The folks at 99% Invisible podcast must be scrambling, and their scrambles are totally worth it.

A few episodes ago, there was a brief digression where Roman Mars said something about the toilet paper supply chain being mostly automated, and a few other things that made me suspect they were preparing an episode about how America receives its favorite under-appreciated fiber product. Now, it’s out, and (as you’d expect) it’s pretty good.

They don’t go into the details about how the toilet paper supply chain is managed, but as usual for a 99% Invisible episode, there are tons of chunky nuggets to flush down. The episode is here: [99pi]

I don’t think it’s cool to plug some piece, and then spoil it by trumpeting the most fascinating details, so I won’t discuss anything from the episode here, other than to say that it’s worth a listen and it ends with a really interesting public service announcement about those “flushable wipes” – as anyone with a septic system knows, they’re not. But I did not realize that the downstream effect of the flushable wipe is much worse for municipal sewage management systems. The 99% Invisible folks didn’t go into tampons and condoms but, briefly, those are not flushable either. Just ask the septic tank crew who had to go out with pry bars and axes to break apart a frozen floating mountain of condoms in -5F (-20C) at a place I know near here, a few years ago. They were vocally unhappy and I don’t blame them a bit. The “flushable” bit was some ‘inspired’ marketing – i.e.: a lie – without which the darned things probably would have never sold at all, for good reason.

We don’t see how much of this is a matter of culture because, for most of us, things are the way they are when we were kids, so “it’s been like this my whole life.” I grew up in an era of ‘facial tissues’ – Kleenex – but my dad grew up in an area of handkerchiefs. Grandpa was a working class guy, a carpenter, who raised my dad that you blow your nose into your handkerchief, then fold the fabric around the precious goop that was delivered, like you are packaging some kind of gift, then put it back into your pocket. When I first started encountering men’s business fashions, I thought the ‘pocket silks’ looked a bit nice to blow your nose in. Laundry bills, and all that. That was when I started to realize that there are all kinds of indicators of social class built into every stitch in our wardrobes. I pointed the practice out to my dad, once, and he thought for a while and said nothing. But then, he quietly switched to disposable paper – perhaps the idea of storing his snot suddenly didn’t make sense anymore. It has never made sense to me.

I’ve been fascinated by the run on toilet paper. It seems to me to be a “market failure” or whatever excuse economists come up with for when the markets don’t behave at all the way the economists want them to. It’s irrational: the coronavirus outbreak has not caused people to wipe their butts substantially more than they did, before. Even if it did – let’s say double the consumption – you’d see a brief shortage that would be filled fairly quickly as everyone adjusted to the new demand level. But, instead, the demand has gone off into someplace that makes no sense at all. Here’s what I think is going on: we’re seeing a downside of “just in time” supply. The computer models pioneered by WALMART and then Amazon et al are actually very good at predicting when you’ll want your next 6-pack of toilet paper, and they know, for each store, approximately how much toilet paper is consumed by the community it serves. Since toilet paper ought to have a rational demand level, the suppliers can make the right amount of toilet paper appear, as if by magic, to be sold – no need to have stockpiles of toilet paper in mouse-proof storage, like I do. You go straight from the truck to the shelf and then to the customer. Maybe, I am thinking, that “just in time” supply chains are individual disasters, primed and ready to go off when something violates the assumptions that go all the way back to where the raw materials are produced.

Supply chain management is a tricky thing, indeed. You can’t just tell the toilet paper guys “turn the machine to 11” and have the short-term demand fluctuation smoothed out – the paper comes from fiber suppliers, and the fiber comes from shredding and bleaching plants and somewhere there are loggers with chainsaws involved. Turning the machine to 11 means getting the loggers up earlier in the morning, or working them later, or getting more loggers and chainsaws. There’s money to be made in doing that, of course, which represents as a premium in the price for the extra loads of toilet paper to flatten the demand bump. I saw somewhere that the Trump administration apparently is trying to get General Motors to make breath masks, and General Motors asked for nearly $500 million to switch production. That’s because things have changed since WWII, when GM had a lot of machinists and metal workers and you can tell a machinist to stop making pistons and make shells instead, because the machinist a) already knows how to do that, they just haven’t been b) the materials are fairly similar c) human machinists are much more rapidly adaptable than computer-driven robots which don’t actually ‘know’ anything – they just lift this thing to a that position and put two screws in exactly that place then wait until their optical sensor tells them it’s time to do it again. The car makers’ supply chains don’t look like they did in WWII because they’re doing a certain amount of “just in time” supply as well – to switch to doing breath masks they need to replace the thing lifting robot with a silicone injection pump and vacuum system and they need metric fucktonnes of silicone and there aren’t many machinists left in the building and, if there were they’re machinists not silicone experts who know how to build a delivery system for the silicone precursors – they’ve never even seen that stuff. I’m betting that the management team at GM said:
Big Shot #1: Sure, we can do that.
Big Shot #2: (mouthing the words and holds up 2 fingers) (“two hundred million dollars”)
Big Shot #1: (thinks a bit) It’ll cost $500 million dollars Mr President. But we’ll cut you a discount and we can try to bring it in for $480 million plus the standard delays and cost overruns.

