jmiskovic
jmiskovic on Hacker News
Nintendo remains a treasured memory of many 90s childhoods. They took great risks to innovate both on hardware and software side for ~2 decades, I'm not defending what they did since.
I'm also not arguing for intellectual property or copyright enforcement. I'm asking what comes next, I'm genuinely curious here.
"Wouldn’t it be nice if Nintendo were more like Microsoft?"
That's... completely unhinged. But, putting aside the mention of the biggest rent seeker in tech world as a role model. What's everyone's take on the gamers cheering against the Nintendo? Who's going to create this generation's own franchises like Nintendo did? What will be their incentives to invest time and effort in any new IP?
I saw
https://github.com/lokutor-ai/oido for this purpose the other day. It's easy to compile and run it on desktop for testing but I don't know how it compares to Whistle.
"seems to be the best" weasel wording from author with vested interest.
Inference vs training. It's training that takes enormous amount of power, both electrical and the compute. I suspect many many models get simply thrown away because they end up being too low on benchmarks by the time they are done. And some are not released to the public. So we learn only about tiny percentage of trained models and their environmental impact.
When you pile them up they burn my eyes. I seem to have skimmed through them.
Be careful with this, you'll get burned sooner or later. Short sentences and breaking up the rhythm are often taught as good writing style. The people will also pick up some LLMisms, especially the non-native speakers. I enjoyed this prose, I felt like author had a very personal perspective to share. My AI BS detector didn't spike and I think your post is the over-confident one.
Did you know you can get certification from Anthropic? And it actually costs real money.
More info? Couldn't find anything about it and now the Quest communities are swamped with these glasses.
Good point. And ambient occlusion, as sort of many micro-shadows. A bit more advanced to implement.
Yes, they are dishonest and only loyal to stakeholders, that's capitalism 101. The other examples you mentioned are around for more time and they have safety regulations, inspections and fines. Another thing that changed is media landscape. Tesla can get caught of altering the logs in cases of autopilot crash and it just gets buried in news instead of being brought up and repeated over and over each time their stock is discussed.
The GP is right, the correct mental model is just frequency of accidents. There's a high chance the accident might not be caused by you, it could be that you're a well behaved pedestrian and you still get hit. Your own cognitive abilities and skills matter, but only so much.
Would you rather that the other driver is a person or an algorithm? What if automated driving causes few % less accidents than humans but there's no liability or accountability when it happens? What if algorithms cause more accidents but far less fatal ones? When is it the best time to deploy them at scale, to enable rapid improvements and a significantly safer roads one year down the road? Difficult questions.
3D on 2D screen is all about giving brain enough depth cues. Having size diminish with depth is a big one. The 2nd biggest thing is occlusion, which is altogether skipped in these wireframe renderings. Another big one is fog, or impact of medium the light is traveling through, very easy to do for the basic effect. It's basically just fading the colors to gray when more distant from camera.
You can have a convincing 3D with just the occlusion and fog (or even just the fog), without simulating the perspective. It's just another artistic direction.
In my case the bad internet also tips the scale.
Incredibly important research. We've reached the point where local LLMs are good enough! It takes less time for local model to take the first action on your task than it does for Claude to validate your login, put you into queue and start issuing the commands. Local models are persistent and 100% predictable unlike any cloud offering. It's better for the power system for the demand to be distributed. During the winter time the GPU also doubles as a 300W in-house heater. Not to mention avoiding personal data collection and re-selling.
I got similar info from my accountant. Along the Peddle and Lemon Squeezy she also mentioned FastSpring. Do you know anything about them?
Thank you for sharing your domain knowledge!
A brand name for an agent in VM they spin up for you.
Have you traveled the EU? Things can always be better, but looking at the rest of the world I'd say we're doing pretty fine.
I've noticed that images and fonts are broken and that it assumes emacs editor instead of using $EDITOR.
I like the general idea and simplicity of it though. I'll try to run a task or two through it.
