Paul Lee

Paul Lee

@strivenword.com

hypertext writer

34 followers73 following112 posts

Every videogame micro sub-genre and graphical style from every era deserves to be explored and innovated upon. (In college, a friend and I tried to spec out a 2.5D system in HTML 5... it wasn't practical using the IDE systems available at the time. Maybe today we could have done it.)

Korzodin (Relentless Frontier)@korzodin.bsky.social

Working on a limb targeting system for the Void enemy in Relentless Frontier. You can shoot them in the leg to slow them down, then line up some headshots to kill them quick! Lotsa WIP here, mind the prototype dust all over the place. #ScreenshotSaturday #Gamedev #FPS

A Runner agent in Opencode uses skills to run a script, then analyzes output, documents results within the repository, and updates a file that triggers the Maker agent in Aider to modify the code. I finally have a reason to use tmux to manage multiple terminal sessions.

Working on the open graph and twitter card metadata for my blog. Funny how we're still traditionally putting a `twitter` property namespace in there, when there has been no platform named #Twitter in a long time. www.strivenword.com/notebook/Sep...

strivenword.comChildren of the First PandemicThe year 2019 saw the outbreak of disease leading to widespread social unrest in Europe. I’m referring to A Plague Tale: Innocence, last year’s historical adventure game that won awards for narrative ...

What are the most universal, non-corporate-defined solutions for open graph data, to define "card" design patterns for links on microblog platforms? #webdev

I just decided that I think AI generated text can never legitimately be included in traditional linear fiction -- i.e., fiction as understood in the formats of novels, novellas, short stories, etc. Digital-native interactive formats are a different story, and a complicated one. 🧵

View full thread

All this to say, I think LLM generated text can have great entertainment and/or artistic value, legitimate in *some* way (laying aside the legal and ethical questions). But the value is not in the stochastically generated text, but in the emergence pertaining to input. Conclusion: 🧵

I think #LLM fiction should be a thing worth doing, an interesting and potentially legitimate thing. But it is inherently a different thing from static fiction and even traditional interactive fiction, where the use of AI invalidates some of their value. So LLM fiction is its own separate thing. #AI

I just decided that I think AI generated text can never legitimately be included in traditional linear fiction -- i.e., fiction as understood in the formats of novels, novellas, short stories, etc. Digital-native interactive formats are a different story, and a complicated one. 🧵

View full thread

That is, for both novels and traditional IF, the art of the story is its synthesis through human authors' experiences. But LLM utilization accommodates user participation in a different way than IF. I think LLMs can utilize stochastically generated phenomena in ways that are contextually artistic.🧵

I think the LLM experience is again similar to the possibilities in non-LLM electronic liturature -- procedurally or randomly generated texts, for instance -- but also different, not simply due to the generative training-based nature of LLMs, but also due to the way they interact with user input. 🧵

I just decided that I think AI generated text can never legitimately be included in traditional linear fiction -- i.e., fiction as understood in the formats of novels, novellas, short stories, etc. Digital-native interactive formats are a different story, and a complicated one. 🧵

For some of the same reasons that AI text shouldn't be in a novel, it makes sense to be able to define traditional interactive fiction (be it parser-based text adventures or choice-based subtypes) as not having AI generated content or utilizing LLMs. #textadventures #IF #interactivefiction 🧵

Decentralized finance isn't supposed to be abusive rich bros gaming the system to become tech overlords, nor lawless anarchists trying to loot and burn the system. It's ordinary people interacting with each other in emergent economies of value that can't be quantified by the system. #DeFi #crypto

Way more fun than hallucinating with an LLM.

Katherine Traverse@kattra.bsky.social

Walking by the river, Ezra spots an #unmoored punt bumping against the bank. Once painted green and gold, flaking away now, black slime at the waterline. Approaching carefully, Ezra spots the name “DATURA” carved into the bow. Pulling back a canvas cover, they find: #vss365 (see options ⬇️ ⬇️ ⬇️ )

I use Alacritty as my primary terminal app on both the Mac and the Chromebook, but the font family I was using had a different name on Debian than on MacOS, so I took the opportunity to select a different font and risk some time on customizations, because joy matters too.

Spent too much time synchronizing my tools between Mac and Debian, where I've been learning the i3 window manager, which is perfect for the small screen. #Linux

I think of Brave as a good version of Old Microsoft and Kagi as a good version of Old Google.

BleepingComputer@bleepingcomputer.com

Brave has announced the public release of Brave Origin, a paid minimalist version of its browser that strips out cryptocurrency, AI, rewards, and other monetization-focused features.

A great thing about Python culture, which I haven't quite figured out yet, is playful humor enshrined within a rather rigid, formal structure. The Python community feels very institutional, but it feels like the best kind of institution. Google as a company, for instance, feels like a loose canon.

