mbforbes

I'm interested by the first pull quote:

> "...analyzing customer data, which has become too large and complex for Excel. It helps me sift through massive datasets..."

The emphasis here is on the size of the data, but I think that's a red herring. Excel can certainly handle way more data than you're going to want to repeatedly pipe through an LLM. I'm betting it's the diversity of---or even lack of---schemas. Big piles of tables and unstructured data you'd normally have to normalize and join.

I spent years doing data analysis slowly with Python (pandas), and it took me a long time to accept that getting better at Excel was just more efficient. Hard to accept that code isn't always a task's best interface. Automating some Excel work sounds even better.

Super helpful input, thank you for writing. I think you're absolutely right that I should work on a video first, at the very least.

I'd never heard of this, thank you for sharing such an amazing reference!

Thank you! I hope you find time to work on it as well. I'd love to try out your demo and happy to give feedback if you'd like - contact into on my website linked in bio. I also think you're totally right on the video demo needed. On it soon!

From watching playtests, I think it can induce some real social stress. But I think the controlled environment and ability to take breaks helps mitigate it. It may actually be decent practice. Please let me know how it goes if you try it!

My NLP background left me on a hilariously level playing field with everyone else to basically become a prompt engineer for a while ;)

Thank you for the kind words, Alex! Stories like these make me wish I could watch more people playing it

Replying to you and parent - thanks both for these comments, and it's a really great idea. I could see a product today (maybe even this) being helpful for practice. But I think to make genuine claims about assessment and improvement, off-the-shelf AI isn't there yet, because there's such an incredible amount of information in the way something is said. Basically the entire study of prosody. Then, there's all the body language---which, if you believe anecdotes, conveys a high % of the received information. Long term, it's a great goal.

I think you're right, there's actually a genuinely useful angle here (at least for the situations that aren't too absurd). I was surprised how much actual social stress talking to a voiced NPC can induce!

This is an extremely fair point.

So lovely to hear, thank you for the kind words!

I'm sorry about that. I was probably overly cautious about my API quotas and server melting. Expect an invite by tomorrow!

Super great observation about the objectives! It started out with lies (er, let’s say “excuses”) but it’s broadened into more conversational goals. I need to highlight this more.

Specifically, re: flattening the strategy, spot on and I’m excited to introduce more long term repercussions for lying. There are some long-range dialog repercussions now, but not many. Having to craft a consistent narrative across the whole game (“Overboard” comes to mind if you’ve played that) would be wildly interesting.

Loud and clear re: demo and highlighting voice!

Thank you Greg, your feedback has been immensely helpful!

Hey I appreciate you pointing this out, thanks for taking the time and writing this feedback!

I love that this and AI Dungeon (perhaps the OG) exist. I think my key insight in the formula for this game is that the gameplay really sings when you introduce more constraints - it feels more challenging and less cheeseable, which makes it more rewarding.

Thank you, great feedback. I need to do this soon!

I worked on building exactly this earlier this year. I was hanging out in Taiwan for a few months and thought, surely the Babel Fish should exist by now.

I did several experiments recording from all the microphones I could on my iPhone and AirPods while out in the wild. My conclusion: it's impossible right now for that hardware given the microphones we have and what they pick up.

So much of what's spoken is at a combination of (a) high distance (b) low volume (c) background obscuration. Something that was clear as day to my ears would barely register on the mics. While context is of course an issue, the raw audio didn't have enough to even translate.

The one caveat is that there might be low-level (i.e., Apple-only) access to headphone microphones that capture the environment to do noise cancellation. I'm not sure though---I couldn't find them on any API.

For cases where you do have clear audio, existing apps (e.g., Google Translate) are so close to achieving this, but don't let you specify audio outputs with enough fine grained control. By default, it will start screaming out of your phone what you were attempting to silently translate.

Thank you, again, for such a warm and thoughtful reply!

I'm working on bootstrapping a software business right now, and I spend a decent chunk of my stewing minutes stewing on what pricing and free trials should look like, so I especially appreciate the candid remarks on those three factors. The conversions fact surprised me, though makes sense once you explain it.

The tricky thing I'm working to balance---for both my own finances and software pricing---is subscription fatigue and subscription inflation.

It's fascinating to now look at things from a "business" angle, because monthly costs I hardly blink at for my business ($10 or $20 / month) feel like hard blockers for my personal life.

