> What I learned from 10 years of work around the semweb is how to build a very sloppy system and apply 'data lake' techniques to full scan it I mean, bully fo

> What I learned from 10 years of work around the semweb is how to build a very sloppy system and apply 'data lake' techniques to full scan it I mean, bully for you - but from interviewing data engineers for the past 5 years what virtually everyone else seems to have learned is "we have to frontload every feature with weeks/months of ontological discussion and then I'm lost because I never actually had to process any data quickly, let alone deal with dirty raw stuff." As a person who was skeptical from day ~1 of the semweb vision, everything I see around it still feels like an "I told you so" but in a really sad way.