fijiaarone

“Close” meant secret in the 1940s. A “close secret” was next to “top secret” classification.

See also the use of the word “close” in literature, eg The Lord of the Rings “Gandalf is closer that ever”.

To keep it close or to hold it close meant to keep it secret.

At what point before the US coup on 2014 was Ukraine not controlled by Russia?

There is an answer, but probably only an actual Ukrainian “nationalist” can tell you. And it was only a couple of years.

It’s like the lesson told in the paragraph above “this is what my team did” and imply “and I am their manager”. This is what I learned and imply “and I can help you learn too”

Does “consumer co-op” mean that you have to buy a plastic card in order to get the corporation money, like Costco?

VCs don’t have ROI. They’re gambling with someone else’s money and taking big fees. ROI implies that there is a return — but they’re already making more than they deserve just for the fee, and they are not investing anything themselves.

100 units isn’t 100% of available cash and equity. It’s the minimum number of units that founder (investors) think they can get employees to work for. Most of the time in well funded startups they can afford to pay 200 units of cash and 500 units of equity. But the founders (investors) would rather take 50000 units of cash and 50 million units of equity for themselves and only pay employees 90 units of cash and offload their risk onto the employees so that they have greater reward and lesser risk.

I definitely thought like the first commenter at first but the realization that you admitted what you did wrong and accepted responsibility for it really won my over by the end of your post.

NCR maintains legacy POS systems with a mostly outsourced labor force.

Splunk maintains a logging aggregator with a mostly San Francisco staff.

A manager’s job at NCR is mostly about cracking the whip to try to get your team to solve boring enterprise client technical problems, but they usually don’t have enough context to understand the problem.

A manager’s job at Splunk is to motivate a team that used to work at a hip, fast growing startup who just got their payout from an Cisco acquisition and feel like the product is getting sidelined.

If the younger generation is too pandered and can’t take criticism or honest feedback, thats the fault of the older generation.

The protocol for interacting with code is files and text.

You need to define an interface for Ai to click buttons? Or to create keyboard macros that simulate clicking buttons? We are all doomed!

Google and Amazon were doing things at roughly the same scale* 20 years ago on slower hardware and less of it.

* They might be serving twice as much (but definitely not ten times as much) as they were in 2005 but mostly that scales horizontally very easily.

Conway wasn’t proscribing a goal, he was describing a problem.

The problem isn’t refactoring, its that it was failed, abandoned, or incomplete.

And that’s usually because the person or small group that began the refactor weren’t given the time and resources to do the refactor, and uninterested or unknowledgable people hijacked and over complicated the process, and others blocked it from happening, so what would have taken a few weeks for the initial team to have completed the refactor successfully, with a little help and cooperation from others, and had they not been pulled in 10 different ways to fight other fires — instead after months and months and expending tons of time and money on people mucking it up instead of fixing it, the refactor got abandoned, a million dollars was wasted, and the system as a whole was worse than it was before.

Mostly I agree with what the author is saying. But there is a clear distinction between the simplest “system” and do the simplest thing.

The simplest thing to do is almost always the easiest, but knowing what is easiest thing to do is a lot trickier —- see Javascript frameworks.

Mostly I agree with what the author is saying. But there is a clear distinction between the simplest “system” and do the simplest thing.

The simplest thing to do is almost always the easiest, but knowing what is easiest thing to do is a lot trickier —- see Javascript frameworks.

But I think I disagree with the author’s second axiom:

“2. Simple systems are less internally-connected.”

Creating interfaces is more complex than not. Even if it leads to a cleaner design because of interface boundaries. At the least, creating those boundaries adds complexity, and I don’t mean “more effort”. I mean it in the sense that creating functions is more complex than calling “goto”. And it took decades to invent the mechanism needed to call functions —- which is probably the next most simple thing.

However, using call stacks and named pointers and memory separation (functions) leads to vastly improved simplicity of the system as the system as a while grows in complexity.

So in fact, using your own in-memory rate limiter may be a simpler implementation than using Redis, but it it also violates the second principle (using clear interfaces leads to simpler systems.)

And it turns the author’s first premise — Gunicorn is simpler than Puma. Because Puma does the equivalent of building their own rate limiter — managing its own memory and using threads instead of processes.

And Gunicorn does the equivalent of using Redis — externalizing the complexity.

What Gunicorn did was simpler to implement (because it relies on an existing isolated architecture - Unix processes and files) but means it has a greater complexity (if you take into account that it needs that whole system to work.

However that system is a brilliant set of reductions in complexity itself, but it runs up against limitations and performance at some point.

Puma takes on itself more complexity to make administering the server less complex and more performant under load. Also, because it is, in a sense, reinventing the wheel, it lacks the distillation of simplicity that is Unix.

So, less internally connected systems are easier to expand and maintain and interface boundaries lead to less complex systems as a whole, but are not, in themselves less complex.

Limitations in the system that cause performed problems (like Unix processes and function calls) are not necessarily “more simple than can possibly work” —- but the implementations of those abstractions are not perfect and could be improved.

Sometimes it’s not clear where to push the complexity, and sometimes it’s not clear what the right abstraction level is; but mostly it’s about making due with the existing architecture you have, and not having the time or resources to fix it. Until the complexity at your level reaches a point that it’s worth adding complexity at a higher level due to being unable to add the right amount of complexity at a lower level.

There's nothing stopping humans from visiting websites. But humans don't want to do that. They want AI slop, they want Google SEO spam.

They don't want to hear from real people. They want the YouTube/Tiktok/Insta-algorithm to auto-entertain them.

And there hasn't been a real recipe published on the internet in centuries.

DynamoDB is actually an anti-product, and MySQL/PostgreSQL costs $10/month for a faster, more scalable version.

Somehow I knew what RFC2549 was without knowing what it was.

Exactly, you're selling inconvenience. Management usually has to work hard all day long to annoy the employees under them.

The goal in sales is to flatter people (with attention) whose job is to spend other people's money.

You're clearly not the type of person who makes buying decisions at a large company.

Ah...but have you considered that there are also CEOs?

That logic probably applies to NVIDIA too.

It takes me an hour to write 100 lines of slop -- code or text. AI can do it in 1 or 2 minutes. That might have something to do power usage.

CEOs don't have any data.

Well, in Intel's case, they were already in a nose dive, so it's not really macroeconomic.

But for most tech companies, yeah -- when money isn't free (or cheaper) there is no business model.

The job of the high paid people in finance at prestigious firms is to look nice in an expensive suit. Know many people in tech with those qualifications?

Yeah, react developers don’t even realize that there is execution time as well as download time for an app.

You’ll add at least 100kb of text for class names styling each HTML element using tailwind.

After IBM & AT&T