Dragonfly's extreme loop-the-loops
41 comments
·January 14, 2025sashank_1509
z2
Sounds about right, 1 million neurons and potentially 1 billion synapses. Back in school I used to think computers and machines were so advanced and complex compared to boring old biology. It was such a silly misconception to be so many orders of magnitude off in complexity.
jurgenaut23
Tokens? Is that the new « bit »?
_joel
I guess we're still not out of the 'flying into a window' stage. https://en.wikipedia.org/wiki/Loitering_munition
noja
Is nature doing that? Or following a much simpler pattern?
sashank_1509
I feel pretty confident saying it is not following a simpler pattern. It is doing something very hard, with far less training data and a much more alien form of compute (which is also much slower in general than computers).
Is there some elegant learning algorithm in nature that enables this? Maybe. But I don’t think there’s any simplification of the task, that nature has found. For example, I doubt we will find a 100 line algorithm to do the motion the dragonfly did, or a simple physics equation to describe it etc. My intuition, stems from the bitter lesson, a lot of these tasks are irreducibly complex and trying to find elegant simple models, tends to be a waste of time.
thoroughburro
Producing complex behavior simply is usually harder and more elegant than using a lot of superfluous complexity.
tw04
Does the how matter if we can’t replicate it the same way?
stronglikedan
I watched this cool video recently on dragonfly movement. Dragonflies Hunt by Predicting the Future https://www.youtube.com/watch?v=8i9WMD6xbuA
mrbonner
In the summer, one of my "hobby" is just to sit in my backyard and watch a couple of dragonflies circles around, swoop down and catch flies/mosquitoes. They have a way to fly close, rendezvous with the targets and then just swoop down in all of a sudden to kill them. Fascinating animal! I could sit there and watch them in constant state of awe.
philk10
same, have a natural pond in my backyard and I can watch them all day
stevenwoo
Not mentioned is the frame rate for data acquisition which can approach 300 frames per second equivalent so they kind of live through time a bit differently than we do. (Credit to An Immense World by Ed Yong)
meitham
That’s exactly the impressive part, the speed of doing these moves. The moves themselves on a lower speed is very common to see in fancy pigeons, though much slower
sparrish
Dragonflies are freakin' awesome. By far my favorite insect.
Not only are they incredible acrobats but they can eat up to a hundred mosquitos (my least favorite insect) per day with a hunting success rate of 95%.
adrian_b
The ancestors of dragonflies and damselflies have been the first insects that have evolved to fly well, not just to fly.
The other insects that are good fliers, e.g. flies, wasps, bees, certain moths, have evolved towards better flying in a completely different way from dragonflies, and many tens of millions of years later.
The ancestors of dragonflies have developed special muscles that can move separately and accurately each of the 4 wings, allowing exquisite flight control.
All the other good fliers among insects use only the equivalent of two wings, either by reducing one pair of wings, like in flies, or by having a mechanical linkage between the front and back wings, which makes them act like a single pair of wings, like in wasps, bees and fast-flying moths.
Moreover, the other good fliers among insects, unlike dragonflies, do not move directly the wings, but their flying muscles are attached to the thorax and contract it for moving the wings. A system of levers transmits the movement of the thorax to the wings. This also allows the wings to beat at a higher frequency than the frequency at which the nervous system can command the muscular contractions.
foobarian
Neat. So dragonflies are bitbanging the wings in a way
praptak
Their larvae are also pretty brutal predators and they also feed on mosquito larvae. Here's a YT vid showing how they hunt: https://www.youtube.com/watch?v=EHo_9wnnUTE
snthpy
That's awesome. Any way to facilitate them to multiply in my garden? There are enough mosquitos to go around.
rurban
You can try to attract bats with bathouses. They also eat tons of these beasts, and even better for fruit farmers
IncreasePosts
Bat houses can take 5 years before bats will decide to live in them - they won't just settle in some strange box because they found it. This lag time makes it very difficult to determine if the bats will like your box, but just need more time - or if there is some aspect about the placement or location of the box that means they will never take up residence in it.
eru
The Internet is your friend. See eg https://www.reddit.com/r/dragonfly/comments/10z03h6/how_to_a...
0_____0
To summarize the reddit thread, apparently no, you can't practically attract dragonflies to your garden
snthpy
Thanks
arnejenssen
I guess it is an issue that they are territorial.
TheSpiceIsLife
Damselflies deserve a bit of attention too.
Random video https://youtube.com/shorts/-ALXaGRl_g8
rednafi
For a moment, I thought this was about the Dragonfly database—the Redis alternative—and my morning brain couldn't parse the title at all.
_joel
I was thinking DragonflyBSD and -funroll-all-loops. Seriously need to get out more.
CarVac
I was thinking the planetary probe to Titan.
wink
> That was a big surprise to Christofer Brothers, who studies dragonflies and damselflies at UC Davis. “Before this talk, I was under the impression they avoided contact with the water outside of females laying their eggs.”
Not to throw any shade, but this sounds... weird? I mean, missing that they do something you can only see with a high speed camera? OK. But missing that they dip into the water? Would you not at least have watched them for a couple hours if you are "studying" them? At a university?
Oh, I study data structures but I've never measured the access time of a HashMap. Am I being too harsh?
thoroughburro
Yes. The behavior in question is not easy to observe.
wink
> About once every 10 minutes, a dragonfly dunks its body in the water
... and you see the water splashing in the video, with the ripples being visible for a bit.
Again, I was not talking about the rotations, just the dragonflies touching/immersing in the water. I just find it incredible no one watched them for 20 minutes at a time apparently?
ZYbCRq22HbJ2y7
This looks like one of the first usages of the phrase "absolutely insane" on science.org.
Cipater
Found a second usage.
https://www.science.org/content/article/how-oil-plume-change...
asadalt
“one in 4 eventually drown”… does this mean it’s a recent evolutionary trick and selection hasn’t settled yet.
tromp
> The dunk-and-spin maneuver is risky, Yarger notes, as roughly one in every four dragonflies eventually drowns.
It's not that risky, with about a 99.99% success rate. But in their lifespan of a few weeks, they will still be doing thousands of them...
foxglacier
It might just be because they have to die somehow. Probably 1 in 4 humans' hearts eventually stop working too.
robocat
Selection is a balance. It is hard to judge what is being selected for when you only look at one factor (drowning).
Aging versus cancer is an example of balance: one theory is that age related diseases are a side effect of selection forces against cancer: programmed cell death after X reproductions (telomeres) are a general anti-cancer defence but the cell-death has aging effects. Also beware that selection is for successful reproduction: death after reproduction is not so relevant to evolution.
mmooss
To see the presentation abstracts, at least, go here:
https://www.xcdsystem.com/sicb/program/fZq6Sh8/index.cfm?pgi...
... click on Search/Filter Program, then search for 'dragonfly'. Note that the results for each day are on separate tabs. The OP research is in at least two presentations.
I don't see a way to link directly to the search results or the presentation pages, which appear in some sort of popup.
If we wanted to make a drone have the intelligence of a dragonfly (which might still be a decade away), we would at least need a billion parameters and a trillion tokens in training data. Just helps us appreciate what nature is capable of.