.

Imagine that someone added some code that identified some specific DNA characteristics and gave always positive (or negative) results. You can only find that if you test all possibilities or if you look at the source code.


Yes...so vet for all possibilities, or at least a significant subset. This is software, it only reacts to data, not physical samples, thus it is possible to give it a very large set of simulated data. The process can even be automated...



posted on Oct, 18 2015 @ 06:21 PM
link   

originally posted by: tanka418
Yes...so vet for all possibilities, or at least a significant subset.

Considering the potential results I wouldn't accept anything less than all possible DNA variations, but I don't know if that's even possible.



new topics

top topics
 
21
<< 1   >>

log in

join