colbyn

Something I’ve been thinking about.

Your eyes are broadly sensitive to the visible light spectrum.

But this is not uniformly distributed.

Emission in the green - yellow band is bright, equivalent intensity at higher energy wavelengths such as the 450nm emission pump of most LEDs can be quite dim.

It makes me wonder, IR is invisible, yet it can be very intense. Because you can’t perceive it, your eyes can’t adjust… Just as you can fry your reta with invisible light, what about ordinary LEDs that have a sharp spike at the 450nm region?

Could your eyes be slightly overexposed because it’s physically brighter than your perception?

I believe some damage does occur. But the eyes can heal. Blue light toxicity is a thing. But I think the threshold is based on irreversible damage… But I sometimes wonder if it’s too generous in favor of LED manufacturers.

Brightness is important, and blue light isn’t necessarily the sole factor of concern.

What you want is to accommodate both. Which has been a subject of recent academic literature concerning industry standards.

Specifically, there are new proposed metrics that accommodates brightness, but weighted in terms of its melatonin suppressing effects.

See the ref at the end.

Because you can have luminaries that are meaningfully bright physiologically speaking but dim to your perception of light or vice versa.

Your perception of “brightness” and also your perception of warmth (ie bright cool = day; warm dim = night) are not reliable heuristics.

Because your perception of light is calibrated in a manner that was adaptive to the species. It’s not a spectrometer. Moreover ipRGCs (intrinsically photosensitive retinal ganglion cells) have a different brightness weight that is distinct from what you perceive as bright light.

See the included document for an example.

But do note, it’s a work in progress, so I can’t say it’ll always be page 4 or whatever it happens to be.

Also, there my focus has been on comparing the emission of different luminaries irrespective of magnitude. But once you know the Mel-DEI, the brightness magnitude is just the product of that and the bulbs lux. Or so I’m inclined to say off the top of my head… This can be a surprisingly complicated topic, everything I know, has been from my own exploration. Whatever I say now, I could have very different opinions a year from now.

If I was more organized with regard to my academic literature reading list, I’d cite that but I’m not sure what happened to that list.

But the one thing I’d say I did very well: beautiful illustration if I dare say.

But I digress…

Here’s the link:

https://colbyn.github.io/lighting-misc-v1/main.pdf

(I’ve been working on it on / off for some time. It’s very much a rough draft at this time. Moreover its function is really to serve as a visual aid when talking to folks. So it’s very light on the commentary.)

> As a native speaker, it feels like reading an impression of a literature book by a high school English class’s most overconfident student who’s only ever read LinkedIn-speak.

You wouldn’t happen to have gone to Westlake HS? lol i may have been that person… Life has since slapped me in the face a few times.

I’ve thought about this. Or at least something like that.

When reading dune there’s a class of women who can control their biochemistry in such a way that they can eg prolong lifespan, or anything else within the realm of feasible biochemistry.

Here I was like: how does the nervous system function in that way? How could signals from the mind control the behavior of molecules?

The mind cannot itself feel pain, similarly the mind has sensory limits within the body. As argued in a book called the body a guide for occupants: cancer is a very good example of a disease that our nervous system should be able to detect but for whatever reason we don’t, probably because cancer is something we can only now treat.

Not all of the body can be made legible to the mind.

Although maybe a better question: why don’t we have a dedicated organ that can sample blood with laboratory like precision and make anomalies available to the conscious mind beyond whatever faculties we currently have?

PS

One thing that definitely should be within the realm of conscious control: body fat. There are ways of forcing the body to metabolize more fat for energy and the biggest problem is managing excess heart (easily becomes lethal). But this could be super useful in cold climates. Imagine being able to literally burn body fat to stay warm? The amount of heat that can be released is enormous. Nowadays most of us could probably afford the otherwise superfluous expenditure of body fat (beyond essential functions).

In my experience Grok is the least likely to push back on crazy ideas out of all major chatbots and it’s more often wrong on technical matters. Although I suppose this isn’t necessarily bad. I go to Grok for subjective explorations.

…

I can’t get downvoted further but as a friendly FYI this page can’t be opened in iOS safari (too many redirects).

But I was thinking some time ago I disabled algorithmically prompted content even though it had its advantages for work related content… In some ways the transition was like quitting smoking. I kept going back to such platforms, getting a blank page, closing the page, then repeating some time later and so on.

Oh I guess I stand corrected. I sometimes wish AI wasn't a thing because any natural talent for writing has now been rendered almost meaningless.

Why should we presume it’s an AI description? The writing seems reasonable enough. Some people are more naturally inclined towards articulate writing than others (seems like projection) and furthermore the OP claims to be a screenwriter.

Design can go a long way when reading long form text. If someone here is in contact with the author please tell them to improve the typography; most notably smaller and justified text for mobile phones. Other designers could probably weigh in. I’m not an expert, but well designed text goes a long way towards comfortable reading.

Apart from that, content wise a preliminary abstract is nice to have. I do like how the author provides a table of contents.

The title could have been a line out of Dune.

It’s interesting that many of us myself included once thought that the butlerian jihad was silly until now. Frank Herbert wrote something that is particularly prescient.

(Usually writers are just a decade ahead of their time. Whatever Podcasters are talking about today, has usually already been discussed in literature a decade ago. Prediction markets come to mind. Socially, over vs under population as discussed in popular books like the rationale optimist or the accidental superpower.)

Nope. For all I know it’s due to the sins of the people. The reckoning of heaven.

For context there are many theories for why younger generations are less “cognitively capable” than older generations. Nowadays we call it the reverse Flynn effect. IMO this article is nitpicking, probably confirmation bias at play.

