TalkOrigins Archive Modernization Underway
Back around 2000, Brett Vickers made some changes in his life and realized that the maintenance of the website and resource he created, the TalkOrigins Archive, was not going to be a focus any more. Brett asked me to serve as the site maintainer, and I agreed. It is something that I’ve tried to serve as a good custodian for over the years, but I haven’t had a lot of extra bandwidth to deal with the technical debt.
The improvements in generative AI, though, have made a path to actually do something about the technical debt issue, and allow a broad modernization effort. When I say ‘broad’, I mean working toward making the whole suite of domains that come under the TalkOrigins Foundation umbrella work together in a better fashion. Many people don’t realize that the TalkOrigins Archive is not intended to be a unitary thing that is entirely separate; it is complemented by the focus seen in other domains. There is the talkdesign.org site, serving as a direct counter to ‘intelligent design’ creationism argumentation. Then there was the http://pandasthumb.org blog, as the Foundation resource aimed at addressing events and issues needing a more rapid response than the slow-twitch process for FAQs had for the TalkOrigins Archive. The Panda’s Thumb is hosted by Reed Cartwright via his University of Arizona lab and Github, but other Foundation domains are served by a host I have full access to for development and system configuration. There’s a domain owned by the Foundation, http://evolutionnews.net, that differed from a domain name used by the Discovery Institute by the extension, and was used to pull an effective April Fool’s prank a couple of decades ago. EvolutionNews.net is now being repurposed as a place to make ‘evidence docket’ materials that will aid people who want to make responses to particular anti-evolution writings or video. The newest addition to the Foundation project domains is http://evo-edu.org, for ‘Evolution Education’, as a place for education (either structured or auto-didactic) in topics in evolutionary biology, evolutionary computation, and ecology, with an emphasis on web apps for interactive use and learning. It features a vareity of Javascript implementations of classic web apps related to population genetics, population dynamics, evolution education, and evolutionary computation, with an emphasis on providing guidance for use and learning pathways based on a knowledge-graph driven representation of source materials. Each web app has pages for the app itself, curriculum guides, and education standards alignment. It is also the initial place with a rollout of a ‘Notebook’ concept, as a compendium of knowledge-graph based materials collated from sources, with source tracking and links into generative AI based mentoring. The Evo-Edu.org site is aimed at effective communication of the scientific content under these fields, rather than the more confrontational stance taken against anti-evolution and other forms of science denialism seen in other Foundaiton domains.
The first and most major step in the process was to move the TalkOrigins Archive away from the HTML+CSS theming of the original site, which was aimed at desktop web browsers of the 1990s. The new site theming is done in responsive design, making it easier to use from mobile devices, while still doing a good job on desktops. That is done, and the site reflects the new theme now. A major feature change is that the TalkOrigins Archive now has Javascript and routing rules to support a multi-ligual presentation. This change is in progress: an initial set of four foreign languages will be the short-term target, and a somewhat larger set will be the long-term target. The initial set of languages is English (our default), plus Spanish, French, German, and Portuguese. The translation process is automated via generative AI, so until there is review of the translations, each will carry a warning banner with a disclaimer about the process and how to enter feedback about problems (via the new TalkOrigins Archive Feedback app). Since only about a quarter of the site’s content has been translated so far, if a linked page is not available in the selected language, the English-language version is served with a banner in the target language saying that the content is queued for translation but not available yet.
There is also a new feature, ‘TOA-Lite’, a downloadable representation of the static content of the TalkOrigins Archive, with an included Javascript site-search capability. This allows you full access to static pages and a search function for those times when internet connectivty could be problematic.
For the online version, that is a multi-corpus site-search capability that can allow searching content from a variety of Foundation sites, including TalkOrigins, TalkDesign, and PandasThumb. Which corpora to search can be selected.
So far I’ve mentioned feedback (.fb) and search (/search), but there is another dynamic app of note, the workbench (/workbench). The workbench is meant to queue suggestions for content, and provide a means to research and update content. It features a multi-role authentication system that uses email verification. This is where the bulk of change to content for the Archive should happen going forward.
The TalkOrigins Archive has suffered a couple of hacking incidents in the past. While hacking can’t be stopped entirely, the new method of development and hosting should reduce the pain that can be caused. Development happens on another machine entirely, and every element of hosting on the production server is now implemented in Docker containers, with each group of static files getting its own server container, and every web app or endpoint gets its own separate Docker Compose stack. This is intended to reduce the ‘blast radius’ of any hacking/cracking, and once recognized, the affected container or containder stack can be deleted, and a new version promoted from the development host. There shouldn’t be a simple path to an adversary gaining control over all the sites on the server through a single vulnerability. This re-organization will also make it easier to have a handoff of the sites; the time to re-deployment on a Docker Compose capable host should be just a little more than the actual data transfer time.
The TalkDesign website has also had the initial phase of conversion to a modern responsive design theme applied, and is available now. The translation queue for TalkDesign pages will be served after the TalkOrigins queue completes.
There are a great many other features to be discussed, but this is the core news on the Foundation domain modernization effort right now.
