Hacker News Reader

Hacker News

Comment by gammateam | Hacker News Reader

Parent Full thread

gammateam

Thats because

> your company is liable for his wages

doesn't cross his threshold of

> In this business .... we will not go to jail

its consequentialism 101. do the consequences deter you.

A criminal charge from the US FEDERAL GOVERNMENT WITH INFINITE RESOURCES vs a civil charge from a bankrupt state against the most capitalized entities on the planet

HMMM TOUGH CHOICE

you may have been indoctrinated to respect due process of all laws as written from any united state and from the national republic itself, thats not the same game that everyone is playing.

View on HN

The game he was playing was "become a billionaire." I was a member of the protected class in the Apple-Pixar-Google etc case. When I left, I got a 50% raise. I estimate that that collusion cost O(10,000) people between $50,000 and $500,000 each (my settlement was less than $10K, and my personal loss was ~$40,000/yr * 5 years).

Given that Ben is discussing executives here, let's say: O(1,000) people what, $100,000 to $1,000,000 each?

Thats insightful

why are you using big O notation?

I often see this used informally as shorthand for "on the order of", which is another way people say "approximately".

"on the order of" is not synonymous with "approximately"

"on the order of" 1000000 means any number from 1000001 through 9999999

I would expect "approximately" to mean accurate to within the next order of magnitude down. e.g. I would expect "approximately 500000" to mean a range of 400000 to 600000; maybe 350000 to 650000 if you're stretching it.

> "on the order of" 1000000 means any number from 1000001 through 9999999

Not exactly. It can also mean anything from 500,000 to 4,999,999, or 316,228 to 3,162,277...

See https://en.m.wikipedia.org/wiki/Order_of_magnitude

But yeah, it doesn't mean approximately

It's because the notation is pronounced "order of", which is also the vocabulary used when you do Fermi estimation: you don't care about exact numbers, but you want to guess the order of magnitude / power of ten correctly. It's linguistic overloading if you want.

People, please stop downvoting this reasonable question

You shouldn't have been downvoted for that, it's a perfectly reasonable question. I had the same question myself.

Here's the funny part: there is no such thing as O(10,000) in big O notation. Big O doesn't concern itself with constant factors, it's about how performance changes with the size of the dataset. If two algorithms run in constant time but one takes twice as long as the other (maybe it's a debug build), they aren't O(1) and O(2), they are both O(1).

So it's not surprising that anyone would be confused by O(10,000). In big O, this would be an impossible value. It just wouldn't make sense.

The best solution, as always, is to use plain English. You will never go wrong by saying "about 10,000".

Formally, the Landeau notation allows for O(10,000)... But 1 = O(10,000) and 10,000 = O(1), so there is relatively little use for writing anything other than O(1).

Reply on news.ycombinator.com