date = 2024-01-03T12:00:00Z
+++
-TODO
+As we enter into 2024 I thought we'd once again look back over the past year to
+see how the project has evolved, in terms of both the core platform and the
+supporting working, including the current funding seen throughout 2023.
### Project Funding
- Book, and shelf-book, sorting made accessible with improved UX
- Updated underlying framework to Laravel 9
- Updated code handling to CodeMirror 6
+- New Languages:
+ - Finnish
+ - Norwegian Nynorsk
+ - Uzbek
- API Additions:
- Roles
- Image gallery
- Added more customization-focused partial views
+### Website Usage & Audience Reach
-### Videos on PeerTube and YouTube
+Looking at the [BookStack website analytics](https://analytics.bookstackapp.com/bookstackapp.com)
+we can compare activity between 2022 and 2023 to understand year-relative growth
+in website usage and audience reach:
+
+
+
+Across visits and page-views, we can see a general average growth between 10% and 20%.
+This is a bit lower than the 45% growth last year, but is not bad considering I've done almost no
+additional outreach or marketing content pushes this year, so a ~15% growth indicates
+a nice steady natural rise.
+Last year we started the year very strong with [reaching the top of Hacker News](https://www.bookstackapp.com/blog/9000-stars-and-the-effects-of-hacker-news/),
+which then boosted activity for a good while, but there hasn't really been any spikes or visitor boosts
+from anything similar in 2023.
+Maybe I need to put more active effort into marketing in 2024 although it's not clear if there's a reason to do so,
+and slow & steady growth has it's own benefits of being more manageable and requiring less resources.
+To be honest, as long as we're trending in the right direction, I'm happy.
+I'd rather not chase numbers for the sake of it.
+
+### Videos on YouTube & PeerTube
+
+Throughout 2023 I've kept up publishing videos, one for each feature release with extra videos
+here and there to provide guidance and/or insight for the project.
+Here are the [BookStack YouTube channel](https://www.youtube.com/c/BookStackApp) general stats for the year:
+
+
+
+You may notice the estimated revenue shown there. Earlier this year the channel met the requirements for monetization
+so I'm now able to earn a little bit of bonus revenue from these videos.
+Money isn't really the target though, as it's very far from covering the effort in making this content.
+These videos are really about building this extra resource and community hub, while providing
+a means for me to celebrate and emphasise work being done for the project.
+Folks dunk on YouTube comments as being a toxic place, but generally the feedback I get is
+wonderfully wholesome which feeds my motivation for the project.
+
+As you can see from the other stats, we've generally had good growth overall, even as
+I recently promote the YouTube content less in favour of the [PeerTube alternative I set-up](https://www.bookstackapp.com/blog/bookstack-on-foss-video/).
+I've been meaning to play around with shorts, just to provide quick summaries for releases, but my recording flow & format doesn't really make this easy.
+
+The usage of the [BookStack PeerTube channel](https://foss.video/c/bookstack/videos) has gone well so far.
+Views and usage is predictably a lot lower than our established YouTube presence, but I've really liked the fact
+we're not forcing folks to view ads, or submit to Google's privacy requirements, when sending visitors to video
+content from the BookStack site. Plus it's nice to be a part of the open social fediverse.
+
+On the subject of video and YouTube, in August [I spotted BookStack](https://fosstodon.org/@danb/110956729212142440)
+used within a Linus Tech Tips video, which was pretty awesome to see.
+From what I could tell, it's likely an instance set up by Floatplane which LMG labs were also
+using at that time.
+
+### AI Impact on BookStack
+
+AI has been a massive subject for this year as LLMs & image generation has exploded in use.
+Personally, I've found LLMs to be quite useful in some select cases, although I'm still conflicted on the
+the morality of the training process and their overall impact on the world.
+They seem to be becoming a bigger part of society though, so it's probably important to keep tabs on their use
+along with their co-existence and potential benefits for projects like BookStack.
+
+In regard to LLM usage in BookStack, earlier this year I came across [Danswer](https://docs.danswer.dev/introduction)
+which I built an integration for which allowed Chat-GPT querying of your BookStack content.
+I talked about this more, along with a demo of this integration, [in my video here from about 4:28](https://foss.video/w/mB67n8JBBHb9mSMYUM5DED?start=4m28s).
+This is probably the most valuable usage of LLMs with a system like BookStack.
+I've also heard some other good ideas like using LLMs to generate out templates, or base page content.
+Another neat idea is using image generation to create book and shelf covers.
+Most of these desired should be possible via existing integration methods for the most part although
+as LLM/AI options, and the desired usages of them, evolve we may start to see areas where we can
+specifically provide interfaces to help these kinds of tools.
+
+Outside of the core platform, we'd also seen AI's impact in our issue management, as well as in
+external news content for BookStack, both of which reflects the downsides of AI:
+
+On issue management, I've had a couple submitted that were clearly filled out using an LLM.
+I don't mind a LLM used to help where it adds value, to assist user written content.
+Unfortunately though, in these cases, it has simply been used in feature requests to back up the request
+with many paragraphs of generic text that has barely a grasp of what BookStack is.
+These cases are a little frustrating as it just serves to waste my time, in attempting to extract any underlying intent
+and value. I'm having to somewhat reverse the content to their original prompt.
+The questions on our feature request form are to understand the value to users, not those imagined by a text generator.
+
+In regard to news content, BookStack [was featured on Laravel news](https://web.archive.org/web/20231109033107/https%3A%2F%2Fp.rizon.top%3A443%2Fhttps%2Flaravel-news.com%2Fbookstack-documentation-wiki-software) which was awesome as a long-term reader of the site myself.
+Unfortunately when reading it was quickly clear the article was LLM generated since many of the features detailed
+reflected a limited grasp on understanding BookStack while some of them were just wrong & misleading.
+Due to these issues, I didn't feel I could celebrate or share the article as that would just propagate this incorrect content.
+Instead I emailed the site to notify them of the inaccuracies, with advised changes, although I had to follow that up [with a Tweet](https://twitter.com/bookstack_app/status/1722457925900280074) a few days later to prompt some action. It was then promptly updated using the text from our website.
-TODO
### Hacks Site?
http://localhost:1313/blog/hacks-on-the-site/
-### Website Usage & Audience Reach
-
-TODO
-
-
### Going into 2024
TODO
projects / website / commitdiff
