Let’s talk about one of the absolute dumbest things in tech, and I’m reminded of this bc of the NFL draft coming up and how every year nfl teams do the same dumb thing people do in tech: over analyze things that don’t matter For a second let’s talk about what actually matters when hiring (or drafting) someone & it literally is just two things

1. Is the person capable, or capable of being capable
2. Can they take a punch to the face

Nearly everything else doesn’t matter in the grand scheme of things

Have a long flight so time to do a thread on a question I get in some form at least half dozen times a week

Where will value accrue in AI from here

Think: foundation models, apps, middleware, full-stack blah blah blah etc Note: I am writing this thread "old style" as broken up individual 280 char tweets. Why? Because I viscerally hate the long post reading format on Twitter and I don't care at all about this getting algorithm love

And, I still like the 280 per tweet limit forcing function

I hold a few very un-popular but likely hard truth views on the world and life

One of them is that most people will never achieve a single large goal they set out to do and it’s primarily because we as a society have taught them a very dangerous concept called ‘moderation’ I absolutely do not believe that if your goal is to do something even approaching very good let alone great, moderation is the way

Moderation is for mediocre. Perhaps more accurately , moderation is for maintaining

If your goal is to maintain your status quo, moderation

If I could give every founder/exec/leader/senior eng one piece of advise above all else it’s to figure out what matters & what doesn’t & literally to spend no cycles on what doesn’t matter, literally none

This is one of those things, though, that cannot be taught imo There’s just too many things that appear to matter in life and work but ultimately don’t or won’t. Experience can guide us here more than most anything else, but some come down to instinct or gut feel. Most people either lack the experience or intuition to do this well

I get asked a lot lot which management books I recommend

None. Management books are mid

Want to be a better leader? Start a new sports team in your local rec league, get players, get better, win the championship, keep that team together for a few seasons. Seriously Unless you are literally unexperienced in any facet of corporate life most management books are a complete waste of time

Instead, do something, start something, or even join something and watch/absorb good & bad

Study leaders in many other fields as much as possible

Will AI take developers jobs?

Yes, but not. And yes, but not really. But mostly no Ok, it's been a lot of fun seeing all the AI advancements recently and obviously I'm obviously quite close to this having incubated GitHub Copilot in OCTO (Office of the CTO, now called GitHub Next) before I left with the amazing team there

I've tried and I've tried and I've tried but enough people have asked me to comment on what's likely to happen to Twitter *architecturally* so I'll do a reluctant thread

*Really* hoping to avoid any other twitter commentary but there is always some overlap 1. Is Twitter likely to go away?

No, not from technical problems

@realGeorgeHotz Hey George, likely most experienced massive infra open source CTO that’s gonna reply here (GitHub, Heroku, Canonical/Ubuntu)

1. need to distinct between twitter the service & twitter the company
2. need to talk datacenter v cloud. Impact on $ v people spend @realGeorgeHotz 3. can't forget to sort out maint, upkeep, "oh shit" scenarios & call rotations
4. also legal liabilities

What I'm saying is the real number for twitter the company is obviously much lower than 7000+ but even for twitter the service at this scale, it's a few hundred infra min

Turning off services running in production on a whim, a thread

I'm convinced that one of the biggest architectural mistakes of the past decade was going full microservice

On a spectrum of monolith to microservices, I suggest the following:

Monolith > apps > services > microservices

So, some thoughts First, these are thoughts, not rules. Anyone that has built a large distributed system knows they don't really work that way and have to adapt with it

Second, stage will be important. If you are reading this at a 5-50 person company...just stick with a monolith. Trust me

Founders, here’s a simple formula for success

Have a big vision, good judgement, & gut instinct

As you grow have 1-3 execs who are world class (not good, not great, world class). Pay them extraordinarily well

They’ll hire less folks, but they’ll be amazing. Pay them too Fill out the rest with above average folks doing necessary jobs. Pay them average. Build processes to allow average to above average folks to function without slowing down the world class

Resist over processing and over hiring at all costs

Become great at saying no

You, a moron, removing debt, reviewing code, thinking twice cutting once, get fired from twitter

Me, a genius, commits node_modules and becomes ceo of twitter

https://twitter.com/GergelyOrosz/status/1588906808395333633

Here’s a question. What’s the cross language/framework ratios? Like, how many lines of Rails equal one line of Ruby equal how many lines of Rust equal how many lines of Go equal how many lines of JavaScript? How about those there LLMs? Do we value them the equally as well?

