I know this is a popular "well actually" to do, but it is not always useful in a conversation. Yes, all cancers are different, but yes, cancer is also one thing: unchecked, harmful division of cells.

Bacteria are also all different, but still they are "one thing", and despite their diversity, antibiotics exist that can deal with many species of them at once. It is reasonable to talk about bacteria and antibacterial medications, it is also reasonable to talk about cancer and cancer treatment. I truly hope cancer will meet its "penicillin" one day (yes I know this is unlikely).

The trick is to generate the PDF normally, then zip this same PDF together with the sources again, with compression level 0, making sure that the PDF is the first file to go in the archive. (Easy to write a script that does this.)

The resulting file, when given the extension PDF, is readable as PDF, and when given the extension ZIP, is extractable as ZIP. So whoever wants the source can rename the file to .zip and extract the source. The instruction to do so can be in the PDF text itself.

Why it works: a) compression level 0 means that the input files are just copied into the stream, so the PDF reader will find the PDF header, decode the rest of the PDF, and ignore the trailing stuff. The trailing stuff contains the markdown sources and the zip directory, making the file a valid archive.

I suspect that tolerances in PDF readers and ZIP decompressors are being slightly abused here, but it works with all PDF readers and ZIP decompressors that I tried so far.

"Higher levels" of AI usage are exhausting and flow-free endeavours.

This is what I've been doing in my Obsidian infodump for a while. If I know that an image is important, I generate a text description (Mermaid if possible, English if not) and paste it after the image in a block. This lets agents see the image if they don't really see it. Though my process is manual, the improvements in outcomes for agents that rely on text search/retrieval is very real and is worth it.

That is not the case, sorry. Pre-2015 Wikipedia was as honest and unbiased as we can get. Now the political, historical, philosophical segments of English Wikipedia is very biased and I cannot recommend or support it.

I believe it's worth than pointless. IMO adding such things to the context "configures" the AI to reproduce the statistics of conversations where people swore, shouted, and were unprofessional (despite the alignment runing and all that), where quality content is rarer to find. So this is bound to decrease the quality of the LLM output.

> It’s just incapable of the thing that makes a real architect valuable: saying “no.”

From my experience Claude is excellent at saying "no". It won't say "no" if the prompt doesn't call for it (it won't say "no" to your direct request to do something, usually). But it offers good critique and happily pushes back if you make it clear that that's a first class option.

There is one hickup though: the Win95 theme relies heavily on the pixel grid for its readability and aesthetics. Displays with very small pixels will either render it too small, or require some kind of upscaling with unpredictable results. 1920x1080 on a 17 inch display is probably as far as it will comfortably go.