Basically, when you come to a businessman with a requirement like that, the price is going to be exactly ${what it costs to build a new factory} because you’re asking GM to start a new business. Adding the capacity to produce toilet paper? Another $500 million. Because there’s no reasonable overlap between those processes. What I’m getting at is that the Trump Administration are a bunch of stone-cold fucking idiots for not going to 3M, which already makes silicone masks, and has the supply chain and know-how and asking them “how much can you ramp production with $200 million if we seed another $200 million down your supply chain so the entire production-line’s capacity is boosted. We’d like it to double, what do you think?” When you’re talking about production lines, you really only have 3 choices:
a) Build a new, parallel production line
b) Run your existing production line twice as long
c) Run your existing production line twice as fast/hard

Each of those has tradeoffs, option a) is costly, option b) is going to increase labor costs and is a non-starter if your production line already runs 24/7, option c) is going to impact quality.

A little googling around shows me what I expected: there are robotic toilet paper production systems available in various sizes:

You can see the small one in action [youtube] but to flatten the bump in the demand curve you’d need gigantic machines – much larger than this. And, you can’t run the machine faster by turning it to 11, because as mechanical systems speed up, some processes become time-critical bottlenecks – for example, there may be a maximum speed at which a certain rolling system can roll paper before it turns to a confetti-maker. You could run it faster but you’d need a whole new machine architecture.

I did a few postings about rope-braiding machines [stderr] back in the day, and they’re a good example of what I’m talking about: some braiding machine architectures simply can’t go that fast because they depend on shuffling bobbins – you need a whole different way of doing it to run 10 times faster. 10 times longer is relatively easy (it has to be more failure resilient) but faster is a different game entirely.

What I’m getting at is that I bet some Washington Dumbass (perhaps Jared Kushner, personally) probably called 3M and told them “turn it up to 11” and they said “pound sand, we’re capitalists, it’s already turned up to 11. In fact it always runs with the knob set on 11.” And the Washington Dumbass said, “screw you then, we’ll ask GM” At which point the executive from 3M said, “good, waste some of their time, I’ve got a breath-mask-making factory to run. Bye.”

The classical response to supply chain management problems is what’s called “vertical integration.” What that means is that you bring more and more of your supply chain under your control, until you manage the entire thing from production of raw materials to owning trucking companies to deliver the product. That’s what WALMART does, on the stocking and selling side. But the first amazingly integrated vertical company I ever worked with was Koch Industries in 1992 (I was looking at the computer security of their remote-controlled refineries) – Koch started out as an oil drilling and pumping operation. Then, they realized they could control costs better if they owned their own refineries. After that, they realized they had a materials transportation problem and they could control costs and make it more predictable if they owned their own pipelines and a trucking company. Once that was nailed down, then distribution and sales became the next problem, so they started buying gas stations. When it turned out that the main barrier to franchising a gas station is the franchise fee and raising capital, they went into banking, too, so they could own and control the loan that the franchisee took to open the gas station to sell the gas that they refined from the oil they trucked from the wells they own. At each point, when you vertically integrate, you can drive costs down because you’re creating an internal transaction that does not have to be exposed to market forces – i.e.: you’re controlling your own little captive market, which is what every capitalist secretly wants to do no matter how much they talk about free markets.

Update: Here’s some about the GM contract [politico] – pull quotes

GM is to deliver 30,000 ventilators to the Strategic National Stockpile by the end of August, with a total contract price of $489.4 million. The first 6,132 ventilators are to be delivered by June 1.

The administration and GM had previously demurred when asked for specifics on the number of ventilators, the purchase price or the logistics of distribution. The only numbers that had been released publicly were in a tweet by Trump, complaining that GM had initially committed to deliver 40,000 ventilators “very quickly” but then switched to say they could “only” deliver 6,000 “in late April.”

Remember the F-35 trick: same price, smaller number of units. So the difference between 30,000 ventilators and 40,000 is a significant mark-up. I bet a dollar to a donut that the 6,000 in late April is what GM can produce by draining their existing supply-chain, while frantically negotiating for components from wherever they can find them.

Elon Musk fits into all of this, somehow. I read one piece by a former GM executive where he said that Tesla was doing something interesting by trying to re-invent the automotive supply chain that Ford and GM had perfected 50 years ago. His attitude was, “Elon’s smart; it may work.” But I thought it was interesting that his assumption was not as Musk’s is – Musk assumes he can walk up to a highly optimized existing supply-chain management problem and melt it away with the power of his own brilliance. You can do that with software because the software supply-chain is very different from machine products. Does Musk understand this? I don’t know. He appears to have one idea right, which is to hire smart people who know the problem, and let them try to tackle it in new ways. That’s a high-risk proposition because what if they are wrong? If Elon Musk announces that he’s spinning of a “roll on demand” toilet paper company, when we’ll know he’s just a dipshit who’s gotten lucky a couple times in a row.