Why a Damaged QR Code Still Scans

A parking-meter QR code near a Chicago bus stop has a rideshare sticker slapped across one corner and a coffee ring bleeding over another edge. Roughly a quarter of the pattern is unreadable to the eye. Point a phone at it anyway, and it scans in under a second.

That isn't luck. Every QR code carries a built-in repair kit, extra data woven into the grid specifically so a scanner can rebuild whatever gets torn, smudged or covered. The math behind it wasn't designed with parking meters in mind at all. It was built to keep signals readable across millions of miles of empty space.

The technique is called Reed-Solomon error correction, and according to Denso Wave, the Japanese company that invented the QR code, the underlying method "was originally developed as a measure against communication noise for artificial satellites and planetary probes." The same math also keeps a scratched music CD from skipping, and it's a cousin of the trick that lets a phone hold a Bluetooth connection through interference that would otherwise garble the signal entirely.

How a QR code rebuilds what's missing

A QR code doesn't store your web address or Wi-Fi password just once. It stores a mathematically derived backup of that data, spread across the whole grid instead of clustered in one fragile block. Denso Wave breaks it down in codeword terms: a codeword is the basic unit of a QR code's data area, and one codeword equals 8 bits. Encode 100 codewords of data at the highest correction setting, and the code adds another 100 codewords of backup. That's 200 total, of which up to 50 can be lost or wrong and the code still reads correctly. Denso Wave calls that a 25% error correction rate, corresponding to what's known as error correction Level Q.

Cover a corner with a logo, let a sticker peel halfway across the pattern, or scratch out a chunk with a car key, and the scanner leans on that redundant data to reconstruct what's gone. It isn't guessing at pixels. The algorithm calculates exactly which data was lost and fills it back in, the same way a satellite downlink pulls a clean signal out of static.

How much damage can a QR code actually take?

Every QR code generator lets you choose from four error correction levels, and the gap between them is stark:

Level M, recovering around 15%, is what most generators reach for by default. It survives ordinary wear: a folded flyer, a grease spot on a takeout menu, without bloating the code into a denser, harder-to-scan grid. Level H is overkill for a business card, but it's close to standard practice anywhere a logo gets stamped into the middle of the pattern, because a logo destroys real data and something has to make up the difference.

Why do some QR codes have a logo in the middle?

That's the same mechanism, deliberately triggered. A logo blanks out a cluster of the code's modules, which looks identical to damage from the scanner's point of view: data that should be there simply isn't. Choose a high enough correction level before adding the logo, and there's enough redundant data left over to absorb the gap without breaking the scan. Skip that step, and the code either fails outright or only scans from certain angles, which is the most common reason a QR code that looked fine on a laptop screen stops working once it's printed on a poster.

Is a damaged QR code always safe to scan?

Readability and trustworthiness are two different questions. Error correction only rebuilds the data that was encoded when the code was first generated. It has no way to tell if someone has pasted an entirely different QR code over the original one. McAfee has documented scammers doing exactly that at parking meters, layering a fraudulent sticker over a legitimate code to redirect drivers to a fake payment page. The error correction that makes a beat-up QR code so reliable is the same feature that makes a swapped-in fake one scan just as cleanly. Nothing about the technology can tell a torn code from a tampered one; that part is still on the person holding the phone.

None of which explains why the pattern exists in the first place. Masahiro Hara, an engineer at Denso, a Toyota parts subsidiary, was asked in 1992 to build a barcode that could hold more data than the striped kind scanners were already struggling to track fast enough on assembly lines. Two years later his team had the black-and-white grid, and buried in it, the Reed-Solomon backup that means a code born to track car parts can now survive a coffee spill on a parking meter three decades later — the same trick that also keeps a crowded Wi-Fi signal from collapsing into noise.