Okay I see. Are people being attacked for engaging in AI research/dev irrespective of their character or other personal attributes?

I’m confused can someone please explain to me why he or she is so controversial?

I have yet to read this article (in full), but I love trees! As an amateur AST transformation nerd. Kinda related but I’ve been trying to figure out how to generalize the lessons learned from this experiment in autogenerating massive bilingual dictionary and phrasebook datasets: https://youtu.be/nofJLw51xSk

Into a general purpose markup language + runtime for multi step LLM invocations. Although efforts so far have gotten nowhere. I have some notes on my GitHub profile readme if anyone curious: https://github.com/colbyn

Here’s a working example: https://github.com/colbyn/AgenticWorkflow

(I really dislike the ‘agentic’ term since in my mind it’s just compilers and a runtime all the way down.)

But that’s more serial procedural work, what I want is full blown recursion, in some generalized way (and without liquid templating hacks that I keep restoring to), deeply needed nested LLM invocations akin to how my dataset generation pipeline works.

PS

Also I really dislike prompt text in source code. I prefer to factor in out into standalone prompt files. Using the XML format in my case.

My thoughts exactly. I’ve been using them for exposing local services to the public internet from my home network. Super convenient for initial proof of concept work…

Cool I sent you an email and included a screen recorded video demo of that project.

Although FYI: the UI in that demo needs some work since updating iOS but it's an interesting proof of concept that took me some time to figure out (in the future I think I'll do full page swipes of the entire downstream conversation tree and not just at the sibling level although this is less power efficient). Also that demo doesn't include my work in markdown rendering since I have yet to integrate the two. I've been working on both of those projects on and off for the past couple years plus some.

As old a rust dev. Any interest in the client side of things? I’ve been experimenting with a few ideas in this regard. Especially, true native markdown rendering and a branching conversational data model. (This complicated the UI and required better layout handling.)

A while ago I figured out how to extend my native TextKit 2 based markdown renderer to support full blown markdown tables with full text selection support. At the time this was quite novel.

I'm now working on a fresh rewrite with iOS support... But as someone who's wasted a lot of time on side projects, could this be profitable?

I almost had a sponsor but he ghosted me. Since then other libraries and commercial apps like ChatGPT and Grok have improved their markdown UI/UX.

I'm thinking about abandoning this endeavor. However it feels wasteful to just drop this and all the experience I accumulated in the meantime. Moreover this proof of concept prototype is still superior in a few ways, notably with regard to text selection and typesetting.

Perhaps because people who are convinced that they’re on the morally right side of history will always assume the problem pertains to others.

Not sufficient. Would people pay money to have these ideas explained to them by a former self educated zealot? (Id have to refresh my memory but if my younger self could get into this anyone can.)

That’s too simplistic because there’s different types of equality. Things can be different but semantically equal in other regards. Equality in elementary algebra is very vague.

Oh also I’d like to add to the above, the TextKit2 API is way more freeform than people think. You could probably implement your own web browser upon it with optimized line by line text rendering (which isn’t that bad). One thing I always wanted to experiment with is rendering markdown content with horizontally scrollable text fragments for table rows and certain fenced code blocks. Super cool and practical idea for native markdown text rendering. Moreover, in some ways it seems pretty easy to do in TextKit2 since you get very fine grained control over text layout fragment rendering.

Oh hey good to see some more literature on TextKit2. I remember diving into it when I wrote my proof of concept markdown renderer:

https://github.com/SuperSwiftMarkup/SuperSwiftMarkdownProtot...

It’s a good example of what else you can do with TextKit2, beyond plain text rendering. Here, I can draw text layout fragments in a core graphics context and use typographic information to also render markdown specific graphics in the background or foreground such as markdown tables. One day I’d love to experiment with a markdown centric spread sheet app.

At the time there was so little online info that I had to figure out everything on my own (and no chatbots didn’t have a clue about TextKit2), but STTextView was a great reference. Overall his open source work was very much appreciated. (I should probably say as much in the read me.)

That experiment was also a ton of work that ultimately didn’t go anywhere. Most people would rather just use a relatively subpar embedded browser (i.e. web view).

I thought this was referring to articles as in the part of speech (i.e. there are nouns, verbs, but also article like “a” or “the”) given the title and something spanning across languages… I wonder what his exact thought process was that motivated all that effort?

Suppose someone wanted to dive into other projects with the ambition of finding high value bugs. Besides chromium what would you recommend or consider? What would be your thought process for deciding what projects to look into?

On a side note some time ago I wrote this utility specifically for optimizing the encoding: https://imager.io/

I often wonder if I should have done something productive with this tool (kinda abandoned it)…

Personally I’ve been thinking about this problem for some time and have had a neat idea that im almost tempted to patent, I have yet to test it via an actual implementation but the core idea is quite simple…

Today it's LLMs/ChatBots, when I was younger it was StackOverflow. Interestingly, when I was just starting out I remember being told to avoid copying & pasting snippets from SO in favor of learning to solve problems on my own, or at the very least, slow down and manually type out the snippets (which actually helped). I used to be against LLMs/ChatBots (as someone who likes to write) but now I've gone all in, it's just such a great productivity booster for very common grunt work that doesn't require too much project specific context... At the same time, while I do sometimes wonder if this will cause my problem solving skills to stagnate, I'm far more concerned about the broader ecosystem. LLMs perform best when the solution domain is well represented in their training dataset and with hard cutoffs and whatnot, I do wonder if this will cause tools / frameworks / libraries to stagnate because people will favor older tools over newer tools for better ease of use. Great for established authors, not so great for new ideas, something that too few people are talking about.