The Envelope Was the Defect

On September 2, 2016, the company announced a global recall of 2.5 million phones after batteries began venting, scorching nightstands and, in a handful of cases, cars. The number Samsung led with was the defect rate: fewer than 0.1 percent of units. That figure was, as far as anyone has ever established, accurate. It is also the least useful number in the entire incident, because the institution that mattered most does not compute in percentages.

The U.S. Department of Transportation issued an emergency order on October 14 banning every Note 7 from every flight to, from or within the United States — carry-on, checked, cargo, all of it. Airlines were required to announce the prohibition before boarding, and to deny boarding to any passenger who wouldn't give the phone up. A consumer electronics product, by name, on the pre-flight safety announcement. Regulators do not do expected value. They do categories, and once you are in the hazard category, your defect rate is a footnote nobody reads.

The engineering is where it gets genuinely interesting, and it is not the story most people remember.

Samsung's first fix was to swap the battery supplier. Units from Samsung SDI came out, units from Amperex Technology went in, replacement handsets shipped within weeks. The replacements caught fire too. A second recall landed October 13; the model was killed outright.

When Samsung published its findings in January 2017, with Underwriters Laboratories and other outside labs, the result was two entirely separate defects. Battery A, from SDI: the pouch was deformed enough to impinge the negative electrode, letting the electrodes touch. Battery B, from ATL: inconsistent welding left burrs on the positive tab — up to 80 microns — that pierced the insulating tape and the separator underneath.

Two independent manufacturers. Two unrelated failure mechanisms. One enclosure.

The industry filed that under "battery quality control," as though the batteries were the variable. The strongest reason to think otherwise does not come from the coincidence at all. UL's own conclusion pointed at the thin separators and the high energy density — properties not of a supplier's process but of the space the supplier was told to fit into. That is a direct finding about the envelope, from an outside lab, and it holds whether or not the two failures turn out to share a cause.

Take the competing explanation anyway, because it is the first one an engineer reaches for. ATL's cells were ramped into volume production in weeks, under recall conditions, to replace a supplier that had just detonated. Welding burrs are a process-control defect, and a line ramped at emergency speed is the textbook way to manufacture one. So the second failure may well share a cause with the first — and the shared cause could be schedule rather than geometry.

That gets nobody off, because schedule and geometry are withdrawals from the same account. Both are line items in one negotiation, and both were won by teams that each got exactly what they asked for. The two-vendor pattern is corroboration, not proof. What the envelope did to the margin is the finding.

The envelope came from the spec sheet. A 3,500 mAh cell, a curved edge display, an iris scanner, IP68 waterproofing, fast charging, and a chassis thin enough to compete with what Apple was about to announce. Every one of those is defensible. Every one is marketable. Every one takes millimeters out of the same budget, and waterproofing in particular means sealed — no vent path, no user-swappable pack, no way for a cell to fail gracefully because there is nowhere for the failure to go. It is also the one item on that list that removes the owner from the loop entirely: whatever the phone decides to do at 3 a.m., you cannot open it and take the battery out.

Nothing on that list is a defect. What's missing from the list is margin, and margin has never once appeared on an org chart. There is a team that owns display, a team that owns battery, a team that owns waterproofing, a team that owns the ship date. There is no vice president of slack. It gets consumed the way a shared server gets consumed — everyone's request is individually reasonable, and the thing that dies is the property that only existed in the gaps.

Samsung's answer was an eight-point battery safety check: more inspection, more X-ray, more charge-discharge testing, more disassembly. All of it sensible. All of it detection. You can inspect for a burr. You cannot inspect for the fact that a burr has nowhere safe to be.

The company recovered, which is the part the case studies celebrate. Roughly $5 billion in losses, $14 billion off the market cap in a day, and the Note line survived anyway, because a phone is a subscription to a supply chain and nobody switches ecosystems over one bad year.

Which is where the aside from paragraph three turns out to be the whole argument. Regulators do not do expected value; they do categories. Margin is the one property no internal owner can price, because every internal owner is optimizing a named line item and margin is by definition what is left after the named items are served. Nobody inside the firm can put a number on it, so it never gets priced from inside. It gets priced from outside, by something that reasons in categories instead of percentages — something willing to put a product name in a pre-flight safety announcement and bar it from the national airspace.

For one product cycle in late 2016, the hazard category made slack worth more than millimeters. Then the last Note 7 came off a plane and the category lapsed.

Every phone in your pocket right now is thinner, denser, more sealed and faster-charging than a Note 7 was. In ten years nobody has written a ninth point for the spec sheet. The eight-point check survived because it is internal and cheap. A ninth point would have to be imposed, and nothing has been in a position to impose one since October 2016.