This site uses Cloudflare for DNS, which also provides basic analytics for this site. That information is completely anonymous and there is no way to de-anonymize it, nor do I have any interest to. Google Analytics and Tag Manager are also disabled as of February 2025, and Google Search Console is disabled as of January 2026 but may be enabled again in the future if appearing higher in search results ever becomes important to me.

If you want to prevent being tracked by the tools and embedded 3rd party websites on this site, visit in Private mode or the equivilent on your browser, or clear your browser's cache (different than history) after visiting.

Embedded content acts the same to your browser as if you visited their site. For example, if a Vimeo video is embedded, by playing that video you will get the same cookies that you would if you visited its page on Vimeo.

This site does not contain any AI generated media. I am against the use of generative AI for a number of reasons:

I have only made exceptions to my anti-AI principles for two reasons, which you can read below.

There are two reasons I've used LLMs and will likely continue to (very sparingly).

The first use is a specific aspect of learning to code. I want to actually learn each coding language I use, I don't use it generate new code, but I will occasionally give a model a piece of code that I've written that isn't working the way I intend and ask what the issue is if I can't figure it out myself. Often it's a formatting error I've missed or another small mistake that I didn't catch and would have been a waste of time to find without providing a useful learning experience. I've also used it once or twice to rewrite a piece of my old code to be more efficient, which I've then tested and studied so I can write it myself next time. For these tasks, I use local models that run offline on my computer to minimize unnecessary data center use.

The second is the only use I've found to date for the large models that (at least for now) can only run on data center machines. The project isn't on my site yet, and I'll link to it here when it is, but it involves correcting the formatting of hundreds of documents and converting the correctly formatted content to HTML, which I can then display natively on a webpage and convert using traditional methods to other file formats. It's a large scale undertaking that would not have been remotely feasible without hundres of hours of menial work before the advent of modern LLMs, so I've made an exception for that project. When it's on my site, there will be a more in depth explanation of the process, as well as disclaimers and an LLM use tag of some sort on its page.

Both of these exceptions use LLMs to change the formatting of content, not the content itself. The point is either to assist in learning which avoids repeat uses for the same task, or to accomplish something that would have been effectively impossible otherwise. All of my LLM use undergoes rigorous scrutiny by me, often to the point of being completely overwritten by the time the project sees the light of day. Any mistakes are my own, and any use of the word "delve" is because I read books.