Thread by @tleam

#chaosday19 @gen_nja is getting started with his presentation “Black tie Chaos: Failing formally”

#chaosday19 @gen_nja : Providing a quick review of Auxon. Left the autonomous vehicle engineering and moved to complex systems at large.

#chaosday19 @gen_nja : Quick poll of who is a software engineer vs. who is a ChemE/MechE/etc "Everyone who is a software engineer, those other engineers don't think you're an engineer" *laughter and applause* "Who would trust a software engineer to build a skyscraper?"

#chaosday19 @gen_nja : Reviewing Fiat Chrysler recall for cruise control glitch. We've been doing this for a very long time and all of a sudden the adaptive cruise control would not stop when you put your foot on the brake.

#chaosday19 @gen_nja : The only way to overpower this was via e-brake OR to shut off the car while you were driving. Again, we've been doing this for ages... so what happened to this fairly standard process?

#chaosday19 @gen_nja : We think it has something to do with complexity. "There are 3 models to any system, the intended the squiggle potato that is the reality, and the third model is the assumption part" [ed: due to complex systems we'll never full match assumption to reality]

@gen_nja #ChaosDay19 We are taking a look at a FMEA analysis spreadsheet now and discussing how we try to identify how things will fail. This is one row, there are normally thousands. Still doesn’t match reality.

#chaosday19 @gen_nja : "We're starting to see the unfortunate transition in our systems where we're bolting on ever increasing layers of complexity and abstraction onto our manual systems" [ed: direct reference to the 737Max issues recently]

#chaosday19 @gen_nja : "To get through the certification and the updated flight characteristics of the new model they accommodated by adding automated compensation. There was a fundamental misstep of building resiliency into the system."

#chaosday19 @gen_nja : "Part of the reason for this is due to the level of abstraction. In civil engineering things have stayed pretty stable for a while. Planes now require many more engineering fields to build a plane"

#ChaosDay19 @gen_nja this is the issue we then run into. Modular development causes issues with integration

#chaosday19 @gen_nja : "How do we understand which of our assumptions are closer to reality and those that aren't right? Chaos Engineering helps surface known-knowns from unknown-unknowns"

#chaosday19 @gen_nja : The point of Chaos Engineering is to figure out how 'true' any of our assumptions are at all.

#chaosday19 @gen_nja : Current thesis - need to be writing things down. Need to put down what your assumptions are. There's little value in setting things on fire just to see what happens. [ed: scientific method! Develop your hypothesis first. FMEA is one format for doing so]

#chaosday19 @gen_nja : Its not very effective to measure experiments along one dimension. It's not as interesting to have diverse implementation but rather diversion observational elements

#ChaosDay19 @gen_nja All potato’s are dangerous

#chaosday19 @gen_nja : Thesis is that the method and results from chaos and input into a formal process. The blend of formal and informal provides a better outcome over time... but we could be wrong. /fin

@threadreaderapp unroll please and thank you