When the label has (millions) in it, it means multiply by 1e6 yourself. If you want to put the 1e6 back in, you'd need to remove the (millions) from the label.

I'm also not clear on the motivation here. The original example was correct.

Yes the original example was correct, but it was unclear because the 1e6 multiplication was implicit and most of these numbers are in the billions. Was motivated to open this PR b/c I was showing this example in a seminar and was tripping over the yaxis.

Since the y-label states it's in millions I would not say it's implicit. However, I think if I were plotting these sorts of magnitudes then I would go the other way: divide by 1000 and change the y-label to say "billions". Would that make it less of a trip hazard?

Would that make it less of a trip hazard?

Agree w. you that I think framing in billions is the cleanest here as it's the 7 thousand million that's causing me to do a double take.

converting back to draft because the decimal points aren't being displayed 🤦‍♀️

You could set a custom formatter. However, it is extremely common to specify values in millions (eg, quickly found example - https://www.apple.com/newsroom/pdfs/fy2023-q4/FY23_Q4_Consolidated_Financial_Statements.pdf). I am still not clear on the advantage of this change.

. I am still not clear on the advantage of this change.

In the example you link to, it's in millions to preserve the significant digits, but here we lose all those digits anyway b/c of the locator/formatter. I realized I'd been mixing up the formatters/locators w/ my data values (I blame sleep deprivation), and I think seeing 7000 million didn't help w/ that.

Basically, if we're losing all the millions significant digits in the labeling then we may as well just label in the billions.

So ended up adding minor locator at every 100 million because that's what I think I was really after, an aide for reading off the count in each band width + billions since that's the significant figure the locator lands on

My two cents:

• the example Stackplots and streamgraphs is about stacking data, the axis labelling/scaling seems to be rather irrelevant in this respect
• billion may be a bit ambiguous (1e9 or 1e12, at least for non-native speakers), although it's clear here in the given context of world population, so I'd either leave it as is or remove "(billions)" or "(millions)" from the label and use a multiplier offset instead.
• I find the minor ticks aesthetically rather unappealing (too dense) and given that it's anyway quite hard to read off quantities from stack charts I don't think they add any real help in reading the chart.

that it's anyway quite hard to read off quantities from stack charts

Yes but you still want to get a sense of proportion and I think minor ticks help w/ grounding, but also changed the minor ticks to .2 to loosen it a bit & the upside is then it more or less lines up w/ the Africa count (which side note, fixed the capitalization).

the axis labelling/scaling seems to be rather irrelevant in this respect

Yes and no - the reason I opened this PR was b/c I was talking about this chart in a seminar (as an example of how the same chart can have different names - stackplot and area chart) and the labeling distracted from the discussion b/c it's a bit clunky.

Since no-one objects strongly enough to block, and @story645 has clearly explained the motivation for the change, I am going to go ahead and merge.

People also write things like "10,000 MWh per day", so, yeah, unit handling isn't what it used to be.

(https://ourworldindata.org/scale-for-electricity)

People also write things like "10,000 MWh per day", so, yeah, unit handling isn't what it used to be.

(https://ourworldindata.org/scale-for-electricity)

😢