Creating a new blog post for Hugo couldn't be simpler. One simply creates a new Markdown file in the correct folder, making sure it contains the proper YAML front matter, and...OK, it's not that simple.

Normally I use a small lisp function in Emacs for creating posts, but that doesn't help me on days that I don't feel like using Emacs^[1].

Another option would be to use the built-in hugo commands, but for some reason I never think of that. Probably because it requires that I cd to the right place and tell it what to name the file.

Mike Hall recently published an update to the Ruby script he uses. It's called hpost.

I used hpost to create this post. Worked fine, but made me think about how I might do it, given some private time with Claude.

• I wouldn't use Ruby. It's a nice language, but dependencies and version management confound me. Same with Python. I bet it could be done using bash. Or maybe Go, which would be Hugo-appropriate.
• New posts would go in dated folders (e.g. /posts/2025/07/my-new-post.md)
• I'd offer a choice to use a Hugo bundle (/my-post/index.md vs my-post.md) to make it easier to include images.
• Including a slug: entry in the front matter would be nice for maintaining URL consistency in case of title changes.
• Daily Notes would have to conform to my own conventions (/journal/2025/07-july/20250718/index.md)

What I'll likely do in real life is to continue using Emacs for creating posts. I've got Dired bookmarks for getting to previous posts, dwim commands for building/deploying, etc. hpost (or equivalent) is still nice for days like today when I'm off Emacs.

If he's a bit manic for you, here's part of the transcript that I've reformatted a bit:

...if you're a progressive, don't do this. Don't go on to the internet and go, "Hey fellow progressives, like maybe we should consider having, you know, a little - just just chill out a bit and have a bit of humility and realise that not everyone's had the opportunity to read the books we've read and not everyone's, you know, learned the lessons that you've learned. Maybe we should apply the Principle of Charity and not assume that anyone who disagrees with us has a nefarious intent and try and hear the best version of the argument, not the worst. And maybe we just need to-"

They go, "Don't you police my tone, you straight white male! Check your privilege!"

And I go, "Fuck, you're absolutely right. I'm sorry, I'll check my privilege." So I go and check my privilege, which I keep in my library on a mahogany shelf in between my BMW keys and a large Fortnum and Mason's tin of 50 dollar notes. And I come back and I say, "I checked my privilege and you're still fucking wrong."

Because, because, because it cannot be okay. It can't. It cannot. It cannot be, can it? If , if the intention of progressives, which I assume it is, is to progress forward into a future of more empathy and understanding for more and more people, it cannot be, can it, that the primary mechanism by which we're going to make that progress is the suppression of empathy and understanding for anyone who doesn't align with our beliefs? It cannot be that unmitigated expression of furious outrage will somehow alchemize into a future of peace and love.

And of course there's something to be discussed about the justification - of course there's many, many reasons to be righteously furious. I'm a straight white cisgendered male and I could be furious about 700 things before any given breakfast. But it doesn't fucking matter if it's not helping, and it's not helping. The tribes are getting further and further apart. And then the point is, it's very hard to change people's minds - but we got there - but it is not impossible.

However, one thing is a hundred percent certain: if you tell someone that they're a fascist and publicly shame them on the Internet for what they believe, you've lost them.

You have prioritised your need to express your outrage and, if we're honest, more often than not signal your virtue over the possibility that you could utilise your educational privilege to reach across this algorithmic chasm.

org-mouse is a built-in package for Org mode that lets one do some handy things using the mouse in Org files. Things like checking checkboxes and toggling subtrees.

(use-package org-mouse
        :after org)

Or just (require 'org-mouse)

Clicking in [ ] Do this thing to complete it is easier than navigating to the line and hitting C-c C-c for each item.

There's no shame in using a mouse, even in Emacs. I sometimes prefer to sit back, grab the mouse, and click things.

Obsidian should be the right answer for taking notes. It's free, powerful, local-first, and does everything one might need. The ecosystem is huge.

Every few months, I find myself questioning my decision to stop using it after the last attempt^[1] So, I re-install and start again, usually from scratch.

At first, it's such a relief. There are buttons for doing the important things. There is a command palette for doing just about anything else. I can easily remap the keyboard shortcuts. The graph view, which I ridicule as useless most days, looks pretty cool, actually.

