Prompt engineering is the new Search Engine Optimization.

Not sure if we called it engineering ten years ago.

The distributed system can only handle one point of failure, and two have failed.

The leader is net split and doesn't care about most of the cluster and the "zookeeper" is happy with the leader.

If the zookeeper doesn't select a new leader the cluster is going to stay in this state.

This is the same problem as outsourcing to third party programmers in another country, but worse.

It really, really is at present. It’s outsourcing but without the benefit of someone getting a paycheck: all exploitation.

I am sure the governments can work it out the same way billionaires with bankers do.

Take possession of parts of their stock in a blind trust and just get dividends or take out loans against the stock and use that cash.

The plan is everybody but Trump loses.

The USA spends close to a trillion dollars annually on K-12 education, so you are only off by about 3 orders of magnitude.

I disagree with Dan due to my experience as an low level employee under Ballmer. He encouraged political infighting and backstabbing and dog eat dog internal competition, while praising and desiring tight integration between teams.

He wanted "cloud first, moblie first" - two firsts! The culture at the time was built around RAID - the internal bug datadbase and that there should be clear prioritization for everything.

The inability to decide between enterprise cloud and consumer client devices held Microsoft back.

Ballmer had customers asking for enterprise cloud in 2000 but he kept listening to people talking about lifting windows sales by 10 percent with search integrated to the desktop.

And then they chose the bloated SQL server for that and wondered why that couldn't run on normal consumer hardware in Longhorn.

The fundamental tradeoffs between something that sacrifices generalization for specialization and efficiency meant that what is good for running server rack NASDAQ didn't work for low powered laptops.

From a low level employee perspective Ballmer was the ruthless guy that wanted people to hate each other at work as they fought for survival lord of the flies style but was pikachu surprised that we could never deliver integrated experiences that worked together.

Satya's two key abilites to me were the ability to actually prioritize in a coherent way and the decision to bring the rank and file infighting down because integrated experiences are hard to build when you want your brother and sister departments to fail so yours gets more budget because thats how Ballmer worked.

So Ballmer-era Microsoft was the inspiration for Amazon's current culture, I see.

This one’s on Jack Welch - a pioneer in short term gain over long term building. You absolutely can juice a company’s performance by going dog-eat-dog, but inevitably when the smoke clears you’re left with jackals and hyenas stretched too thin.

Always worth mentioning that this culturally altered America in a way that we’ll probably never unwind.

> Always worth mentioning that this culturally altered America in a way that we’ll probably never unwind.

I think this about a lot of things, such as certain events in politics or generative AI. I'm curious how you apply this to ruthless cutthroat policies at a handful of (admittedly quite large) tech companies?

Culturally unwinds corporate america.

Go to any family business that scales a niche and you find golf course dealmaking and nepotism humming along with good ole fashion quiet cartel work.

Combined with the massive popularity of private equity in so many business areas now we're unlikely to see 100-year companies again.

I worked a long time ago in the same company that Steve Ballmer worked, both as juniors. Same culture. I think I know where he got it from.

off-topic: Love the username :)

There used to be a seperate job for this and they dumped it on startup engineers (otherwise failing businesses) and now they dump it on all engineers.

Yes it was QuakeWorld Team Fortess. Funny story of my own. I had just started at Microsoft and met Robin Walker at a lan party with a bunch of other QuakeWorld TeamFortress fans.

As soon as they hired Robin and John i knew they (valve) were going to be a giant hit. As a former windows PM, it seems that Gabe knew how valuable mod authors as first time game designers were. Valve may not have been the first to make their game engine a platform, but they were the most dedicated to it in that era.

I didn't know their contract with Sierra was so much like a music industry contract and that they didn't own their Half life IP. It makes so much more sense why they recruited the founder of counterstrike, which is an even bigger hit than TF2.

If they had been public i would have bought into valve before 2005. I missed the boat on apple, amazon, and so many others, but that one had a competitive advantage of understanding the industry that I understood.

300M in annual revenue is a key number for me.

Right now you're making 1 million dollars a year per employee or a 10x or 20x revenue to average employee cost.

The reason your software providers probably suck is that I'm not hearing how permanent or consultant employees can provide the kind of ROI that it will cost to build even basic software.

Let's say you're based in Detroit, MI - somewhere dirt cheap - 3 engineers for 1 year is 350K in salary, another ~300k in benefits and then you'll take 2 to 3 people normally earning money off their real work to be the domain experts.

Let's say 1 million in developers and 2 people providing 10x the value of their 100k salaries as well, and that's 3 million dollars of cost + lost revenue from productive employees.

If your company has a 15% profit margin on revenue and 250M in revenue - you're making 37.5 mil in profit, and you'll be giving up roughly ~10% of the company profits on this venture.

So the question is, do you think you have greater than 70% chance of getting a 3x return on their cost? Maybe 9 million dollars more worth of revenue?

If not, the project will fail just from the ROI expectations of the bean counters.

You could try to outsource stuff to somewhere with cheaper programmers but that just means the 2 million in lost revenue from productive employees becomes 20 million in lost revenue as the outsourced engineers need 10x the handholding because they aren't onsite, watching the people that do the work and interacting with them with a tight feedback loop.