HomeKnowledge base › Forensic reconstruction
Technology

Forensic reconstruction of shredded paper: what is actually possible?

The question regularly comes up during audits: “but can someone puzzle it back together?” It is not an academic question, because there are offices that make their profession of reconstructing shredded documents. What they can do depends on one thing: how small are the shreds? A tour through history and science.

The Stasi archives: the original version

When the Berlin Wall fell, Stasi staff tried in November 1989 to tear up their archives by hand. 16,000 bags of paper were left. Since 1995 work has gone on to reconstruct them. Until the late 2010s mostly by hand, with armies of students and pensioners sorting shreds by colour, font and tear line. Result after decades of work: a few thousand pages reconstructed from billions of shreds. Important detail: this was not machine-shredded material but hand-torn documents with shreds the size of beer mats.

The DARPA Shredder Challenge (2011)

In October 2011 the US research organisation DARPA put up a 50,000-dollar prize for a team that could reconstruct five strip-cut shredded documents. The winner, the team All Your Shreds Are Belong To U.S., used computer vision to cluster shreds on pattern features (font, colour, tear line) and then manual refinement. Time to completion: 33 days for 10,000 shreds across five simple documents.

The scientific conclusion: strip-cut documents are reconstructible with enough compute and human time. Cross-cut documents at P-3 or P-4 produced much longer times and, in most pilots, were not completed within a reasonable period. Above P-5 the possibility of reconstruction effectively disappears in practice.

Mathematically: why smaller particles are exponentially harder

Reconstructing a torn document is a combinatorial problem. How many possible orderings are there for N shreds? At N=20 it is billions. At N=200 it is numbers with more than 300 digits. A document at P-2 consists of about 30 shreds; P-3 about 60; P-4 200; P-5 1,500; P-6 4,500. The difference between P-3 and P-5 is mathematically the same step as between a crossword puzzle and breaking a random 80-bit key.

Read our article on cross-cut versus strip-cut for the exact mm² classifications.

What helps the reconstructor?

What helps the owner (you)?

P-5 is not overkill for a business with customer data. It is the threshold below which reconstruction becomes mathematically unfeasible and your audit story holds up.

Realistic threat models

The question is not “could someone reconstruct it in theory?”, but “who would put in the effort?”. For most businesses the threat model is a curious third party with brief access, not a state actor with an unlimited budget. P-4 keeps curious third parties well at bay. P-5 also puts motivated criminal reconstruction attempts beyond reach.

For lawyers, notaries, banks and care providers the picture differs. There the threshold is not only technical but legal: a confidentiality professional must be able to demonstrate that reconstruction is not reasonably possible. P-5 is the minimum there, with a DIN mention on the destruction certificate.

What if your document is in the street?

If a bag of shredded paper accidentally ends up in the street and someone takes it, what they have is only a data breach if reconstruction is reasonably possible. For strip-cut: notifiable. For P-5: in almost every case no notification, because reconstruction is not feasible within a reasonable time. The AP will look in concrete cases at particle size and the nature of the data.

Want to clear the mathematical threshold?

Our mobile shredders run by default at DIN 66399 P-5. The exact level is stated on the certificate, and with it the underpinning of your security measure.

Request a quote

If in doubt, request a quote via desnipperaar.nl. We look together at which level fits your documents.