The vendor lock-in trap does not apply here. Users don't actually even run the original weights, they mostly use open source inference engines with open source agent harness. If AI provider decided to start charging, a) released models would not be affected, and b) people would drop it instantly and move to other models.
The actual problem is that we know nothing about the training set of any open-weights model. They could be intentionally biased to influence users, from political censorship to brand advertising, or general shaping of cultural norms. You run the model on own hardware not knowing if it is designed to act against you. Having whole chain open source would allow audit and reproducing the results.
The Qwen team published the same sampler settings for 3.8 and presumably they used those while testing on benchmark. Do you believe they could have achieved higher result with top-n-sigma?
I like that there is a fallback to read the full text, but why are there two such app modes? I think if the main (non-popup) one should absorb the other's features like scrolling and bookmarks.
For some texts the layout is also important to understand the context. Poems, structured text, paragraph breaks. If you read word by word and stumble into block of structured text, it's a lot of friction to figure out. I would love a background low fidelity representation of paragraphs, like the minimap in code editors, with the current word highlighted there. It could be passively monitored for places to switch mode to reading the full text, and to keep spatial orientation around character dialogs and such. It would have to not draw attention to itself though.
The theme circle icons are not useful, they should preview text background and color.
I see Dyslexic font is already built-in!
Code is not available? It would be fun to hook this to browser extension to fast-read article in a ReadKinetic popup.
I didn't mind the controls at all, they are obviously for touch and not mouse and it seems inspired by video player controls. This is a kind of app that you spend hours with, so better to have more features and options than to be beginner friendly. Very nice app, I like it so far.
Computing power is crazy cheap and does not help. The perceived slowness is because the taken measurements are averaged over second or so. When you step on the scale your weight shifts around and that affects the measurement by far more than half a kg. Averaging gives you better accuracy and more confidence in the result but it doesn't improve the measurement precision.
The scale precision comes from calibration of the measuring mechanism done on factory line at certain weights. If your specific body weight is far from the weights used in calibration, or too much time passed, then the calibration doesn't help much.
A better microcontroller doesn't help at all. Even a cheapest uC from decades ago is good enough. Better ADC and gauge sensor would help, but even more important is good analog engineering to produce self-correcting circuits with decent noise rejection.
In general this category of products is more for tracking changes in your body weight than getting the precise absolute value. And the body weight changes a lot just from hydration level, so the ±.5kg tolerance is considered good enough.
Bad code is result of not following enough good practices rather than following one too many. Goto creates too much cognitive load. It had to be phased out to make room for better ideas on describing the intention with code more clearly.
Two recommendations for puzzle lovers that like more open ended challenges:
Deadly Rooms of Death has same movement style as SSR. It is amazing series of games each with unique mechanics. It often evokes that feeling of being in active communication with puzzle author through evolving level design.
I've also enjoyed Rhell: Warped Worlds & Troubled Times, a recently released gem. It is completely opposite of SSR, with how broadly you can approach the puzzles depending on which spells you have available (with many parts becoming too easy if you postpone them). Many games design their puzzles with a single solution. Rhell takes approach of "you cannot pass here, let's see you try" which is a fun spin on the genre. I also enjoyed how you gain abilities through game. In many modern games you unlock higher jumping by collecting skill points and spending them on skill tree. Here the same ability is emergently and organically incorporated into game mechanics. Over the playthrough the player collects new spell runes; at one point you gain rune to apply the spells to yourself. By assigning a combo strong-raise-on-self to a hotkey, you empower the character with the higher jump ability and stop using the normal jump. There's all kind of crazy and innovative ideas in there and the entire game is delightful.
This is engine for HL1 while OP talks about HL2.
I'll add, if you have a VR headset, modded HL1 runs beautifully on it with full hand controller support for gun aiming and crowbar smashing. I've also heard lots of praises for HL2 VR mod bringing the game to new levels, I have yet to try it myself.
That's unfair. Binary driver blobs are blackmail: "you bought the hardware, but parts of the laptop won't work unless you agree to run this mysterious bundle insecurely". Open weight is more like "here's a frozen brain you can thaw in a safe harness to do your bidding".