AeonFelis

joined 2 years ago
[–] AeonFelis@lemmy.world 4 points 1 hour ago

I can bearly wait to can some bears

[–] AeonFelis@lemmy.world 13 points 12 hours ago (7 children)

dumb lighter for idiots

Meaning a tool idiots can use to set the dumbs on fire?

[–] AeonFelis@lemmy.world 2 points 1 day ago

Clearly a duck-billed platypus.

[–] AeonFelis@lemmy.world 39 points 1 day ago (15 children)

Of course not. What kind of man would want to see his family burned with him in a fiery death trap?

[–] AeonFelis@lemmy.world 10 points 1 day ago

Science isn't about WHY. It's about WHY NOT. Why is so much of our science dangerous? Why not marry safe science if you love it so much. In fact, why not invent a special safety door that won't hit you on the butt on the way out, because you are fired.

[–] AeonFelis@lemmy.world 1 points 2 days ago

Housing? As in - for people? Where are the data centers supposed to be then? Ever think about that? No. You only think about yourself.

[–] AeonFelis@lemmy.world 19 points 2 days ago (5 children)

I'm sure they can work out a mouth next

[–] AeonFelis@lemmy.world 1 points 2 days ago

Considering how these companies are losing money because they subsidize these tokens - I doubt that cost is really absorbed.

[–] AeonFelis@lemmy.world 7 points 2 days ago (2 children)

The proper response to dystopian prophecies is not "challenge accepted"!

[–] AeonFelis@lemmy.world 1 points 2 days ago (2 children)

$20k is what it would cost you or me, but it’s just free for them.

No it isn't. This is not regular software where the bulk of the price is the licensing. With slope-as-a-service, the bulk of the price is the data center operation cost - which Anthropic is certainly not getting for free.

[–] AeonFelis@lemmy.world 3 points 2 days ago

Around the turn of the decade there was this big movement to rename the master branch to main. GitHub, too, made that switch and when you create a new repository the master branch is called main. The original flowchart was from 2010, when the main branch was still called master, so it's called master there.

The AI generated flowchart, of course, is not a plagiarism machine and it's exactly like a human being that is merely inspired by the source material. Surely it used the up-to-date name for that branch?

[–] AeonFelis@lemmy.world 6 points 3 days ago (1 children)

99% pass rate? Maybe that’s super impressive because it’s a stress test, but if 1% of my code fails to compile I think I’d be in deep shit.

Also - one of the main arguments of vibe coding advocators is that you just need to check the result several times and tell the AI assistant what needs fixing. Isn't a compiler test suite ideal for such workflow? Why couldn't they just feed the test failures back to the model and tell it to fix them, iterating again and again until they get it to work 100%?

 

Encountering one of these embedded tweets in a blog post, my hand instinctively moved to click the X and close it. That took me to the website.

Could this be a clever ruse to generate more visits? Is Elon Musk actually more cunning than we give him credit?

 

I have this idea for a certain game development tool, but before I start another side project I want to check if something similar already exists.

An important part of game development is fine-tuning numeric values. You have some numbers that govern things like character motion, weapon impact, enemy AI, or any other game mechanic. For most of these there is no "correct" value that can be calculated (or even verified!) with some algorithm - you have to manually try different values and converge to something that "feels right".

The most naive way to fine-tune these numbers is to have them as hard-coded values, tweak them in code, and re-run the game every time you change them. This, of course, is a tedious process - especially if you have to go through long build times, game loading, and/or gameplay to reach a state where you can test these values (that last hurdle can often be skipped by programming in a special entry point, but that too can get tedious)

A better way would be to write these numbers in configuration file(s) which the game can hot-reload - at least while in development mode. That way you can just edit the file and save it, and the game will reload the new values. This is a huge improvement because it skips the building/loading/preparing which can drastically shorten the cycles - but it's still not perfect because you have to constantly switch between the game and the configuration file.

Sometimes you can use the game engine editor to tweak these while the game is running, or create your own UI. This makes the context switches hurt less, and also lets you use sliders instead of editing textual numbers, but it's still not perfect - you still have to switch back and forth between the game controls and the tweaking interface.

Which brings us to my idea.

What I envision is a local fine-tuning server. The server will either update configuration files which the game will hot-reload, or the game could connect to it via WebSocket (or some other IPC. But I like WebSocket) so that the server could push the new values to it as they get updated.

After the server deduces the structure of the configuration (or read it from a schema - but providing a schema may usually be a overkill) you could use its webapp UI to configure how the values would be tweaked. We usually want sliders, so you'll need to provide a range - even if the exact value is hard to determine, it's usually fairly easy to come up with a rough range that the value must be in (how high can a human jump? More than 5cm, less than 5m). You will also decide for each slider if it's linear or logarithmic.

The server, of course, will save all that configuration so that you won't have t reconfigure it the next time you want to tweak values (unless there are new values, in which case you'll only have to configure the sliders for them)

Since this would be a server, the tweaking of the values could be done from another device - preferably something with a touchscreen, like a smartphone or a tablet, because tweaking many sliders is easier with a touchscreen. So you have the game running on your PC/console, gamepad in hand (or keyboard+mouse, if that's your thing), and as you play you tweak the sliders on the touchscreen until you get them just right.

Does anyone know if a similar tool already exists?

 

Narrative scripting languages like Yarn Spinner or Inkle were originally meant for writing dialogue, but I think they can also be used for scripting the world progression even when no dialogue or even narration is involved.

Example for something silent that can be scripted with a narrative scripting language:

  1. When the player pulls a lever...
  2. Move the camera to show a certain gate
  3. Open the gate
  4. Move the camera to show something interesting behind the gate
  5. Return the camera to the player

Even though no text nor voice are involved here, I think a narrative language will still fit better than a traditional scripting language because:

  • Narrative languages describe everything in steps. Scripting languages will need to work a bit harder to generate steps the actual game engine can use.
  • Narrative languages have visual editor that can help showing the flow of the level as nodes.
  • The interface between a narrative language and the game engine tends to be seems to tend to be higher level (and less powerful) than the one with a traditional scripting language.

On the other hand, flow control seems a bit more crude and ugly with narrative scripting languages than with traditional scripting languages. It should probably still be fine for simple things (e.g. - player activates a keyhole. Do they have the key?), but I wonder if a game can reach a point where it becomes too complex for a narrative language (I'm still talking about simple world progression, not full blown modding)

view more: next ›