Hear from PSF's @pypi.org Support Specialist Maria Ashna on what her day-to-day looks like, how she cleared multiple months-long backlogs, and the future of PyPI Orgs in this Behind the Commit episode from @clytaemnestra.bsky.social. ▶️ www.youtube.com/watch?v=OGIz... #Python #PyPI #OpenSource

When I was 13 in 2002, I first learned to code by following the Inform Beginner's Guide. Always grate to Roger and Sonja for that. Used Gemini to make a similar guidebook for Python based on the text of the PDF. Then used #Opencode to build the 3 example in #Python.

Using #uv with #invoke in my #Pelican website to rebuild on every change and free up the localhost port if it was already in use when listening for changes, also closing the port before quitting the script.

Still conflicted over the proper stance toward privacy and personal ownership of one’s digital footprint in the latest version of the digital landscape.

View full thread

what relationship the use of Pinterest (which is not an AI platform) has in relationship to total profiling, i.e., to the total information available to ad vendors or commercial AI companies for personalization purposes. Finally, I would like a reasonable projection of what the expected outcomes

of heavy, willful, deliberate use of all sources of information profiling with consensual allowing of cross-indexing would be for a responsible adult; i.e., how the decision to allow maximum cross profiling will affect a responsible adults life for the next 5 years.

Still conflicted over the proper stance toward privacy and personal ownership of one’s digital footprint in the latest version of the digital landscape.

View full thread

across corporate tech platforms leads to integrated information profiles—i.e., using ChatGPT from an associated Google account. I want to know whether API or developer usage of AI tools from Anthropic or OpenAI, etc., can be associated with personal data. I want to know

what role things like Palantir play in all of this. I want to know whether use of ChatGPT can influence ads seen on YouTube. I want to know whether heavy personalization in MidJourney can affect advertising or any other information dynamic across the web. I want to know comparatively

Still conflicted over the proper stance toward privacy and personal ownership of one’s digital footprint in the latest version of the digital landscape.

View full thread

Please do some research. I want to understand the nature of data footprints in the post-AI digital world. I want to know how profiles from trackers across the internet interact with datasets accessible by AI. I want to know how people's own AI use,

when open to the same networkers of traditional profiling, interacts with that; i.e., whether data in someone's OpenAI account resulting from heavy AI use might be related to add data from DoubleClick or whatever ad network. I'm interested in whether connected accounts

Still conflicted over the proper stance toward privacy and personal ownership of one’s digital footprint in the latest version of the digital landscape.

I bounced on Logseq for day task journaling. Now using the Todolist plugin on #TiddlyWiki, which has interstitial and task-based modes and allows me to take it anywhere, now that I finally figured out a comfortable way to run my local TiddlyWiki server.

I kind of like the idea of copy-paste LLM game scenarios, such as the one entered into @ifcomp.org. I've done this on my own and played AI Dungeon early on. But a bigger goal would be to use sophisticated LLM prompting and RAG to create a virtual GM for D&D, etc. I don't want pre-built agentic stuff

My problem with #Obsidian is that I can't easily make meaningful or useful metadata, and my problem with #TiddlyWiki is that while I can do exactly that with fields, the effort involved in implementing a custom way to render them for each project is discouraging.

I’ve decided to swich to MacPorts to manage #Python versions on my laptop, with the exception of using `uv` where I need to manage project dependecies--my static blog published through Pelican, and a custom #Vim plugin that restarts #TiddlyWiki whenever I change a `.tid` file.

The discussion about the ethos of open source in regard to the WordPress drama in Lex and DHH's podcast got me dreaming again about the ideal New WordPress. I see how Rails encapsulates a lot of the hacker-builder energy that the old WordPress community thrived on.

Obsidian, TiddlyWiki, and Logseq—I’m making a website to review and compare the three of these personal knowledge management systems, with corresponding example zks. #zettelkasten

I got caught by Logseq, chiefly, trying to learn more about it, always just one more thing to read. But I still plan to launch this comparison website some day.

Used Neovim to solve a problem faster than I could have without it. Probably will never compensate for all the time I’ve put in to trying to learn it. Oh well.

New “start slower” option in Dino Run. It’s part of core Chromium, not just Google Chrome, seen here in Brave. I love that this easter-egg is so deeply embedded in the experience of the Web.

I’m discovering Vue.js thanks to my SUNY class, and intrigued that it’s so easy and so hupertextual. It actually reminds me of TiddlyWiki, an entirely different kind of Node app. I think I can develop for Vue with transclusion techniques I learned from TW.

Oh my Goodness–ThePrimeagen just name dropped a specific scene from The Wheel of Time books on the Lex Fridman podcast! (At about 24:40.)

Mindblown by ES6 after studying JavaScript seriously again for the first time in about 10 years. I just have to wait around a while, and the hard stuff I struggled through years ago is suddenly easier. const hardThings = () => abstraction;

I’m finally fluent enough in TiddlyWiki to use it comfortably for day to day practical purposes. Now, if only I had the motivation to do so. And if only I knew how to make a widget out of an element inside of an SVG... for my main project.

You are up to date.