Web archiving was designed around documents. It handles text-based sites remarkably well: a crawler fetches the HTML, the stylesheet and a handful of images, and what comes back is recognisably the page. Comics break that model, because on a comics site the HTML is a wrapper and the substance is a large binary file that the crawler treats as an asset. When the asset is missed, what survives is a perfect copy of an empty frame.
What web archives capture well
Crawlers are good at anything reachable by following ordinary links from a starting address. On a webcomic site that means the front page, the section indexes, the prose pages — about, news, links, the update log — and the top layer of any straightforward directory tree. If a site had a real text archive listing, there is a decent chance it was captured, and those listings are often the best surviving evidence of what a strip actually contained.
Archives also capture addresses even when they fail to capture content, and this is underused. A capture index for a dead domain will list every path a crawler ever saw, with dates. That index alone reconstructs a site's structure, dates its rebuilds, and identifies which sections existed in which years — before you open a single page.
What they capture badly
Five failure modes account for most of the loss.
- Deep sequential runs. A crawler visiting a page-turner archive follows the "next" link a certain distance and stops. Early pages of a long strip are usually well covered; the middle is patchy; the recent end is often absent.
- Navigation built from scripts. Where the next-page link was generated by JavaScript, or was an image button wired to a script function, a crawler may find no link to follow at all and the archive ends at page one.
- Large images under a size ceiling. A high-resolution page scan can exceed the limits a crawl applies to individual assets, so the page is stored and the artwork is not.
- Anything behind a query string. Galleries and forums generate effectively unlimited addresses from parameters, and crawlers throttle them hard. This is why the community layer is the least-preserved part of these sites.
- Downloads. Files offered as issue downloads were frequently linked from a page the crawler reached but never fetched themselves.
The compound effect is that an image-heavy archive can appear preserved and be substantially empty. Always check several capture dates rather than one; artwork missing from a 2011 capture is often present in a 2008 one.
The metadata problem
Even where everything was captured, the resulting archive is close to unsearchable. As covered in site anatomy, headings were images, navigation was images and the comic was an image, so a fully preserved page may contain no machine-readable words except a copyright line. Nothing in the file records the title of the strip, the name of the arc, the publication date or who drew it.
That is a preservation failure with a trivially cheap fix that nobody applied: descriptive alternative text on every image, a real text heading above each page, and a publication date in the page itself. Any one of those would have made these archives findable twenty years later.
A practical routine for creators
If you drew a strip during this era, or you are drawing one now, the following takes an afternoon and is the difference between an archive that can be moved and one that can only be lost.
- Keep the masters, not the web exports. The 300-dpi scan or the layered working file is the preservation copy. The 600-pixel-wide compressed version on the site is a derivative, and it is the only thing a web archive will ever hold.
- Store them in a flat, sortable structure with meaningful names. One folder per strip, files named with a zero-padded sequence and a date. Avoid spaces, capitals and anything that ties a filename to a specific site layout.
- Write a plain-text index alongside them. One line per page: sequence number, arc, publication date, original address, contributors. This is the single most valuable artefact you can produce and it costs an hour.
- Prefer open, well-documented formats. Lossless formats for line art, widely supported lossy formats for painted work, and plain text for the index. The Library of Congress maintains detailed assessments of format sustainability that are worth ten minutes of reading.
- Keep three copies in two places. Local disk, external drive, and one off-site copy. Optical media burned in the 2000s is not a backup any more — read those discs now, while drives still exist that can.
- Request a capture before you move or close anything. The Wayback Machine accepts on-demand submissions of individual addresses. Walking your own archive and submitting each page is tedious and it works.
- Leave permanent redirects behind when you restructure, and keep them indefinitely — see URL conventions for why this matters more than anything else on this list.
Bit rot and the media question
The physical side is more urgent than most people assume. Recordable optical discs from the mid-2000s are now well past the point where dye degradation shows up, and the failures are silent — a disc reads fine for years and then returns errors on a handful of sectors, which in a compressed image file means the whole file. Consumer hard drives left unpowered for a decade fail at meaningful rates too.
The Digital Preservation Coalition's handbook covers this properly and is written for people without an institutional budget. Its central point applies directly here: preservation is an activity, not a storage location. A copy nobody checks is a copy nobody has.
Where this material belongs
There is a real gap between how well the print end of independent comics is collected and how badly the web end is. University and museum collections take small-press and self-published comics seriously — the Billy Ireland Cartoon Library & Museum at Ohio State is the best-known example, and several other research libraries maintain substantial comic-art collections. Web-native work from the same creators, in the same years, is far less likely to have been deposited anywhere.
Cartoonists who ran strips in this era frequently assume their work is not of interest to a collection. It generally is, particularly with the physical convention material, the printed samplers and a documented digital archive alongside it. The worst outcome — and the common one — is a folder of files on a dying drive belonging to somebody who has stopped thinking about it. Everything on this site about how these archives die is ultimately an argument for acting during the years when nothing appears to be wrong.