Timing Technology | Hacker News Reader

Timing Technology

gwern.net

63 pointsordiblah16 comments

> Someone could have built the Rift in mid-to-late 2007 for a few thousand dollars, and they could have built it in mid-2008 for about $500.

That's not true. The thing that enabled the Rift was the rapid improvement in cell phone screens in the early 2010s. I know because in 2011 I was trying to design a VR headset that was basically the Rift. Readily available screens at that stage were nearly, but not quite, good enough (most were around 720x480 iirc). I spent ages playing around with lens and mirror setups and fixed foveated rendering etc. and came to the conclusion that it wasn't doable (yet).

My failure of imagination was not realising that, while screens in 2011 were inadequate, they were improving in resolution and response time rapidly and were about to cross the threshold from useless to useful.

A year later 1080p screens were the norm and the Rift was a reality, but in 2007 it was a pipe dream.

Also what always separated the wheat from the chaff in VR, was latency. (That kind of timing.) Naive approaches did not take low latency seriously, causing users to throw up needlessly.

Yeah, there's an excellent series of blog posts by Abrash going through all of the different triggers they found for VR sickness and how they tackled them.

I believe that quote was in reference to the Rift DK1, which had a 1280x800 full persistence, slow response panel but pretty poor panel utilization. The usable resolution per eye was something below 600x600. It was plausible to build a DK1-like headset in 2008. That is, something just good enough to show that compelling gaming content could be created for VR and that a device could be delivered at a consumer price point.

Getting to an actual comfortable consumer-quality product would still have required solving for low-persistence, fast response, higher PPI, and positional tracking. We couldn’t have delivered that in 2008, but if we had DK1 in 2008, we could have pulled it in a few years from the consumer Rift launch in 2016.

I love everything written by Gwern. One day I want to meet him if he lets me.

But I am not sure I agree with this one. At least in my personal experience, I had several good guesses at where tech was going since about the late 90's, but couldn't capitalize much on it until very recently (e.g. by investing with my micro-VC fund).

Perhaps it's my experience that's skewed, or perhaps I think too highly of my past intuitions.

Even though you didn't have the money, did you identify specific companies that were worth investing in? Did those companies succeed, or was it others with similar ideas?

Yes, absolutely. Public and private companies. And also some specific technologies.

Edit: let me give you a simple example. In late 2005 I proposed an investment to a wealthy relative: buy $100,000 of GOOG and $100,000 of BIDU. Any loss would be covered by me as debt, to be repaid over the next 10 years, with a stop-loss at -50% ($100,000). Any gain would be split half and half. Hold it for 10 years, then sell. Unfortunately my relative didn't accept. Baidu did 30x in 10 years. Google did ~6x.

Did you write down all your bets? Not just the winners, but also the losers?

This is a bias that a lot of investors often make ...

I think a lot of people are surprised when they start using a prediction market or prediction registry and track all their predictions, not just the ones they remember decades later. (I thought Google was a good investment in 2005 too. I thought a lot of things back then, but I don't remember what else I thought.)

That's true, and it's partially my bias as well, of course.

You could almost certainly have done better by borrowing money and buying the right kind of options. If you're willing to give up half the upside for a stop-loss, you would collect quite a premium and should more than offset even usurious interest rates.

You want exposure to 50% of the upside: this means you want exposure to 50k worth of each stock. You don't want to lose more than 100k total, and are ok being wiped out if both drop 50%.

Seems to me like you'd want to buy 25k worth of call options at a 50% strike price. The options would cost 25k and would reflect shares worth 50k. You get exposure to 50k of shares on the upside, but your loss is capped at 50k total, which is twice as good as your deal. All you need to do is pay interest.

Oh, I thought it would be about timing instruments, crystal oscillators and such, maybe time transfer protocols...

"Hindsight is 20/20."

Said someone better than me.

Elliot waves and quantum entanglement.

Reply on news.ycombinator.com