bensandcastle

This is a mainstream media piece. Technical writeup and papers will be forthcoming.

The scaling issue the author identifies is exactly what we've built Strong Compute to solve. Dynamically scalable super computing.

"In this environment the problem with a single big unit is that significant updates to it are likely to be quite expensive and you probably simply won't have that money all in one lump at once."

This is true if you approach it as a long term lease, or a fixed purchase, which is often a requirement given even if using on demand or spot there's still a lot of setup work to do. Our focus to crush this issue. Zero code change from a small VM to a large cluster.

On board with all of this except conferences, which I still approve sometimes, but typically see poor yields here. My preference is to pay for 1:1 expert training.

Highest end dev laptops available. Dev replicas of prod infra. Separate high spec servers for almost everything. Highly automated deep testing pipelines. Enterprise grade Internet.

We also shared how to build 2x-10x faster github actions runners: https://words.strongcompute.com/p/maximising-github-actions-...

Dev time is precious.

Other focus is Management by Context: Solid physical and tech environment. Clear goals of next feature with real testing in front of users. Dive into details where something isn't working, otherwise hands off.

The natural light and diffuse light are good tips.

Next is to get a big screen eg. 85" 4K and put it 1.5m away. That should be your main display. I don't have that all the time, but then I get some variety, 85" @ 1.5m a lot of the time, laptop some of the time, driving/walking etc. for longer range.

1.5m is the midpoint of focus for the muscles in your eyes.

I built augmented reality displays and this was the focal plane we selected for to minimize eye strain and the felt sense of vergence/accommodation conflict.

We could then throw graphics as close as ~30cm, or at infinity using vergence adjustments, even though the accommodation was at a fixed 1.5m. Graphics felt best at that distance, but they also felt ok in the range 0.5-10m, which suited nearly all productivity scenarios.

marginally relevant. space based dawn dusk LEO solar infra is the answer. vastly more power than we'll ever get on the surface of this rock and then onto Sol.

was a stated objective before the launch: "faster/harder booster catch" https://x.com/elonmusk/status/1858867695233425734

self built runner box how to including overclocked tuned high spec CPUs, and how to deploy self hosted runners.

how much analysis have you done on this? I'm working on a series around Kardashev II work - space solar, asteroid mining, dyson spheres and the real work that is happening to support this now. contact me if you'd like to discuss - @bensand on X

Ah yes! I remember working on this. It was born out of asking all the YC founders what they wanted to see - and anyone else we met as well. Formed the basis of our second kickstarter which we self hosted and was much more successful.

Yes, correct. As in their degree of openness (not much) is what's letting them down. I see how you could have read it as I meant their too open, which I definitely don't think is the case!

a bunch at Vision Pro, some at Zuck's Meta, some at Hololens, some doing other things. Meron is doing BCI, I'm doing AI infra - strongcompute.com (YC W22)

Yes, got my founders ed. pre-order in, can't wait!

Having built similar tech (Meta, YC S13), it's been a great year with Vision Pro, Orion, Spectacles and more coming out.

Currently at my co, seeing most day to day use out of XReal, and keen for Visor.

AR/XR/MR/VR app I'm most looking forward to is a 360 location share with the sharing user in AR, and the receiving user in VR, with additional virtual objects shared between. Orion would be great for the send side, with a few extra cameras and Vision Pro on the receive side.

The main thing letting down tech today is how open the platforms are for external developers.

The lack of projecting black I don't see as an issue, clip on something for VR (ok 70 degress isn't quite enough but getting fairly close), or just dim and use gradients for day to day work.

I think we're still at the most basic level in terms of understanding optical physics and ultra high resolution much smaller devices will come out, probably not too soon though.

This strategy makes sense when the odds of winning are known, and the jackpot is not shared.

The powerball jackpot is shared so you would need to factor in the likely number of people you would share with which would be complicated to figure out.

There are cases in gambling where the jackpot is not shared, for example poker machines.

The odds of winning remain fixed, yet the jackpot slowly increases, moreso when there are many machines linked to build the same prize.

At a certain level of jackpot the math works out and it's a sound strategy to pour money into the machines.

See: break even point https://en.m.wikipedia.org/wiki/Progressive_jackpot

Here's a person who's written about dedicating their life to this approach: http://www.amazon.com/Million-Dollar-Slots-Peter-Liston/dp/0...

If you were going to invest in lotto with shared jackpots I've heard anecdotes that choosing numbers 32 and higher is a better way to go because people tend to pick birthdays. I think looking at past drawing results and whether there was a winner or not you could analyse if numbers are picked disproprotionately, but I'm not sure what sample size would be needed to generate meaningful results.

All that said, I can think of many things I'd rather develop expertise in than reverse engineering the payout odds of poker / progressive jackpot machines.

There are no gates. There is a sign. http://travel.jeffersoncampervan.com/Pictures/PixoriginalSit...

Cross border travel within the EU is like moving between states in the US, easy to miss.

Is it possible to get videos of the battles?

xkcd store has some good ones,

Here's a few I've been planning to have printed:

Human Spaceflight, everything to scale http://theorysend.com/uploads/bdcffe08190411f5095eabcc8860d3... from: https://www.reddit.com/r/space/comments/2niljz/human_spacefl...

Tree of Life, over time: http://e.fastcompany.net/multisite_files/codesign/slideshow/...

Evolution of US politics: http://xkcd.com/1127/large/

I was planning to get most of these printed on a board of some kind, so they be more durable than posters, although I haven't chosen a supplier yet.

Depending on your use of the poster and the printer you go with you may need a rights release from the artist, so often it's easier to find a commercial print and get it framed, if you are willing to have all that glass.