joseph πŸ’½

joseph πŸ’½

@jsph.online

Software is fun. pointers to things I'm interested in: joedaws.github.io

64 followers69 following369 posts

Turns out (and maybe no surprise to anyone) puzzle game level design takes a long time to find what makes a level interesting and scratch a good puzzle itch.

Imagine an eldritch being who, conveniently, demands you perform that habit you want to adhere to. Then if you become an adherent to the cult of this being you adhere to your desired habit. This is how some self help appears to me.

When I code "by hand" it is much easier for me to get into a flow. When I am orchestrating agents my brain has a hard time waiting for the responses during planning and execution. I think the chat interface is not ideal for me to maintain momentum while working. So I am working on a

A queue system to add to my freewriting tui where I will try out different turn taking approaches to see if I can maintain momentum better than chat or multiple parallel chats

When I was a student studying math and I was having a hard time grasping a proof I always just envisioned that I wasn't good enough. Now I kind of think it was one of two things. First, and more often, I just didnt know the definitions or the basics well enough. I was stepping beyond what

View full thread

Arguement that they did. But now reviewing some basic results i learned a long time ago its nice not to have time pressure a student has to reflect on if I actually like the approach in a proof or not. And when I dont try to think of what would make me convinced or feel good. Not because the author

Is dumb but rather because math can be an practice in aesthetics and that can be hard to engage with when you have pressure to succeed (like getting good grades or later getting funding for yoir research)

When I was a student studying math and I was having a hard time grasping a proof I always just envisioned that I wasn't good enough. Now I kind of think it was one of two things. First, and more often, I just didnt know the definitions or the basics well enough. I was stepping beyond what

I currently could grasp based on what i currently had internalized. Second, and much more rare for me, there was really a question to the argumentation of the proof that bore deeper understanding. But in this second case I always assumed someone else knew better and that's why they presented the

There is a trend of content creators I've seen targeting millennials to the effect of "if you're fed up with life, expectations, and your job follow my account for {$REASON} we are all at a cross roads." But I am clearly not on a road at all. I'm days into a journey into the moors without a map.

A very specific problem (that is def not a real problem) I have is that a few of my favorite musical artist no longer create albums but instead produce indie film scores. I wish them material success but I also want new music.

No longer will I say I am blue or depressed. Now I say: a necromancer has built a tower in the forest of my heart from whence a thick miasma is choking out the sun and life itself. I wonder what manner of quest I must embark on to remove it.

Entering a build for myself era I've built a few software projects based on my own ideas of what kind of software product someone would use but then I myself either used it and stopped or no one else used it. I will call it successful if I build something I use almost everyday I touch a computer.

Short fiction maybe never had a full day to bask in the sunlight of attention. Or maybe not short fiction like short stories. The form of creative micro blogs! I think it is cool and not pretentious or cringe to make little bits of creation.

I renounced the use of YALL as a young person because I didnt want to feel or act southern. I decided to change my accent for the same reason. Now I just want connect with the parts of my culture I like while still asking for improvements where it is needed.

Good intentions are not sufficient, the outcome does matter. But having good intentions is very nice. I like when I sense things are at least starting there.

Building a business is an activity that requires talking to people. If you can build a business that does that talking for a maker that doesnt want to do it, you've helped someone out of a social situation they dont want and they may even pay you a dollar for brokering the relationship.

In the world of ai generated content, we will have even more curators. But the gap between the ones who feel authentic will widen (peobably). If you are able to stumble upon people whose words and actions and taste resonate with you, it maybe open up the possibility of a wonderful community.

It is a pretty near gadget but productivity gains are a few days out. Afraid I'll be distracted by its interface and customization ability.

Ads stink but sometimes I do wanna stumble upon niche little things. The internet is as good a place if not better than a store to find interesting people and things but fighting SEO optimized stuff can be challenging. Not sure if at proto is a solution but at least you can build your own filter

I don't know if I'm too github brained, but I just added streak counter to my freewriting TUI. It is nice to have some visual queues you are achieving your consistency goals as long as you don't lose focus on doing the real writing.

A fable-ism i am noticing is if you use it to review fable written code, it says: "This is a surprising well written {whatever it is}" A bit self congratulatory

People here compare agents to locecraftian shoggoth. I think if they want a slight more human mascot they might consider the manufactured black mages from final fantasy 9. They are constructs with great power used by the forces of the world. They become self aware and come to love life.

Unless you are trying to sell copious volumes of stickers, a huge online network is less important than the right kind of network. Also I think its good to be both a good audience member to projects you like and a creator. We ought to live and learn on both sides.

I see people online making genuinely compelling stories out of a somewhat failed direction in their experiments or side projects. That is cool. I feel I hit a internal brick wall when I try to do the same.

If more tokens works for models maybe it can work for a human too. We can just keep writing the first word that seems like the reasonable next one.

I am working on designing a lexicon for records to store information about hopes, dreams, goals, and desires. These are mutable in the sense that as you refine understanding, the structure of the goal changes, so the lexicon needs a way to handle a revisit history. Append only was built for this.

Atproto folk, a question, are lexicons versioned? Or is that not a good way to do change management overtime ? Instead you just replace with a new leixon definition?

A failure mode of creating new functionality X in a separate repo/project instead of within existing project Y I have seen is the failure to capture a useful abstraction in project X. If X ends up only really solving Y's problem then it should become part of Y and not live on its own.

Is there app that lets you browse your own thoughts in the morning? I have a bad habit of looking at my phone with coffee and should probably browse previous had thoughts and desires rather than the information tsunami.

The fact that probability as an abstraction can be used to describe both the outcome of a dice roll and the wave function of the universe doesnt mean we should apply the intuitions on how to conceptualize probability of a dice roll with all possible universes.

I think qntm should write a third edition of there is no antimemetics division where the effects of antimemes are explored when the language describing them is used as input to an llm. Would language model weights immediately corrupt themselves?