Alan Burlison wrote:
Garrett D'Amore wrote:

Trying to use a program to replace human oversight on this is, IMO, wrong. Until you have a program that has a full natural language parser and can understand the difference between an assertion that content is proprietary, and other legitimate uses of words like confidential and proprietary, this is going to cause a lot of grief.

The problem is that the humans who have been asked to do this haven't done a thorough job in some cases. Trusting people to be diligent clearly doesn't work.

So, instead of finding the outliers that need to be fixed, and fixing those manually, you're applying a draconian process whereby *all* materials for a case will be redacted if any of the "evil" words or phrases appear anywhere?

This is uncool. As I said, *every* case I've touched in the past couple of years was fully redacted because of errors either in your script, or nits in .ms files or errors made by the ARC coordinator in flagging the minutes "ENGINEERING ONLY".

Btw, I believe "ENGINEERING ONLY" carries no legal value. It doesn't indicate that the material is confidential, and is instead used to indicate that ARC meetings (all PSARC meetings) are generally only open to engineering. (Historically, Marketing folks and such were not invited to PSARC, only to prevent discussions at PSARC from drifting away from engineering content and into issues that properly belong with the business teams.)

Tracking down and fixing each of these cases is a PITA. Especially since I'm quite confident that none of my cases have any incorrect assertions (beyond that ENGINEERING ONLY comment in the ARC coordinators notes, which were *not* posted into the case by me or any other ARC member or participant) that any of the material is confidential or restricted in any way.

This feels like make-work to me.

As a "stop-gap", I'd be happy if the redacted case materials could be restricted to just the materials that have the incorrect assertion, rather than redacting the entire case just because some (useless IMO) ARC coordinator or .ms source file contains evil language in it somewhere.

   -- Garrett

IMO, there needs at minimum to be an override mechanism, where a file can be blessed as not having any bad assertions, without requiring the *content* of said file to be altered.

There is one - 'manual' exposure.  That's not been done properly either.

I had at one time believed that the process of marking a case open (which does involve additional human review of the materials, btw), implicitly performed this "whitelisting" step.

No, it doesn't.


_______________________________________________
website-discuss mailing list
[email protected]

Reply via email to