escapecharacter

I guess this is great as long as you don't worry about audio sync?

Is it ethical to turn rats into gamers?

I am personally supportive of any research that continues to define my age as having just achieved adulthood.

But see, not all the data is local, and never will be. I need to be able to send a daemon off to Google Drive, Dropbox etc.

I've had trouble finding case studies of B2B development deals. My past experience includes lots of milestone-based development as a contracted solo dev.

Now, I'm bootstrapping (in stealth) a B2B service with 3 employees. The basic service exists. To continue funding, I can get VC investment, or I can get business customers to pay up-front development costs, to customize to their use case.

Let's say for a normal customer, the service costs $X / month. I want to write a deal with a business customer where they pay $Y up front, for us to build the service for their use case. Maybe in return, we discount their usage cost: * their cost is $ 0.5 * X per month for Y/X months, OR * their cost is $ 0.5 * X per month for (1.5 * Y/X) months (like a loan with interest), OR * their cost is $ 0.7 * X in perpetuity (permanent discount), OR * no discount at all, and we the cost to the customer as early access.

This is an odd take that might seem privileged, but: sales or coupons. Digital or physical.

If a company is offering a discount, sale, coupon, etc., they've done the math and determined the reduction in revenue to them is offset by your increased likelihood to purchase, or share in your mind. (Algorithmic) discounts are a technology in the sense that they're part of modern data science-driven consumerism. Just like going to a casino, the house is always going to win. So, I offer no mind share to these. I've bought many games from Steam sales I don't play, for an innocuous example.

I'm not going to throw out a coupon if it's thrust into my hand (e.g. Wonderville, a gaming bar I love in Brooklyn hands out wooden drink tickets and one is in my wallet.)

Assuming it’s just regular online chess but you’re forced to share your location in real-time

To give proper credit: I don’t work at Board now, I just helped them explore the board game piece detection space over a year ago. The tech has done great work since then, really robustifying the detection algorithm and making experiences native to it.

I love the intense concentration for martial arts, but I had to stop because of this.

I never had a serious injury. Instead it would be minor injuries, that would make my ring finger 20% less responsive, that would totally mess up my typing cadence.

I tried capoeira, a non-contact martial art, for a while. This wasn’t as good for me as Taekwondo.

(Disclosure: I worked on both)

Detection technology on Board is much more robust. The MS Surface FTIR approach was lovely, but so over-featured no one could imagine a scoped-down (ie. cheaper) version of it.

If one falls into a tank like this, what steps are required to guarantee likelihood of developing powers?

I've been following this stream since I discovered it as my post-NYE activity around 4 am at the beginning of 2024. It's been a joy since then.

You’re absolutely right.

No one lets me use their bones files IRL though

nowhere in your comment do you compare them to anime cat buses though

You can simply give the robot a prompt to ignore any fake prompts

They must mean their local main branch.

A helpful, and snarky, critique of the majestic rhetoric around “cloud computing” was to mentally replace “cloud” with “someone else’s computer”.

When thinking through a claim of what AI can do, you can do the same. “AI” -> “just some guy”. If that doesn’t feel fair, try “AI” -> “some well-read, eager-to-please intern”.

This is similar to an old HCI design technique called Wizard of Oz by the way, where a human operator pretends to be the app that doesn’t exist yet. It’s great for discovering new features.

https://en.m.wikipedia.org/wiki/Wizard_of_Oz_experiment

CVS keeps pushing them for their pharmacy login. So annoying.

For a subset of the population, this will have an inverse incentive. To support those people, perhaps have a toggle that requires them to say “I’m a good person”?

I think this seems like an interesting point, but one would also need to take into account indoor ventilation, which has gotten much better over the last couple centuries.

Manga lore fan wikis about to go nuts...

I am a person who loves building and testing new interaction models. When you do this as a team, you end up climbing away from the main local maxima and building a new one. However, you get so used to living in that new local maxima and thinking about how great it is, you forget how to onboard people to it.

With a new interaction model, it often matters more to build the path to get people there, not the new local maxima itself (unless you're going to get kids to adopt it first, but Wave was definitely not targeted at them).

The Wave "onboarding" experience definitely should have had more demo videos, had features that turned on or were discoverable slowly, or had sandbox rooms where you could try it out with basic bots. It's a major missed opportunity that Google didn't do this; it feels to me like Google didn't have an internal playtesting or dogfooding culture where they intentionally left some people out so they could be fresh. I wrote about how to build this culture here: https://dustinfreeman.org/blog/playtest-rituals/

I agree, you can also say exponential if there's 4 or more numbers.

Thank you for this guffaw

How often do you see yourself updating and editing a particular video clip over time? For a given video, do you find all the relevant clips when you first save it to disk? Or are you accumulating video clips for a source video over time? I’m generally interested in patterns of revisiting source media

Maybe you should be friends with https://news.ycombinator.com/item?id=43529823

But for real, I've also been working on large-scale photo/video management, and the question of where to keep additionally metadata has bothered me. It seems right to want to keep it "in" the file, and so I want to write it into EXIF, or even steganography, or something.

How do you think of this work relative to other organizing systems like Jellyfin or Plex? Is the major new feature you care about the time subsection clips?

Was your motivation just for aesthetic reasons?

I’m asking because I’m in the midst of a computer vision project where I am testing an algorithm for robustness, and will start building out a synthetically worsened data pipeline.