45 comments

[ 4.7 ms ] story [ 77.6 ms ] thread
Throws it in the food processor.
To translate the final answer from math to human (as I’m going to be explaining this to my mother when I chat with her next!):

Imagine the half onion is a half rainbow. You know there’s another half rainbow lurking below the surface, the onion’s ghost of the sphere it once was. Place your knife as usual for each of your ten dice cuts, but instead of cutting straight down towards the cutting board, angle it slightly inward towards the end of the onion’s ghostly half-rainbow sphere below the board. Check your fingers for safety and then make your cut. Assuming your knife isn’t a plasma cutter, you’ll be stopped at the cutting board without ever reaching the onion at the end of the rainbow, and that’s cool. Set your knife at the next dice point and try again :)

(This still improves on the other dicing cases and only costs 1% uniformity by using 100% radius as the target.)

From their 2-d diagram, I'm having a hard time understanding what they mean by "vertical cuts" and "radial cuts"
Enjoyed reading this. I've followed J Kenji Lopez-alt for a while and I've practiced the "aim below" method for a few years now.

I also like that the article ends with the perfect Kenji-ism. "Yes, technically my method is statistically ideal, but like, it's home cooking and it doesn't matter, heterogenity isn't the enemy". Reminds me of Adam Ragusea (https://www.youtube.com/watch?v=5cWRCldqrxM), we're not making fancy french cuisine, we don't need a perfect brunoise!

Why limit it to just two horizontal cuts?

I’ve always just made equal horizontal and vertical cuts, then slice the onion crosswise.

This results in pretty much no large pieces, and only some smaller pieces (which I prefer over larger ones, anyway).

I don’t care about standard deviation — I only care about minimizing the maximum size (but still without turning them to mush).

(Also, I know this was more of a fun mathematical look at chopping onions vs. practical. But still the “two horizontal cuts” thing seemed to be practical guidance, when it seemed like just equal horizontal and vertical cuts is far superior. But, granted, it’s a little trickier to do.)

EDIT: looking at Youtube, looks like the 2-cut thing is normal. But adding a few more cuts isn’t that much harder, and eliminates the larger pieces from the 2-cut method. I’ll stick to my method, even if it’s a little more work.

I dislike easily 90%+ of the images I recognize as AI-generated, but the ones on this internet web site I think are a good use of the tech.
This ignores the obvious solution of not cutting all the way through. If every other radial cut is only through half the layers, you avoid making the inner pieces too small. It's funny how common it is for people to claim some sort of optimality with lots of math and analysis while completely failing to consider a better possibility. Never take seriously claims that someone found a "mathematically optimal" way of doing something. They didn't.
Real-world optimal is often just "whatever works best without making you cry (too much)"
> It turns out that making horizontal cuts almost never helps with consistency.

They made the horizontal cuts evenly spaced between the cutting surface and the top of the onion, which is nonsensical to me. I believe that a single horizontal cut at around 15-20% height would be better for uniformity than a horizontal cut at 50% height.

Evenly spaced horizontal cuts might make sense from a modeling standpoint, but not from a practical one. In real-life onion dicing, no one's slicing horizontally all the way up like that, it’s usually one or two low cuts to help break up the base
Once upon a time, my father, who could not cook, harshly criticised my onion chopping technique. This knocked my confidence in the kitchen quite a bit. I refused to learn the fancy techniques of the TV gameshow celebrity chef that my dad was enamoured with.

In my opinion, so long as you are chopping onions, all is well. Sure it could be dangerous, with fingers and egos at stake, but far worse is to not be chopping onions as that means ready meals, take out meals and having a poorer diet.

Standard deviation is a poor measure because you care more about avoiding big pieces than small ones. Penalizing for having a few tiny pieces doesn’t make sense.
A few tiny bits will just melt away or brown faster, no big deal. But one big undercooked chunk in a bite of otherwise soft onions? That's noticeable
(comment deleted)
Lol love the extent of clarity of the experiment, findings and interface. I think that for practical purposes, it would be better if the std-dev of pieces with size above a certain threshold is observed. From my experience, pieces above a certain size cause inconsistency in cooked onions. But maybe it depends on the recipe.
For sure this research can be nominated for an Ig Nobel Prize :-)

On a more serious note, thanks for posting this and letting me (us?) know about "The Pudding".

But is uniformitiy the goal?

If you want diced onions, the cook generally wants onion chunks below a certain cubic mass, so they cook and dissolve easily and uniformly. It does not matter if some pieces are 50% of that size, some are 20% and some are 80%.

With that, 1-2 horizontal slices and a bunch of straight downward slices are the safest and easiest way to achieve that.

That technique also expands to onion rings, sauteed onions and such.

As long as no chunk is big enough to stay crunchy while the rest caramelizes, you're good
Yeah it's an interesting theoretical problem but the practical applicability is limited.
Burnt bits add unpleasant acridness to the finished dish. And pieces that are 20% of the general size are very like to be overcooked when the rest are properly cooked, so limiting those is important.
It's funny seeing people dunking on taking this much effort to analyze onion cuts. The rewards for improving your onion dice are indeed probably low. But in mainstream western cooking, you need to do it almost every meal, and the analysis/learning is a one-time cost rather than a cost applied at every prep. Seems like an extremely reasonable thing to noodle on!
I figure since it's only an onion I'm glad that that mathematical optimization is not really necessary.

After all there are many more approaches that can be more mathematically rewarding, might as well enjoy it when you can ;)

The authors rightly point out how the pieces near the bottom get elongated in a vertical cut, but don't realize that the whole point of the horizontal cut is to cut those elongated pieces in half. It's not meant to be a cut halfway up the onion.
I make fresh Pico de Gallo twice a week so I chop a lot of onions. Besides an even dice, I’m interested in not dicing my fingertips. Radial slicing a 180 segment or adding horizontal slices is too unstable.

My method is to cut in quarters, give a quarter a vertical dice, rotate 90, do another vertical dice, then go longitudinal.

This remembers I have a challenge to figure out with some friends.

How to split a round cheese in in 5 perfectly without using any tools except the knife.

Assume you have the ability to cut in half perfectly always

Assume that if you can slice it in 10 equals pieces it is also a valid solution because you can just give two pieces for each

What I want is a cutting technique that’s good enough while still being practical for people to do. I am not sure I’m dexterous enough to slightly and consistently tilt the knife as I go through the onion.
So using some complicated angle gets you 4% less std dev than just doing it the easy way that everone already does it. Ok.