rhtgrg

> there was more that happened before that bit

It was after this bit. Probably when Bezos realized the optics.

> or which was edited cynically, got an inaccurate impression

Ironic to be posting an inaccurate comment, while lamenting inaccuracy...

Here, you can see just how early this happens:

https://www.youtube.com/watch?v=NSNXBvpLb9o

In this case consent, should it be possible, would also be a questionable approach. See: the vampire problem.

    The trouble is, in this situation, how could you possibly make an informed choice? For, after all, you cannot know what it is like to be a vampire until you are one. And if you can’t know what it’s like to be a vampire without becoming one, you can’t compare the character of the lived experience of what it is like to be you, right now, a mere human, to the character of the lived experience of what it would be like to be a vampire. This means that, if you want to make this choice by considering what you want your lived experience to be like in the future, you can’t do it rationally. At least, you can’t do it by weighing the competing options concerning what it would be like and choosing on this basis. And it seems awfully suspect to rely solely on the testimony of your vampire friends to make your choice, because, after all, they aren’t human any more, so their preferences are the ones vampires have, not the ones humans have. --  L.A. Paul

The platinum rule is generally better. See proc0's reply below. The golden rule is...somewhat narcissistic.

> What is the point?

Well, I don't see you signal boosting any of the less annoying ads here...

They don't even question that it is sinking. They take that as granted -- sinking 1-2mm according to "sattelite" data. Then they spend the study demonstrating why this is attributable to the weight of the buildings they estimated.

Funnily enough, the author of this book started out as a systems admin.

> I have a memory like a sieve, forgive me if I can't recall everything.

Well, good thing I didn't say all specific examples.

Anyway, thanks! I didn't know they had copilot in 365.

> Microsoft has already started doing

Can you point to some specific examples of things announced in IO that Microsoft is already doing?

It seems pretty remarkable to me that despite how relevant and ubiquitous AI is becoming (e.g. generative search is coming to Google soon, and from their latest I/O conf it's clearly a big part of nearly all other G products), Hacker News discussions on the topic are of such low quality that it's desirable to block them out entirely. Any sufficiently advanced technology discussion is indistinguishable from tragic.

I learned the hard way that smaller vendors may have no clue what customer service is and in many cases will not stand by their own return policies -- the one I dealt with would not even let me return the item on my own dime suggesting that the item was 'used' (even after I told them they didn't have to process a refund until after inspecting the item).

In the end I just had to do a chargeback, which I imagine is hurtful to their business. They just left me no choice.

Yeah, they can be quite a time sink. :)

Glad you're enjoying them!

I highly prefer http://thecodelesscode.com to this, but this was a nice try.

By the way, when writing a koan, it would be wise to choose between a parable and a riddle. These seem to be trying to be both (short like a riddle, but more of a story like a parable) and aren't really succeeding (in my view) as a result.

The title makes it seem like LoZ changed something outside of itself, when the article is actually just talking about how LoZ has changed over time.

Also, Cloudflare is one of the easier free gateways to set up for your domain. Lots of people are using that (and Cloudflare apps) and they probably enable these protections without giving it a second thought.

I had the exact same issue for months, and I just checked and it is gone (on Firefox), which has to have happened in the last day, without me having changed a single thing on my end. I'm certain it's because you made some noise, so thank you. It was absolutely ridiculous how many websites Cloudflare was able to render unusable in Firefox. Truly a terrifying power for a company to wield.

Can you elaborate on this? What event are you referring to?

To add to this, using ChatGPT feels great in the moment, because it seems to work so well. For example, asking it for an itinerary while traveling gives you something that looks great.

However, once you actually start using it and see that the "ten minute walk" is actually an hour walk, or that a full third of the attractions it has shepherded you to are permanently closed, you realize that building that itinerary yourself from scratch using Google or TripAdvisor would take you less time than manually double checking everything ChatGPT says.

It's also quite surprising that people still think ChatGPT is capable of logic. Even for a complete layperson, all it takes is asking it to draw someone's family tree as an ASCII chart to see that text prediction only goes so far and there's not enough of a relational concept in there to comprise knowledge. There are many examples of asking it to solve famous puzzles with minor variations where it fails spectacularly.

The marketing behind ChatGPT is genius, but there is only so far you can go before the honeymoon is over and people start to really question what you brought to the table. Aside from that, ChatGPT isn't unique in what it can do, and others (including open source) are catching up fast.

That being said, I'd still use it for something like language learning (and other types of learning), where follow up queries (such as why you'd use one word instead of another, or how to rephrase something to be more polite) unlock a significant amount of value. It can also be useful to write trivial code, though I doubt a serious professional would do this (for several reasons, such as privacy and liability). Ultimately, ChatGPT fits squarely under "tool" and not under "intelligence".

