Journals of a Time Traveller

Thetford Forest Archaeology Portal & Local History Hub  - Live website.

I’ve lived through plenty of chapters—more than I suspect most people do. In one, I was a licensed radio amateur (G0AGP), keying in Morse code, studying trans-equatorial propagation and the 11-year solar cycle on the 10-metre band. When I had to move on from that world, I needed new learning curves; I needed to seek the patterns beneath the surface.

Or, as it turned out, on the surface of disturbed soils. I found myself drawn to recognising and searching for prehistoric struck flints. That, too, was a lesson in pattern-matching—a growing fascination with siliceous beauty and conchoidal fracture. I would scan the ground, filtering through the chaos of natural stones for the telltale signs of human intervention: bulbs of percussion, ripples, striking platforms, and flake scars. Even with my eyes closed, I could feel the distinct conchoidal curve of a flake struck by a knapper millennia ago. Soon, I saw how these patterns could scale up, mapping themselves onto entire landscapes. With no formal training at the time, I developed my own field techniques and methodology.

Then, computers arrived. I quickly grasped spreadsheets and databases, seeing instantly how they could help me organise and arrange these spatial patterns. Next came dial-up modems. The driving catalyst for buying a 56k modem wasn't to browse, but to build—I wanted a website to share my archaeological data.

That pursuit unlocked a new obsession: web-building. A metal-detecting webmaster of the era once joked on his links page that I would radically transform my website every few weeks. He wasn't wrong. It was a relentless search for a sense of resolution, found only in error-free perfection. In the strict syntax of HTML and XHTML 1.0 Transitional, I found a new universe of patterns.

Twenty-five years ago, web-building was a radically different landscape. There were no AI agents, no VS Code extensions, and no Git. Software packages for building websites were bought on CD-ROM; professional tools were prohibitively expensive, while the cheaper alternatives generated notoriously horrible code—bloated, messy, and unfriendly to browsers.

As an amateur webmaster working on a shoestring budget, I decided the only elegant solution was to learn how to write the code myself. The gold standard at the time had recently shifted from HTML 4 to XHTML 1.0 Transitional. I mastered it, and even taught it to my young daughters. By hand-coding XHTML in basic text editors and early syntax-highlighting code editors, I built a 120-page digital estate—a cluster of interconnected micro-sites. I even wrote a tutorial over twenty years ago on how to code by hand.

Timeline

• 1999 – 2006: Writing XHTML, building and maintaining my websites.

• 2008: Interests had moved on. I stopped paying the server bill and the websites were erased. However, the Wayback Machine had archived copies on web.archive.org. Many of the images were lost, but much remained.

• 2026: A conversation with an LLM triggers a decision to attempt a total restoration and revival of my old sites.

Step 1: The Extraction (Excavation)

Gemini AI guided me to the Archivarix service. All I had to do was select the range of dates over which the Wayback Machine had originally captured the websites. (A little tip for anyone else trying this: choosing a wider date range might have captured a few more of the original images). I decided to pay the $10 USD fee (around £7.38 GBP)—a service well worth paying for. A short while later, the ZIP file was ready.

I operate from a modest Acer Chromebook. Initially, I had some issues trying to extract the files. That process would have been much easier if I had decided to reinstall the Linux (Crostini) environment right away. I have plenty of Linux experience using a variety of distros years ago, and I had used Crostini on this Chromebook before, but right then I was reluctant to set it up again. That was a wrong move. Instead, with Gemini assisting me, I bypassed the extraction issue for the moment and took the archive straight to the next stage.

Step 2: Upload to Live Status (Conservation)

Twenty-five years ago, you either used one of the awful, ad-funded free servers like Yahoo! GeoCities, or you paid for hosting and uploaded your files via FTP. Things have moved on. I prompted the AI to help me find a free home for my website, and found myself at the front door of Netlify.

Initially, I used an AI interface to help process the raw archive data. I fed the archive into the workspace, watched the AI perform its magic, and suddenly the core structure was back online. Next, I updated the URL, choosing https://thetford-forest-archaeology.netlify.app/. Naturally, twenty-five years ago, the .app top-level domain didn't even exist. But boom—there it was.

