foxtr0t

Apply this to everything not just your day job. People who know how to fix things that break, improve things like their house, fix their car, their neighbor's fence, etc tend to be more well rounded. Doing physical labor should be below nobody. Buy nice things and make lots of money, sure, but also make sure you work with your hands from time to time. This helps me stay grounded and humble. I think many people could use this advice.

FWIW I work as a SWE at a FANG, presumably a "high or upper middle class" job.

This is a patently false statement from first principles. Anticompetitive behavior to keep a monopoly is illegal, but that is not the only case. Cartels, predatory pricing, and price fixing are all illegal with or without monopolies and are all considered anticompetitive practices.

Antitrust enforcers care about anticompetitive behavior not monopolies. They certainly should care about rent seeking behavior that Apple is engaged in.

Saying something is or is not a monopoly is distracting from what actually matters. It's deflecting.

There is some notation you must know (product, sum, set, etc) but the math is very straightforward, which makes sense, game theory in its most basic form is relatively straightforward logic that one can reason through with needing complex mathematical tools or abstractions.

Yeah that is understandable, k8s is a platform with a plethora of capabilities. On the other hand, the vast majority of organizations using k8s have multiple users that have multiple work streams and multiple applications (not true by definition, but close). This lends itself to a code as configuration approach, and makes managing orchestration through a UI risky and error prone. We review code, not clicks.

Further, K8s exposes an YAML interface because it is relatively human readable and of course is machine readable. I'm sure the k8s community encourages users to build on top of this. Want to really do code as config? Write a nice DSL in your favorite scripting language to produce these YAMLs. Submit that as source. You've now automated away the vast majority of toil related to k8s YAML config, because your code can be:

1. modular by design

2. tested

3. tracked as code

etc etc

We've gone full circle:

1. Pre-kub: orchestration is manual through AWS EC2 ui

2. Kub: orchestration is automated and configuration is treated as code (YAML, JSON, proto, whatever) and kept in revision control just like source code

3. Post-kub: orchestration is manual through a ui

In all seriousness I'm sure this is useful in limited cases, but for most the right answer isn't some fancy UI, its middleware or some layer on top of kubernetes that allows you to automate away the toil-y parts of YAML, but keep the flexibility.

EDIT: line breaks.

Yeah, this problem has a pretty obvious solution that is implemented by many teams: find a unit of work that we can measure success and failure over, report an aggregate metric over a specified time interval for availability tracking and plotting:

# units successful / total # of units

Hard to not view this as dead on arrival. The House won't take up any major legislation emanating from the White House/DOJ in an election year, especially given Barr's corruption as AG.

> people... underestimate the benefits of agglomeration, underestimate the negatives of remote work

Absolutely. If you're trying to solve hard problems and don't have incredibly talented people who have 5 or 10 years of experience working on teams, know how to execute tactically and how to develop high level strategies, etc, etc, 100% remote isn't going to work well. You'll get some things done, but the next company over that can develop a new grad into a highly performing engineer/product person/[insert other position here] in a little over a year is probably going to outperform you in almost every aspect of the business.

Way way off. North America is way behind in terms of residential internet speeds even in the most dense areas like NYC and San Francisco etc. Sonic just recently started ramping up around the Bay and if you're lucky enough to have service its great. Still, probably 90% of people in the Bay and 75% of the city are stuck with 30 Mb/s up max from Comcast.

Starlink won't be high a replacement for fiber for a decade if we're being optimistic. These projects have huge ramp ups, tons of expensive launches, especially if they're successful, tons of users means slower speeds for everyone. And that all assumes success.

It is disappointing that Tesla uses much of its engineering prowess to design products that are inherently difficult for owners or small shops to modify, maintain, or repair. Taken directly from Apple's playbook.

Counterpoint: founding a startup and being financially secure isn't some inalienable right. As you point out, writing small checks will skew the population of founders, but by doing otherwise for the reasons you've brought up, you're addressing a symptom very inefficiently. It might be more effective to take an approach similar to affirmative action. I think the idea is totally valid. Pouring money into a small set of founders can have diversity implications as well, it's hard to guarantee you're distributing investment equitably if the batch size is small.

I think this definition of abstraction is lazy. Abstraction is indeed hard to define precisely, but it may be more accurate to describe it as the process of implementing interfaces that are conceptually familiar to users, often through metaphor, such as the unix "pipe".

Creating the "right" abstraction is _not_ the process of bundling up repeated code and only questioning how "abstract" it should be, it is the process of creating an interface that is familiar and conceptually easy to grasp. This is done through naming, comments, the use of metaphors, etc. From this viewpoint there can be many correct levels of abstraction, some more useful than others. We should aim to create good abstractions at all layers of code, even if finding abstraction bliss is unattainable.

That sounds like network latency, not an issue with switching audio sources.

>He’s a privileged rich white kid on the spectrum

Trump or Zuckerberg?

This is an oversimplification of already simple market dynamics. What sets the market is the supply and demand. If demand drops sharply after an event like covid-19 and supply is held constant, prices will go down except in extremely distorted markets like the ones ISPs and telcos play in.

>the latter will drive you more to banks, economic institutions, insurance, government contractors, exchanges and similar jobs.

