A friend of mine recently got back from a trip to Japan and says their tour guide told them something remarkable: the Japanese spend more on adult diapers these days than they do on baby diapers.
Huh. But then he asked Microsoft's Copilot, which is based on ChatGPT, about the United States. It said the same thing was true here as well. This got me curious because it seems unlikely. In Japan, the ratio of elderly to babies is a whopping 19x. In the US it's only 7x.
Unsurprisingly, it turns out that Copilot screwed up: It mixed up US sales and global sales. If you compare just US sales, baby diapers are approximately a $15 billion market compared to only $5 billion for adult diapers.
I say "approximately" because I had a very hard time dredging up this number, and in the process I noticed there were a lot of estimates of the global diaper market from market research companies. Here they are:
This is sort of amazing. First, because there are so many formal market research reports, and second, because they vary wildly. The lowest estimate is $41 billion and the highest is $83 billion. Growth rates are estimated at anywhere from 4.3% to 7.3%. There's not even the remotest agreement about the size or growth rate of this market.
There are two lessons to take away from this. The first, of course, is that I can create charts about even the most trivial stuff. The second, and more important, is how much disagreement there is. The next time you see something in, say, the Washington Post about how "_________ estimates the market for _________ is X and it's growing at Y," just keep in mind that it's probably just a wild guess. Ask somebody else and you'll get a whole different guess.
Just disposable ones or everything?
I'm more interested in a different claim, namely that Nestle (and others) add sugar to formula for poor countries, but not rich ones.
Obviously if you're in the "Capitism bad, corporations evil" camp, that's all you need to know and claim. But suppose you're interested in real, as opposed to cartoon, motivations, then what's going on here?
(a) It's simply not true? One more lie from people who have a history of lying.
Possibly.
(b) It's done for good reasons. Like babies in poor countries don't get enough calories (are fed less often, were born weaker, etc) so adding sugar to formula makes good medical sense?
(c) It's demanded by parents. Who may believe (implicitly or explicitly) something like b, even if it's not true. They believe the babies need sugar, and won't accept a product without it?
(d) Something else? (eg regulations in those countries?)
I haven't read that report, but one reason to do it would be to cut the expensive ingredients with a cheaper one.
So, it comes down to poor piss research? …
Saw an interesting discussion recently where a programer said that when AI first came out they used it a lot for code, but recently it has become pretty useless and error ridden. They thought that the reason was that LLM's when first debuted where only ingesting human content but as time has gone on they are ingesting more and more AI produced nonsense, and that has made them unreliable particularly if they are programed to have recency bias.
GIGO will not be ignored…
That concern was my first thought when LLM's burst onto the public scene a couple of years ago. When they become so common that most content available for training is just stuff produced by other LLM's, they will all be echoing each other. That does not bode well for quality control.
You're right about QC. On the other hand, it mirrors a lot of the rest of the real world.
Because Godzilla is getting on in years and you can imagine how much an issue that could be.
Now, now for Market Research firms one should not be unfair.
"Wild Guess" is quite harsh.
Finger in the wind guess.
I used to have make some of these reports. It is really hard and really expensive to do a good job. In my case I was one guy trying to estimate current and future world markets for a product in a small industry. I wound up taking the good information I did have for a few companies and correlating that to other, larger, industries then I went around the world and calculated a result from the estimated business of those larger industries. If I got within a factor of two I figured I was doing well.
I guess those reports were made in a similar fashion. A handful of people doing their best to estimate something that can’t be directly measured.
Gartner and IDC track PC shipments. This is big, important business, and yet, sometimes they diverge one saying shipments grew and the other saying shipments fell.
The CIA fact book used to be really helpful. It estimates all sorts of business for countries around the world.
In Kevin’s example, the only way to get a real number would be to get all of the manufacturers of those products to open their books and tell you what they sold. Not going to happen,
There's one firm on that graphic that needs to get its act together. "Verified" reported $73B and $57B in the same report.
"Unsurprisingly, it turns out that Copilot screwed up: " It didn't screw up. by it's vary nature, anything chatgpt says is inherently fiction,, any semblance to reality is purely coincidental.
Did this post really NEED a chart?
What is the estimated internet blog market for charts?