25 Comments

Only if the oldest, most experienced members of the exchange each get a lot of info, and increasing amounts of info about the ‘stock’ before a decision is made.

Expand full comment

You're missing the significance of fact that bees will actively head butt other bees they disagree with. After enough of this negative signaling (harassment) behavior, the dancing bee with stop dancing.

Fun fact: a pushdown automata is only Turing complete if it has two tapes. Theoretically this means that you need both positive and negative signals to run the algorithm. The thought experiment of Buridan's ass (particularly in the vein of Leslie Lamport's view on it) suggests that the bee's search algorithm may be subject to the halting problem, and so we hit a speed/accuracy trade off (hence the 95% figure) due to physical time constraints subject to riks/reward processes.

A hive that can't find a site to live will die from exposure or predators, or simply not be able to continue searching due to running out of food.

Expand full comment

We can think of bees as cells of a single organism, but only to a degree, same as workers in a corporation, or citizens of a nation. The model is useful, but can be deceptive if you overapply it. Yes, the individual bees' interests are highly aligned, but, because the queen mates with multiple drones, the workers and virgin queens fall into genetic factions.

But you raise a good point: the interest-vs-knowledge question is key to recognize for any group-decision-making situation. If a group is trying to find the objectively correct answer, e.g. the weight of a cow, you might not call the individual estimates "votes". If the group contains interested factions (e.g. cow buyers and sellers), the estimates would be votes.

Expand full comment

Honeybee queens, unlike the queens of other eusocial insects, mate with multiple drones (a dozen, on average). Colony success correlates with this genetic diversity. The diversity has at least one other benefit: if a worker lays an egg (happens sometimes), another worker will eat it, maybe because it is likely to be from a half-sister, not full sister. Honeybees, to my knowledge, are the only eusocial insects whose colonies survive the winter (up to 5 years). Other species' colonies die or dissolve come winter.

Expand full comment

From '91 "Less than a decade ago, many experts considered bee colonies to be large, genetically homogeneous units. But recent work by a number of research groups has shown that honeybee colonies consist of subpopulations with slight genetic differences that may help determine what specialized tasks the insects take on." https://www.nytimes.com/199...

Expand full comment

It's not clear to me that individual bees have different interests to any significant degree.

If we think of individual bees as cells of a single organism (the hive), then bees are not really an animal that votes.

Instead, what we're seeing is the decision mechanism inside the hive "brain".

Expand full comment

Sounds a little like anxiety, if I understand correctly, they decline in expressing their opinion and then (either usually or there's no data) a bee never attempt to express anyone's site location opinion again;-- Considering the 95% accuracy for the best site it'd be interesting to know the distribution of the 5%, did they give up dancing too soon? are those sites very close in quality to the winning site? In this group of the oldest bee's searching it there potential for a large age range and does that influence their ability to dance?

Expand full comment

"...bee swarms seem to pick the best site available to them about 95% of the time."

If this study on bees is to be relied upon, then doesn't it mean that an unregulated stock market would be the best collective decision tool for human society 95% of the time? lol.

Expand full comment

Thanks Joseph. FWIW based on your and Robin's inputs I see no opportunity for selfishness at either the individual worker or the worker-coalition level to muck up the process. Quite a contrast with typical human group decision-making.

Expand full comment

Are we assuming that a change in explicit communication content indicates a change in signal for humans?

Expand full comment

To my knowledge, honeybees are the only animal other than humans that votes. By “vote”, I mean make a group decision among an arbitrary number of candidates. Under a looser definition (make a group decision) African wild dogs will grunt to make a yes-no decision on whether to go hunt or not—not very impressive. Under a tighter definition (group decision among candidates *when the group members have competing interests* and *when those group members can consider the possible votes of other group members*), then no. However, because honeybee queens mate with a dozen or so males, the workers are in factions: sisters vs. half-sisters. They thus have a bit more genetic variation, and presumably some different preferences. So honeybees make a fine mascot for voting reform:http://unsplitthevote.orgHoneybees are also a mostly-vegan matriarchy, they pollinate our crops, and they give us honey—what’s not to love?

Expand full comment

I read a few books on honey bees while I was preparing to be a bee keeper. There are some assumptions in the following question, I'd like to clear up: "Is there explicit evidence that the scouts are better at evaluation than non-scouts?"There are only 3 types of bees: * Queens* Worker Bees* Drones (males)The drones are only good for mating with the queens when just after the queens are born they leave the hive to mate. Other than that the drones are dead weight on a hive. Other than laying the eggs, all work in the hive is done by the worker bees. