Notes on talks from the 2019 Game Developers Conference in San Francisco.

Highlights here for me were:

The idea that much of indie success today is built on YouTube and streaming, so you should aim to create a game that appeals to the kind of people who want to generate that content (Decorticating “Dead Cells” talk); more than anything, that means building sticky “endless situation generators” rather than “consumable” games that you play through once and put away (The Shape of Success Before & After the Indiepocalypse talk)

The idea that your marketing communications should highlight the changes you make based on community feedback as a way of fostering that community (Decorticating “Dead Cells” talk)

The fact that the Dead Cells developers spent $50k on a trailer for a game whose entire development budget (including that trailer) was roughly $400k (Decorticating “Dead Cells” talk)—if both indies and AAA studios see value in these super “produced” trailers, we should probably try them out.

The insight that you should mock complex systems like physics using real data, rather than trying to either a) write the mock behavior yourself (like “always return 1 here”) or b) deeply couple the complex system to the test (Automated Testing Process roundtable)

The repeated recommendation for Michael Feathers’ Working Effectively with Legacy Code as being relevant to game dev (Automated Testing Process roundtable)

The insight that much of the roughly 2× difference in app developer revenue between iOS and Android can be attributed to the fact that Android is way more popular in low-spending countries like Russia & China, and if you compare just the US & Europe, you find much more similar sales (Paid Acquisition on Mobile: A Scrappy, Practical Guide talk)

The idea that QA should help determine up front (before development starts) what an acceptance test for a new feature will look like, with an eye toward helping devs design things in a way that is testable—this ensures QA knows what a feature is supposed to do too! (Preventing Bugs: The Dark Matter of QA talk)

Learning about AMD’s μProf profiler and seeing their sample code for calculating the number of hardware threads available on their CPUs (Ryzen Optimization talk)