I worked for a gov't subcontractor (for the USAF) and a bank prior to moving to the bay and working at a FAANG. The opportunities to make 500k+/year don't exist, outside a few private funds that grind you to dust. If you are super well connected and you can start a military contract firm, you can make tons of money, but that has nothing to do with things you'd learn in a CS + X double major. Think tanks and other economic agencies don't compare with a good FAANG salary. You'll start at 50k and be lucky to make 200k by the end. 200k is a starting salary for a facebook engineer in most markets. I have two friends who are engineers at insurance cos, they certainly don't work there for the money.

If you're talking about jobs that don't involve CS applications, I wouldn't have any idea, but this isn't what OP was asking about.

Not everyone picks a job based one what "fulfills" them. Some people are trying to pay down college debt, support a family, support a lifestyle etc. And if what fulfills you working with very smart and driven individuals on very difficult and ambiguous problems, I think you'll enjoy working for a large tech company.

> Worst case, you'll be good at whiteboard interviews and can get a job at a FAANG company.

Since when is this the worst case scenario for a CS grad? Last I checked, FAANGs value and compensate their engineers more than basically every other industry.

I double majored in economics and comp sci, then got a masters in cs/ml. I didn't plan on going to a FAANG until I realized a few things:

1. Staying in academia was clearly not an option for me. I still have friends who are miserable trying to get through their postdocs and I can't deal with the hobnobbery that some departments exhibit, the political infighting etc. I felt like if I got into the wrong program for a PhD I'd miss out on 5-8 years of my 20s and early 30s.

2. Did the startup thing. Found that most smaller startups are 90% talk and marketing and nothing really groundbreaking happens there. There are a few exceptions to that and if you're lucky enough to find one of those companies, good for you.

3. Other industries that don't focus on software often don't value SWEs the same way. It wasn't until a few years ago that the IT guy who maintained everyones desktops at a small financial firm and the person who wrote the internal software were the same individual. There is nothing wrong with those roles, but if you have the skills most CS grads acquire, you can specialize much more if you find a more software focused role

4. Starting my own thing isn't an option for me personally. I started some companies very early on when in doing my first degree (2009-11) and they failed, great experiences, don't need them again. If this is for you, then good, go for it.

5. FAANGs and other large software corps give you the structure to focus on work, like a laser, and then turn off and still maintain a healthy work/life balance. There are still politics and other bullshit to deal with, but your chances of finding a good team that minimize bullshit are high, in my experience. This doesn't happen at most startups where you have to play many roles. At a FAANG if you are really into a project, you're free to dive into it 24/7. If you're at a different phase in your life, say you just had your first child, you can take a step back and slow down for a bit. Nobody is going to ask "why haven't you pushed 50 diffs this week?" And obviously there is the pay.

Last I checked almost all antitrust laws are enforced through civil cases.

Isn't settlement far more likely than losing all the way through appeals? And in what way is it straightforward? Does it not require a new interpretation of antitrust law? Google isn't using monopoly power to increase prices for consumers.

>Also google PR is launching active campaign to muzzle voice by downvoting critical comments and blogs. Happened yesterday to my comments on HN and might happen to this comment as well.

Source?

>As google is faltering on revenue in cloud

Source? They don't break this out in earnings so I assume you're just reading tea leaves here?

Until you ride the T extension in 2026 and it's filled with wet garbage and excrement.

You do realize that macOS isn't the only piece of software that Apple makes?

So, specifically, Siri is definitely 3-5 years behind alexa/google assistant, I won't engage with someone who claims otherwise. iMessage is garbage unless everyone you talk to is on apple hardware. Maps was garbage until recently though I still get bad directions (thanks carplay!). Icloud works alright but there are better _and_ cheaper solutions...

So ludicrous, I know.

Why does that distinction matter? Is there language in campaign finance law that says news orgs are exempt from certain restrictions?

By that logic wouldn't Fox News be the subject of a campaign finance investigation?

All that and for the average consumer most of their software is 3-5 years behind everything else. The desktop experience is silky smooth 95% of the time and that’s the only thing keeping me there. That 5% is starting to become a problem for me, though. I got an LG 5k display that apple helped develop, it was the display they tried to sell with MacBook pros in apple stores. If I try to use it with my lid open the graphics hw overheats and the screen freezes and flickers. I have to use the screen with the lid closed. Unacceptable for a $5k setup.

That’s in addition to a series of other annoying hardware issues with that laptop. Don’t get me started about the keyboard.

Define "nobody." If you check earnings you'll be surprised to see GCP is quickly growing, even if it is a few years behind AWS. For large customers, their deployments are increasingly becoming platform agnostic, meaning they shop for price, not platform. Yegge's post, while relevant to almost every cloud customer at the time, is less relevant to large customers that can avoid lock-in today.

Last I checked there was basically zero serving story for pytorch. The trade off seems all to common, tensorflow optimizes for enormous production applications first while pytorch optimizes for developer ease first.

If you are trying to apply one of these libraries to a production system that doesn’t get a lot of throughput you probably shouldn’t be using them (try a linear model first). If you have a high throughput application you probably want tensorflow and just deal with the shittyness.

This still doesn’t make sense. There are portable systems that do the same, and have fully managed options, such as kubernetes.

In my mind the thing that makes lambda “easier” is they make a bunch of decisions for you, for better or worse. For real applications probably for the worse. If you have the knowledge to make those decisions for yourself you’re probably better off doing that.

The parallelism argument doesn’t pass muster because you can do the same thing with a cluster of free tier t2.micro machines with any good orchestration platform, not just lambda.

This argument is basically: no counterpoint to the original post, but you can do things that are also easy on any other comparable platform.

Tell me again what I don’t understand?