melvinmt

Let's take this example: https://beta.openai.com/examples/default-factual-answering

95% of the token usage in this example would consist of the prompt and I would only get 1 sentence in return. So if I wanted to generate another sentence, I would have to pay 95% of the cost towards the prompt again and again... Isn't there a way to create a template for the prompt so you only pay for the generated sentences?

In the simplest case, if your prompt contains 10 tokens and you request a single 90 token completion from the davinci engine, your request will use 100 tokens and will cost $0.006.

It's weird that the prompt characters are also being counted towards your token usage. So you're penalized if you have an elaborate prompt with lots of examples (as shown in the docs)?

lacked transparency (Substack)

What do you mean exactly by this?

Yes, part-time PhD's are pretty common in Europe. They also commonly take 3-4 years instead of 6...

Oculus will also no longer be known as Oculus: https://techcrunch.com/2021/10/28/facebook-errr-meta-uncerem...

Doing More would be completing those two screens and then taking on a third screen that's just like it. Yes, this would help move the project along faster and make your manager happy, and that's great, but in the long-run, More doesn't give you much.

Always doing Extra (and not being appreciated for it) is a great recipe for Burnout.

At most private schools donations are a way to ensure your children have a better chance of getting into the same school. Legacy students at Harvard for example have about 33% chance of being admitted vs the overall acceptance rate of 4.6%.

As someone who has spent a substantial amount of time reviewing top MBA programs

As someone who has spent a substantial amount of time with friends who graduated from Stanford GSB, I can tell you, an elite MBA may not guarantee anything but it opens many, many doors for you (it's up to you to also close it once you're in the door). I'm talking about regular lunches with VC's and billionaires, getting a soft landing when your startup fails, and leading big teams at FAANG's a couple years after graduation.

That's just the perks you get in Silicon Valley. Wall Street is another level entirely. They only hire elite MBAs. A great book to read on this is Liquidated by Karen Ho, which details the many privileges elite graduates get on Wall Street.

Sounds to me this type of language can also backfire; next step of Congress could be to then subpoena someone who did work on the subject matter and did have C-level exec access...

It looks like Zuckerberg doesn't have a personal Twitter though

He does: https://twitter.com/finkd

Good, that's one of my biggest pet peeves for CarPlay.

I'm also not a fan of Siri, would love to keep it disabled as I normally do, but CarPlay doesn't even start without it :/

Elon Musk talks a bit about it in the infamous Joe Rogan podcast: https://youtu.be/ycPr5-27vSI?t=1124

This is the right link I guess: https://news.ycombinator.com/item?id=20412622

Yup, that was the issue! I just launched a new version (1.9) which fixes this.

Sorry about that, I just launched a new version (1.9) which will fix the crash.

I have identified it as an issue with Android 8.1+, will submit a fix to the Play store shortly.

The idea is to create a persistent chat so UDP and other forms of pairing are not suited to sustain the session.

I really like the laptop stickers idea, I might try it out!

It's not just the SSID, the public IP address also needs to match (which is harder to spoof), see my other comment here: https://news.ycombinator.com/item?id=19792374

The external server is needed to create a persistent chat that stays around even after every participant has left the network.

Great question. The app uses a combination of SSID and public IP address as identifiers to create unique channels so both your SSID and outgoing IP will need to match to access a chat.

Weird, do you happen to use a VPN connection?

Edit: it seems this issue only occurs with the android app, I will dig into this. Thanks for reporting!

Yeah, I've thought about creating a "Slack for every WiFi" but before I commit to adding more features, I'd love to collect more feedback. So try it out with your colleagues and let me know how it goes!

Who says the employee is not being rewarded?

It’s not just what you wear and how you look. There are many more ways for AI to identify you: the way you walk, the way you talk, the way you move your hands or tilt your head, etc. Keeping track of all these subtle giveaways of yourself would be relatively easy for AI, much harder for humans to keep track of or to hide.

That is actually very funny. Silicon Valley material.

But firstly, no one wants this infinite set of email addresses. Gmail already provides this in the better form of plus labelling.

What is the difference between dots and pluses? They both have the same flaw: to Netflix they will both be distinct addresses.

What is a remora app?

Move Fast and Break Things.

You really want to charge the price that maximises your net present value.

This goes both ways. You can definitely also increase net present value by pricing higher as long as value > price. Even better would be to create segmented/dynamic pricing that would maximize value capture up and down the demand curve.

If the market pricing is apparently so inefficient that a lot of potential value on the high end isn’t being captured (in economics speak we call this the consumer surplus), then this means there is also potential for arbitrage on the high end. Clone any service and charge 4X!

I moved to the US in 2011 with my first startup (and was lucky enough to be able to stay here). I don't think I would be able to do the same under Trump: https://www.nytimes.com/2017/12/20/us/trump-immigration-slow...