Looking back over my relatively short 30 year career across an assortment of tech companies (hp, google, microsoft, apple, etc) I would add that this mostly changed due to what was rewarded. Around 15 years ago when the first vestiges of OKR review processes and the idea of "impact" started to form; We shifted our designs from things that made our long term outlook better (simple) to things that were easy to explain to management how they had HUGE impact (complex).

Be right back, writing a new design doc for a messaging service and protocol spec to go with it that I can use to pad my next review cycle.