How to make better technical decisions

@mthenw

Maciej Winnicki Engineering Manager @ OLX @mthenw

@mthenw making decisions is hard

@mthenw bounded rationality

@mthenw for his pioneering research into the decision-making process within
economic organizations Herbert A. Simon Nobel Prize in Economics in 1978

@mthenw cognitive biases

@mthenw for having integrated insights from psychological research into economic
science, especially concerning human judgment and decision-making under uncertainty Daniel Kahneman Nobel Prize in Economics in 2002

@mthenw The four villains of decision making

@mthenw

@mthenw narrow framing

@mthenw conﬁrmation bias

@mthenw short-term emotion

@mthenw overconﬁdence

encounter a choice analyze options make a choice live with
it @mthenw

it narrow framing makes you miss options @mthenw

it narrow framing makes you miss options the conﬁrmation bias leads you to gather self-serving information @mthenw

it narrow framing makes you miss options the conﬁrmation bias leads you to gather self-serving information short-term emotion will often tempt you to make the wrong one @mthenw

it narrow framing makes you miss options the conﬁrmation bias leads you to gather self-serving information short-term emotion will often tempt you to make the wrong one you’ll often be overconﬁdent about how the future will unfold. @mthenw

@mthenw

@mthenw Widen your options

@mthenw multiple options reduce ego

@mthenw knowledge

@mthenw read books not blogs

@mthenw conferences

@mthenw

@mthenw GitHub stars are

@mthenw cognitive load

@mthenw The Magical Number 7±2

@mthenw

@mthenw total cost of ownership

@mthenw

@mthenw Reality-test your assumptions

@mthenw experiment

@mthenw ask an expert

@mthenw RFCs

@mthenw

@mthenw https://engineering.squarespace.com/blog/2019/the-power-of-yes-if

@mthenw Attain distance before deciding

@mthenw 10/10/10

@mthenw Prepare to be wrong

@mthenw post mortem

@mthenw 5 whys

@mthenw Users weren’t able to log in. Why? One auth
header was missing. Why? We changed an auth library and the new one expected additional header. Why we didn’t know about the new header? Because we didn’t read the docs. Why we weren't able to quickly revert the false release? The release pipeline was failing. Why? Production credentials were not in the environment. Why? We wanted to speed up deployment and disable protection on the master branch. Why? We didn’t have hotﬁx process. Why we deployed the faulty release? There are no tests covering authentication. Why? It was written at the beginning of the project when we didn’t have time to write tests. Why code coverage tool didn’t catch it now? Because it was misconﬁgured.

@mthenw pre mortem

@mthenw widen your options reality-test your assumptions attain distance before
deciding prepare to be wrong

@mthenw Thanks!

How to make better technical decisions

How to make better technical decisions

More Decks by mthenw

Other Decks in Technology

Featured

Transcript