German government giving up on digital reassembly of shredded Stasi files, resorting to manual puzzlings

When it collapsed some 30 years ago, the GDR (East Germany) was the most invasive surveillance state ever attempted, employing one out of every 60 people in the country to snitch on the rest, running dirty tricks campaigns against anyone deemed "subversive."

When the GDR began to crumble, the Stasi gained a new preoccupation: destroying millions of files so that the identities of their collaborators and the crimes they were complicit in would never come to light. The Stasi burned out their shredders, overwhelmed their furnaces, and eventually resorted to tearing sheets of paper up by hand.

Since 2008, a €8m project to piece those 1.5 million shredded documents back together has deployed the best in computer image-processing to recover the files, but they're stalled out and the government is ending the project.

The hand-shredded ones (which have extremely similar shapes thanks to the pages being folded prior to ripping) are especially hard to recover. Many of them have been recovered by hand-sorting at a federal refugee agency in Bavaria.

The ePuzzler works by matching up types of paper stock, typewriter fonts, or the outline of the torn-up page. It has struggled with handwritten files that were folded before being torn, leaving several snippets with near-identical outlines.

The ePuzzler has also required human assistance to feed in paper snippets and check over the completed jigsaw puzzle, further slowing down the process.

"We currently don't have a scanner that we can work with," said Hovestädt, adding that her agency was hopeful that technological progress would allow the archive to resume reassembling destroyed records this year.

Stasi files: scanner struggles to stitch together surveillance state scraps [Philip Oltermann/The Guardian]

The Stasi's Tiny Torn-Up Analog Files Defeat Modern Digital Technology's Attempts To Re-Assemble East Germany's Surveillance Records [Glyn Moody/Techdirt]

(Image: Uhlemann, Thomas, CC-BY-SA)