The same people that knocked me for sleep/fitness/diet/weight lifting 15 years ago all now preach it, are the same people that told me I had to move to SF (who left themselves now), are the same people that said my leadership approach wouldn’t work

Few things I’ve learned… It never really bothered me about most of those topics mostly bc I knew many of the things I was doing were counter consensus, which I’ve always been fine with

But the fitness one weirdly did bc of the ‘dumb jock’ label and also, honestly it was just odd

A very specific mental change that helped unlock a new level for me a long time ago was stop trying to convince every last person and instead state expectations and move on

Here's what I mean... I always start out the same, giving the high level overview, tying it back to the now, giving a bunch of context & some parameters including knowns & unknowns. I also jump into 'why' a lot

This is normal work & didn't change including the usual supporting materials

Ok, let's actually talk remote/hybrid/in-office work for tech (not VCs)

There are likely some better suited to talk about this than me though the number isn't high, particularly execs. Been remote since 2010 for Canonical (started in Australia), Heroku & GitHub (Victoria) 1. Who likes remote?

There are two categories of people who like remote: the hyper competent and the hyper incompetent. It's actually very simple

The best of the best will demand freedom to live where they choose because they can get the job done no matter what

One thing I abs believe to be true but have no data to support, a group of people either avgs UP towards its singular top performer(s) or completely levels down to its bottom tier group, no real in between (particularly in the fullness of time)

The slope of the company In sports this is easy to see with a great player makes others better around them

Or some other teams where a great player either underperforms themselves but for sure the team doesn’t average up to them and underachieves to expectations

Back in '99/'00 I had just graduated, was working at IBM and decided to join a startup doing video streaming on the internet before it was cool

Insane to think about but we did live streaming of sports(mostly golf), movies(we had the entire Fox something catalog) & news I & 90% of everyone in the company were laid off the day before Thanksgiving 2000. The company had raised $75m (which is just an insane number for dotcom era, not like rando startup today raising that) and had actually bought or installed substantial portion of fiber in the US

I see two equally bad patterns play out at companies all the time

Playing Company
&
Playing Product A. Playing Product usually shows itself at BigCos that can't make product & their revenues are dominated by a one or a few nearly monopolistic offerings ie Google, Microsoft or no-longer-startups that are established but no longer "producting" ie DoorDash Dropbox Twitter etc

Some of the feedback I've gotten over the years

From high school guidance counselor - shoot for vocational school or comm college

From college department TA (2 of them!) - have you considered biz school instead? You seem good but it only gets harder from here From manager & PM when I was first time manager - Everyone really likes you and seems to want you to be their manager and we don't know why. You should clean shave everyday and consider buying a semi-expensive watch to wear when meeting with other managers

Jan 2022: welcome to planning season. It's basically like football mini-camp when everyone comes together & prepares & plans the season. Everyone is cheery & everything looks great on paper

Then the tradeoff start, someone suggests OKRs, &/or "we need to triple staff" convos After doing this for way too long, some suggestions for founders/execs

1. KISS - like seriously, really simple
2. Think big/long, plan short/small
3. Give/make people own things, no dual owners
4. Boulders, rocks, pebbles, sand

Weirdly emotional day for me today

GitLab IPOs & naturally folks talk about GitHub & our acquisition

& as the executive primarily responsible for the product turnaround & the architect of the sale, I have major pos/neg feelings on it it all For those that don't know, I joined GitHub in May 2017 and I joined a company essentially in crisis. GitHub hadn't shipped much for a few years, had a toxic reputation in the ecosystem, and largely was dysfunctional internally excluding pockets of competent leadership