Behind The Tech with Kevin Scott

There are scattered genuine insights - the LinkedIn data-center kill drill, the potential obsolescence of app paradigms, and using AI for root-cause triage - but the AMA format fragments every topic before depth is reached, and large portions of the transcript are filler agreement and vague platitudes about AI regulation and agents.

“at random points every week, just take a whole data center offline, uh, to make sure that all the fault tolerance systems would work”

“I don't know that you need apps in this world... the user interface telling someone they got to go learn all of the complexity of some software... versus just say what they want done”

The LinkedIn data-center chaos-engineering practice and the Ada Lovelace framing of software's unchanged paradigm offer mild novelty, but the AI regulation, agent, and rural-connectivity discussions recycle widely circulated takes without adding a genuinely contrarian or first-principles argument.

“the way that we've been building software hasn't really changed since Ada Loveless... this whole process of algorithmic thinking... We've been doing that for almost two centuries now”

“without telling Them, you just kill the whole system and you'll know real quick whether, uh, their application's robust or not”

Kevin Scott is genuinely high-caliber - CTO of Microsoft, former VP Engineering at LinkedIn, 40 years of hands-on programming - and he has demonstrably done the things he describes at scale; the AMA format, however, prevents him from showing the full depth that his experience could justify.

“at LinkedIn, we uh, used to, at random points every week, just take a whole data center offline”

“I've been programming for 40 years, um, so I'm 52. I started when I was 12”

The episode has pockets of genuine specificity - the LinkedIn weekly data-center takedown, a 300k DSL uncle, 1100°C raku firing temperatures, rural India AI diffusion anecdotes from Satya Nadella - but most of the AI regulation, agent timeline, and infrastructure scaling discussions remain abstract and unquantified.

“my uncle, who lives, uh, just a few miles away from them, is still on some kind of, like, crazy 300k DSL connection”

“I've got a set uh, of tea bowls that I uh, am firing in the classic raku style at 1100 degrees Celsius”

The host is consistently deferential - responding to nearly every answer with 'I love it,' 'great stuff,' or 'yeah, yeah, totally' - and rarely pushes for precision or challenges vague claims; the one genuine follow-up (asking whether the data-center drill was Scott's initiative) briefly surfaces something useful but the pattern of unchallenged agreement dominates.

“Yeah, yeah, totally. I mean, lots, lots of things to think about.”

“was that a process that um, started before you joined or was that something that you um, asked them to do?”