It seems that as of right now, the killer app of ChatGPT is the boost in views you get by putting it in the title of your YouTube video.

That's why I said in my opinion. Which is to say, in my opinion the opportunity YC presents is more compelling for non-FAANG founders. That doesn't necessarily mean anything for what the demographics actually end up being. I'm sure at least some of those FAANG founders are favoring prestige over a better financial deal.

> very little respect for domains outside of technology.

The HN community can also be out of touch with technology. If you are an AI researcher this becomes plainly obvious the moment you visit any of the AI threads (if it isn't obvious already from the 6 to 12 month delay on the posting of cutting edge developments here). Sometimes it's hard to argue with zwj.org's assessment [0].

[0] https://cdn.jwz.org/images/2016/hn.png

Unfortunately, I know exactly why I'm being downvoted, and that's not it. It should be rather obvious that "screen touches" are not the category of comms I was talking about.

Perhaps you meant to say "this is why I downvoted you", in which case, I appreciate you taking the time to state your piece.

Textual inversion involves providing self-created images, which should confer copyright in the same way AI images of DC's superman are considered to fall under the copyright of DC. In other words, commissioning fanart still allows the original owner of the IP to exert copyright -- shouldn't that be the case here?

Yes, removing CarPlay is not a great move, and is rightfully being bashed in this thread -- but these kinds of OEM integrations are far beyond a smartphone subsitute, especially given what's coming down the pipe for auto. As you allude to in your comment already, CarPlay is a "projection", a one-way deal -- whereas most of these integrations make heavy use of data from car sensors, which enables a lot of high fidelity features (including self-driving and advanced HUD indicators) that simply aren't possible with a smartphone, no matter how "independently upgraded" it is.

In the future, perhaps phones can integrate better with cars and enable that kind of experience, but we aren't there today.

You might be able to find some of those, but you probably don't want to work with incubators or accelerators that consider anybody -- let alone such a large absolute number of people as "ex-FAANG" -- a "shoo-in". Just think about the second order consequences of that.

And yes, when it comes to startups (i.e. business) and funding, word of mouth is very powerful and not to be underestimated.

It depends on what your goals are and what your business needs. I would recommend reaching out to your local angel investors —- even if you don’t plan to fundraise with them, they usually have an avenue for you to get in touch directly with investors —- and they can interview you about your startup and dispense free advice that is much more informed than what I can give you here.

Assuming you are at a very nascent stage with your startup, there are still all sorts of options out there, including but not limited to —- Kickstarter, loans, raising from friends and family, running on a shoestring budget offshore, raising convertible notes from local investors (you probably have some of these people in your extended network already), selling preorders, self-funding (e.g. via collateral loans), grants, incubators (other than YC), etc.

There are a lot of variables, and several options to choose from depending on your particulars.

YC will absolutely coddle you if you manage to become one of their “darling” startups, just like any other incubator or VC firm.

That’s the type of thing that you have to earn, and if you earn it you don’t need it.

YC used to be for people who investors wouldn’t normally take a chance on…and imo, it still is. If you have a FAANG attached to your name with multiple years of experience, you can probably find better deals elsewhere. YC has done a relatively good job of scaling themselves, but it has never been a “one size fits all” funding solution, and if you are a founder you’d be wise to consider your options carefully.

That being said, removing the relocation requirement was a big mistake — it was partially a forced move, but it’s not at all surprising that you get a significantly degraded founder experience if you are not in SF.

> if you used ai to design a new superhero and then added pink shoes, yellow hair, and a beard

Wouldn't that depend heavily on the prompt used (among other factors such as image to image and ControlNet)? You could be specifying lots of detail about the design in your prompt, and the AI could only be generating concept artwork with little variation from what you already provided.

If I'm already providing the pose, the face, and the outfit for a character (say via ControlNet and Textual Inversion), generating <my_character> should be no different from generating <superman>, that is to say, the copyright already exists thanks to my work and the AI is just a tool, the output of which should have no bearing on who owns that copyright (DC is going to be perfectly able to challenge my commercial use of AI generated superman artwork).

Looks like GPT is the new 'blockchain', except it's the unwashed masses calling for it rather than VCs (I suppose the latter have learned some hard lessons recently).

Who am I kidding, VCs will be VCs.

Perhaps, though I now recall that one of the options was free. I tried that one too, but I abandoned it first in favor of the others since the checkout and "payment" experience failed in strange and unpredictable ways with that one (the others at least showed me a 'sorry this is broken' style of message).

Recently I paid to book better seats on an Air France flight and since the website was a hot mess (specifically the payment portal) I tried various VPNs to try and get through. To my surprise, prices were significantly different (about ~20% variation) depending on country, and it was most expensive in...France.

This was on a flight that had already been booked from the US long ago.