If I try to tease apart why this is, I think there's a couple categorical differences going on, plus one quantitative. The categorical ones are (1) my personal life is in the category of "cost only," whereas my business is "revenue minus cost". (2) something I'm not used to paying for, I'm reluctant to pay for. And then the quantitative (ish) one is about the value of $10 / month when it feels like every service has that as their target price. There's a slippery slope feeling, where if I'm willing to spend $10 / month on this slice of my life, there are twenty-eight other places that will happily take that as well. I wonder whether $2 or $3 / month subscriptions will emerge, or whether that just doesn't make sense economically.

Anyway, thank you for reading along as I poorly rediscover the difference between consumer and B2B apps from first principles ;-)

Thank you, that's a very generous reply.

Have you considered bumping up the free tier to absorb any of this market? I've never thought of what my $/subscriber value might be as someone running a mailing list, but that's probably because I'm just small fry.

I like this author's idea of curating bookmark lists, but I think they are most effective when two criteria are followed: (1) keep the list small, (2) write small notes about each entry.

For example, the bookmarks list the author links to (https://www.marginalia.nu/links/bookmarks/) has 48 URLs annotated only by category. That's too many for my tiny brain to handle and I move on.

A webring like Hundred Rabbits' (https://webring.xxiivv.com/) has 203 entries. For me, this is in the same category as 48. (It also reminds me of those "Awesome X" lists on GitHub that end up flooded with hundreds of links.)

To attempt an example of what I mean, here's the bookmark list I publish on my website:

- Bret Victor (http://worrydream.com/) • interaction and abstraction

- Craig Mod (https://craigmod.com/) • long walks, atmospheric photos

- Hundred Rabbits (https://100r.co/site/home.html) • physical and digital minimalism from first principles, extensively documented

- Mu-An Chiou (http://muan.co/) • rhythm, space, movement, color

- Steven Wittens (http://acko.net/) • 3D sorcery

(edit: formatting)

Really nice. I tried removing someone behind a wire grate, and it correctly identified them, replaced the wire grate and filled in the grate holes with decent background imitation.

Did anyone else read the generations? They’re… really poor quality, right? Im not sure whether I’m misunderstanding though.

I use Sketch to export SVGs, and animate them with anime.js. Here's an example of some illustrations where I did this [1]. It's not without issues though; using external (i.e., not inlined) SVGs is surprisingly tricky, and you quickly run into edge cases with groups and transforms---some notes here [2].

I like this approach because it decouples drawing and animating, and with SVG it is resolution and bandwidth-friendly. I was already used to making figures for research papers with Sketch, so I just suck with that, and added anime.js on top. However, for more complex motion like you might be hinting with your post, I could imagine a GUI tool with a proper timeline and tweening to be more ergonomic.

[1] https://maxwellforbes.com/posts/every-phd-is-different/ [2] https://maxwellforbes.com/garage/animating-svgs-from-sketch-...

I was shocked and delighted to find that, after pairing a Bluetooth keyboard to my iPhone, the default emacs-lite Cocoa keybindings work there too. C-a to beginning of line, C-e to end, etc.

I wrote a tutorial for fixing this issue, in case it might help: https://maxwellforbes.com/posts/fixing-mobile-page-layouts

Better link: https://www.seattletimes.com/seattle-news/politics/inslee-or...

Thank you so much for your kind words! As a frequent reader but rare poster, I find HN more daunting than I probably ought to.

Those unit conversion libraries look awesome! Way better than my pile of hacks. I only added the units I happened to use in recipes (surprisingly few). The most hacky thing is probably all the cooking-specific ways of saying "thing" (clove, pack, fillet, sprig, scoop, etc.).

The most interesting problem I ran into with unit conversions was actually around the user experience: when did I want units converted at all? Right now, the app simply always does it before rendering anything. This means converting to the smallest unit of that type, and then picking the largest reasonable unit to display. This can lead to annoying instances where you put an ingredient in one unit, but another is displayed. I also went a little overboard on the fractions, so you see things like "2/5 tsp" which is just absurd.

Totally, takes some getting used to with a trackpad. Move + look at the same time a lot for the brain to do. Just pushed experimental gamepad support if you have a controller lying around.

Yes!