Restoration of a Broken Website

My websites were once again live on the World Wide Web, no longer confined to a dusty archive. Yet I had massive amounts of work to do. The Wayback Machine had graciously saved the text, but many of the image files were gone (specifically those that lay too many clicks away from the index file). I managed to retrieve some of these from other old repositories, my old Flickr account, and alternative archive sites, but dozens remained lost forever.

I had a lot of tidying up to do. Furthermore, I had hand-coded these 120+ files in XHTML 1.0 Transitional, only to find in 2026 that the world had moved on to HTML5. My pattern-matching tendencies wouldn't tolerate such a structural mismatch. As a hyper-systemiser working in partnership with Gemini and Claude, it took me about three weeks of editing to rationalise everything down to 95 clean web pages, alongside their supporting JPEG, CSS, and XML files.

Step 3: The Coding (Reconstruction)

I finally did what I should have done at the start: I enabled the Chromebook's Linux feature and updated the system. My Linux commands were rusty, but I had my AI friends to guide me. I generated a website folder within the Linux partition and, using Bash commands guided by Gemini, I extracted the compressed files there. Now I had a full local clone of the website sitting in my Chromebook's local storage.

A tip for Chromebook Linux newcomers: The Linux folder partition won't automatically appear in the native ChromeOS Files app unless you have opened the Linux terminal at least once during your session to mount the container.

Next, I installed Flatpak via the terminal, using it to download Visual Studio Code (VS Code) as my environment of choice. While there are countless AI extensions available for IDEs, many require paid tokens or API credits. To keep this project entirely free, I utilised a simpler workflow: I would copy my code blocks, paste them into the free-standing browser interfaces of Gemini or Claude for structural analysis, and paste the corrected code straight back into VS Code. Money saved.

What to do with the websites themselves? Four distinct components were salvageable:

1. Thetford Forest Archaeology

2. Portuguese Thetford

3. Wesley's Metal Detector Finds

4. How to HTML! (My old coding tutorial)

I didn't want to completely modernise everything; I wanted to preserve their history while ensuring modern browser compatibility. My approach varied across the four sites:

• Thetford Forest Archaeology: A hybrid approach. I converted the skeleton to HTML5 but maintained the period-accurate CSS and styling of the early 2000s. I updated all .htm extensions to .html. This was always my prized showcase website—the original motivation for me learning to code.

• Portuguese Thetford: I let Gemini take the lead on rewriting the CSS. I didn't just sit back and blindly accept prompts; I meticulously updated the structural tags, but permitted the AI to modernise the responsive presentation layer while keeping the original text fully intact.

• Wesley's Metal Detector Finds: This old site is a classic end-of-the-20th-century creation, complete with an animated GIF and a tiled background. I chose to preserve this one completely untouched, keeping the original code exactly as it was, save for updating the broken hyperlinks.

• How to HTML!: I had entirely forgotten that I wrote this! Originally coded in strict XHTML with .htm extensions, I left the core tutorial preserved, simply removing old CSS references to background graphics that have been lost to time.

To bridge the personal knowledge gap between XHTML and modern standards, I provided Claude with a link to my old "How to HTML!" tutorial. I asked the AI to use my own 20-year-old text to form the basis of a personalised transition course into HTML5. It felt as though an earlier version of me from 2005 was reaching forward across the decades to teach the 2026 version of me how to code for the modern web.

Isn't that clever? Perfect for a Time Traveller.

Step 4: The Relaunch (Publication)

To handle updates without relying on automated AI build tools, I shifted to Netlify’s direct manual folder deployment feature. This method is entirely free and bypasses the need for automated build credits.

The deployment process is incredibly elegant. It uses a local checksum mechanism: I simply select the entire website folder from the Linux directory on my Chromebook. The interface might flag that hundreds of files are being processed, but it instantly identifies the precise files that have changed and uploads only the modifications. Updates are seamless, uncomplicated, and incredibly fast. The key is simply maintaining that perfect, updated clone within the Linux environment.

I am pleasantly surprised by how straightforward it has become to maintain a web presence today. With the files live, all that remains is some foundational SEO work—submitting updated XML sitemaps to the search engines and tidying up metadata.

Digital Archaeology.  AI Image prompted by Gemini to illustrate this post. An archaeology website that has itself, become a digital artefact.

I just checked Google Search. The index appears using the string: Thetford Forest Archaeology. May the SEO prosper, and a new generation of archaeology students learn from how an amateur did it twenty years ago.