jerrre

There is this: https://www.ycombinator.com/topcompanies/ Not exactly what you asked, but gives some insight

> hope to go live in 2020

CURRENT_YEAR + 1

Isn't that the whole VC model? 1 big success pays for 99 failures? I think it's more about what could happen if one blows up, than about how big the chance is that it will.

I interpreted the rules as only decompression needs to be in 8 hrs, training could be arbitrarily long? That would also lead to the theoretical strategy of finding the sequence somewhere in pi, and just storing the index and a pi generator....

Since it's about compressing one specific sequence of bytes, and compression time/cost doesn't matter - only decompression - wouldn't finding a set of [neural network inputs + weights + processing code + error correction] (I think in the direction of GANs or auto-encoders) be a way to have a high chance of finding improvements? Not 100% sure it would be in spirit of the contest, and if the cost of training would offset the reward...

I very much agree with your first part, and slow software make me physically uncomfortable sometimes. But I think you overestimate your last point. Can you name examples of software that could (partly) be recreated faster as a side project?

More programmers == more code, more code == more bad code. But that's probably not the direction you wanted to go...

Had a lot of fun reading the names and shopping lists etc they chose for all these countries, not sure why this always intrigues me...

It only now dawned on my why Digital Ocean's VMs[1] are called droplets...

[1] not 100% if that's what they are.

> Be careful taking advice from someone who has no intention of paying now.

This applies to so many situations, could be a great advice for many product creators! Saving this one

> very surprised to see it was an article by John Carmack

John Carmack works for Oculus/Facebook, so that's probably part of the reason he posts on Facebook, but you might mean you were surprised because the URL gave no indication.

The information provided for dating, could be interesting for many other purposes....

> using a Mac SE/30 to display what A Spotify client/player is playing.

It's quite a nice philosophical question who does the playing. Isn't the Spotify iOS app also just displaying what is being played by the iPhone ADC/MPEG Decoder/Network Library/Spotify Server?

Would 25 multiple choice questions be enough to detect random answering with certainty? Surely many combinations come up way more often than others when answered truthfully.

0.001% of the world is about 75k people [1], there are about 2200 dollar billionaires [2], so only(!) ~3% of these people are billionaires.

Not a real point to make, but I was wondering, and it might save someone else time.

[1] https://www.wolframalpha.com/input/?i=0.001%25+*+world+popul... [2] https://en.wikipedia.org/wiki/The_World%27s_Billionaires

I wonder how much Auto-complete could save you here.

On a more positive note:

What I really like about the Sketch story (correct me if I'm wrong), is that a small product-first focused company can make a big enough dent in an existing market. Kind of refreshing between all these marketing/sales-first VC fast growing in employee size companies with a bad fluff/action ratio.

> Omvlee was very smart to target a large segment of the graphics sector’s largest, most powerful incumbent, as doing so created major tailwind for Sketch. Omvlee came to market with a differentiated product that was far better than even Adobe’s flagship product.

Great tip: just create something better than the household product for a large segment. Why did I never think of that....

Interesting point, but not a great analogy. A reflection is a "live" copy, you can move your arm, and see your reflection move its arm. Your scent experiment is more like recognizing from photos.

https://duckduckgo.com/?q=duck&ia=web works fine for me ;)

Might be a nice step towards rebranding for a bigger audience? I use DDG as a default search engine, but the name doesn't roll off the tongue nicely at all...

Edit: It doesn't roll of my fingers either, I didn't even type out the full name...

A strange differences between OSes and Browsers, is that Browsers need/want to conform to standards.

No-one expects Windows programs to work on macOS or Linux. But with the web that's different.

> I'd like a fast mode, where you just get immediate feedback on the answer (with correct answer displayed)

I'm really annoyed by unnescesary delays, this mode with improve it tons for me. Even leaving the info could be fine, but it took ~1s delay for it to appear..

> 64M people making ~>$750/month

I read it as 64M people making enough to pay ~>$750/month in taxes

Did you read the description of the downloading process? If content discovery was the goal he could've just linked to the source

NAL but I'd say the profit is not the problem. Rehosting it (also without credit) is.

Then you loose/cut off all previous comments/views/likes

Any pointers where to start, if I haven't read/seen anything (nor even heard about) Tien, but I find it interesting to see what he has to say?

Title is slightly misleading (the shortened version of this post even more). Together with the fact that min-max is very similar to minimax[1] it made me expect an (language) AI article

The article is mostly about optimizing a specific function/data structure for fast displaying of waveforms (still interesting :]).

[1] https://en.wikipedia.org/wiki/Minimax

I don't know where I read it, but it always stuck to me:

identification is more like a username, not a password, and should be used as such.