hejja

you are right. demonitized means unsuitable for advertisers, which means no ads.

true, yet the same can be said of:

- any new pharmaceutical - diets (keto, carnivore, etc) - 5g

etc

point being: modifying behavior due to decade time-horizon unknowns is not really pragmatic.

...about anything. not just covid.

9 figure deal, 10 figure implications:

"Spotify shares dropped 8.8% Wednesday (Sep. 2) morning, shaving as much as $4.81 billion of its value, following a report Joe Rogan's back catalog debuted on the platform Tuesday without episodes by right-wing personalities"

Case and point: employees at Spotify wanting to censor Joe Rogan.

Whether or not you agree, you can objectively see how this would jeopardize a presumably 10-figure deal for Spotify.

in a perfect world it would. but it doesn't.

Libertarian.

pair programming, for sure

if done right, whiteboarding should be almost the same as pair programming -- an iterative dialogue

100%.

if properly trained, interviewers ask questions that build without throwing the kitchen sink at you.

whiteboard interviews (when done holistically) assess for signal, not binary correctness

- do you have a solid framework to build a solution?

- are you considering multiple approaches / data structures / complexities

- referencing similar problems, vocabulary, situations to illustrate breadth of knowledge

sure, the stress component isn't ideal, but there are multiple relevant capabilities being assessed in a very short amount of time...

in other words it's a much better test of tacit knowledge than alternatives

the truth is probably more nuanced

objectively considering information is further than most people ever get. most people are just out here trying to win arguments

so let's assume you're open minded

you could weave components of this new info into what you already know, and discard the rest

but what do you keep?

does the most convincing empiricism trump everything? are personal values involved?

that's up to you

my $0.02: it DOES seem similar to bitcoin in that

- the technology is intrinsically interesting - there is a ton of hype - but it seems ultimately commercial viability of projects is questionable

Why do I say this?

Well, the "solution" here is instantaneous text generation.

even if it is 99% believable, that 1% error is probably a dealbreaker most use cases

example a: generating code

sure you can generate some simple react components, but snippets already do that.

for anything more complex / production ready, you still need to fine tune it manually

That said, I hope I'm wrong and some cool AND useful applications come out of this

In fact my initial reaction was pure hype but now I'm going the other way

Hey Jash, Curious: with your email list, do you send value-giving emails at regular intervals -- ergo a weekly digest

Or is it more just for paid offers whenever you have one? Or both?

I feel like content strategies are a double edged sword. Sure you could have YouTube, Twitter, email, and a blog but then what time do you have left to work on product?

Anyway thanks for the info

May be a symptom of people sold on "blue ocean strategy" / "zero to one" narrative

"just start a monopoly" the literature says... implying every venture that's not would surely be a waste of time

but even (and especially) if you're the first mover, sharks will come

will you use their presence as a convenient excuse? or are you up for the fight?

many of our positions are emotionally-driven, and then backwards-rationalized

reality: this is too complex for me rationalization: I didn't think there was market fit

reality: It's hard to learn technology x rationalization: OOP / framework / language sucks

love the article, upvoted and saved. but the background was absolutely killing my eyes.

to me, it feels overly optimistic in 2 ways.

1. it makes the assumption that failure is always positive

self improvement dogma: "fail faster, you learn from each failure"

peter thiel: "each failure is a tragedy, it is multivariate and therefore often too complex to truly learn from"

2. dunning kruger syndrome and learning something "the wrong way" is possible in more than one field.

Overall it was a nice read though

not op, but i think it's good enough to warrant reposts so new people can find it.

I interpreted the "reader app" definition in their policy as:

either - app value is content driven (netflix) OR - data-driven (bloomberg) OR - or physical (airbnb, amazon)

in cases where you could say, the client is the "main driver of value" they are pretty consistent.

take a game, for example -- or hey's proprietary filtering system.

that said, I don't agree with 30%, jsut my understanding

this is a salient point, fixed rate would make much more sense

what a difficult comment to read

imo this is less clear-cut than people are making it out to be.

I'm not saying I'm with Apple. In fact I hate platform fees.

But if I have a game and say, "my game is subscription based" to circumvent in-app purchases, how is that different?

In other words, where do you draw the line

for me it's hard to separate it from reading SICP, which as a self taught developer, I feel brought me to a different level of understanding concepts like immutability, the shortcomings of oop, streams, eval apply and so on.

this article feels lacking in nuance, strong claims that aren't universally true lead to bad reasoning.

to any "young developers" read I would say not to blindly adopt these opinions without understanding them.

you could also use gatsby to generate static sites in that case -- imo react offers enough developer quality of life features where it's pretty hard to justify not using it

100%

I'd think anything in the LISP family would qualify

what / where is "the edge", in your mind?

makes sense. thanks.

models I have used seem to have their usefulness greatly outweighed by performance demands.

scaling and economics are another question entirely.

Perhaps we were spoiled with democratized web tech and it's wishful thinking to want everything to be that.

so what you're saying is...

you either swallow the CRUD app red pill, or you live long enough to become the "AI a la carte" manager

genuine question here: why is using deep learning to do something useful impossible?

I can think of a bunch of potential use cases for gpt 3 alone.

or do you mean its impossible to build useful models from scratch because all the "easy" problems are solved?

this also seems like a limited mind set.

context: I'm a ML noob