It's so nice being able to just sit back and click around in the UI without the usual set of keyboard contortions in Emacs.

What a refreshing change!

Then, I'll dig into some recent blog posts or YouTube videos from people sharing their latest configurations, plugins, workflows, etc. This is where things start to go wrong. I overcomplicate things right out of the gate. Too many plugins, too many folders, too many settings. If I wanted to make my notetaking tool into its own hobby, I'd keep using Emacs.

Obsidian's biggest flaw is that it's not Emacs.

Still, I'm determined to give it a fair shake, again, just for grins and to alleviate boredom.

This time I'm not installing any 3rd-party plugins until I'm sure I need them (for whatever meaning of "need" is appropriate). I'm not creating a bunch of nested folders. How about this, as a start?:

├── _attachments    # all attachments. No hierarchy
├── 00 System       # templates, scripts, etc.
├── 10 WIP          # Work In Process, projects, etc. Subfolder for each project.
├── Clippings       # Web clippings (to be filed or deleted later)
├── Daily Notes     # Daily Notes, obviously. Yearly folders.
└── Knowledgebase   # All other notes. No hierarchy.

I'm also avoiding superfluous front matter. Good grief, I sure can overthink the front matter. My default template is just a title and (created) date. Speaking of default templates, why the hell can't I set a default template for all new notes? I'm trying to avoid installing Templater, but it's going to be tough.

So far I've only installed the Calendar plugin. It remains a mystery that this isn't built in.

Anyway, it's always fun for a while. Let's see how long it lasts this time.

previously, previously, previously

Glass is a nice way to host and share photos. I've used Flickr for 20 years, so I didn't think I needed Glass, and planned to let my subscription expire.

The single most frustrating thing with Flickr is that there's no good way I know of to keep a public photo off my profile/timeline. If I want to dump a bunch of images to some album, they all show up on my profile. It's always bothered me.

Playing with the new(ish) "Series" feature in Glass, I noticed that there is an option to "Hide from profile".

That's all I ever wanted from Flickr. And while Flickr is still the best way to share and host photos, it's no longer much of social spot. This is such a shame, but here we are.

So I started thinking. What if I could replace my "Film Roll" posts on the blog with links to the Series on Glass?

Like this: 2025 Roll 035 (Rolleiflex 2.8D), a Series by Jack Baty.

That lets me use a nice, social photo sharing platform for my film rolls, without needing to maintain a separate photo blog, or deal with posting photos here using Hugo.

Yes, I know, "What if Glass shuts down!?" Fair point, but I don't think I'll worry about it right now. Let's try it.

I like to have Make and Model information available in film scans, and use exiftool for this. I have a lisp function in Emacs that does this, but sometimes I'd like to do it from a terminal instead.

So I asked Claude for help. The result was camera-exif-tui. It's a tiny Go app that launches a TUI that allows me to select a make/model and a folder full of image files. It calls exiftool and updates the images with the selected camera info.

There's a cameras.yaml file in ~/.config/camera-exif-tui/cameras.yaml with my cameras configured.

cameras:
  # Leica Cameras
  - make: Leica
    model: MP
  - make: Leica
    model: SL2
  - make: Leica
    model: M3

  # Nikon Cameras
  - make: Nikon
    model: FM2n
  - make: Nikon
    model: FE2
  - make: Nikon
    model: F100

  # Olympus Cameras
  - make: Olympus
    model: OM-1n
  - make: Olympus
    model: OM-2n
  - make: Olympus
    model: Stylus Epic

  # Hasselblad Cameras
  - make: Hasselblad
    model: 500C/M

  # Rolleiflex Cameras
  - make: Rolleiflex
    model: 2.8D

These quick, handly little scripts are ideal for AI. People seem to ignore this when railing against LLM use. The whole thing, including writing the README file and setting up the git repo took less than an hour. I would never have bothered with this without something like Claude. It was fun! I don't want to learn the details of Golang, but I'm happy to tweak whatever Claude spits out. Now I have something that I wouldn't have had before, and it was easy to do. I don't care if it's "correct" or not. It works for me.