After praising the utility of a couple of LaTeX classics, I'm a bit more mixed about Donald E. Knuth's "The TeXbook". It's a lovely artifact and piece of history, but we've not made use of it yet. It's great to have it and we plan to get hold of the rest of the series.
And we had these excellent books on our bookshelves already. It's nice to reward the hoarder's instinct twenty years after first getting them. Both have helped us create a production layout pipeline that works for us.
The team has worked with LaTeX-formatted books before, and some of us even typeset our undergraduate thesis with LaTeX. When it came to typesetting for a new publisher, we chose Markdown plus LaTeX. This combination is open source, scriptable, and has everything we need to typeset technical content.
Caveat: we don't constantly test LLMs on indexing tasks, but have checked with various models every six months. It could be another AI revolution or it could just be that indexing is uniquely human.
🥸
#IndexSky
What about AI in indexing? So far, LLM indexing is *by far* the worst product of AI we've come across. There appears to be something uniquely slippery about indexing that LLMs have not been able to learn. Even a brute-force index formed from a table of contents is magnitudes better.
#IndexSky
Then we find all the keywords marked by the author or editor and slip them into the final index, with a few extra based on the indexing read-through.
Like all start-ups, we're iterating on this. But remember: indexes are cool.
Ullstrom's advice for how-to books: base the index on the table of contents. Programming books are how-to (and ours lean heavily on the why? too). Therefore, our indexes start life as an alphabetical table of contents. It's crude, but want to build something? Look up "Building ..." in the index.
Indexing books is difficult and not something the Project Q team had done much of before. So we turned to a couple of great books: new-wave "Book Indexing" by Stephen Ullstrom and old-school "Indexing, the Art Of" by G. Norman Knight.
Love them both.
@stephenullstrom.bsky.social
#IndexSky
After a lifetime in technical publishing I have a fair few books related to making books. It always makes me glad when I glance up from my desk to look at them.
I'll look at a few of these this week, including the big reveal of a spiral-bound classic. Any ideas what it is?
#publishing
Why write for Project Q?
Run by technical people: the founder has an AI degree and regularly writes code/contributes to open source projects.
Personal GitHub: github.com/mattmoodie
Professional GitHub: github.com/matt-project-q
www.projectqpub.com/write
#github #AI #opensource
Why write for Project Q?
Books-as-code: our open-source tool chain allows us to treat books as code. This makes your content flexible, trackable, and accountable. You're not tied to any platform, so work the way you want.
www.projectqpub.com/write
#latex #pandoc #GIMP
Why write for Project Q?
Developer-friendly workflow: we use GitHub and Markdown by default. We work with the tools you're comfortable with to give you a distraction-free writing experience. If you want to write on another platform, we can work it out.
www.projectqpub.com/write
#markdown #github
Why write for Project Q?
25 years' experience: we're run by an editor with over 25 years' experience. Matt has edited hundreds of technical books and overseen hundreds more. You'll get all the support you need.
Have a look on LinkedIn:
www.linkedin.com/in/matthewmo...
www.projectqpub.com/write
Why write for Project Q?
We're building our launch list, so are starting with a series of posts explaining why authors should write for us.
In brief:
- 25 years' experience in technical publishing
- Developer-friendly tools
- Books-as-code
- Run by technical people
www.projectqpub.com/write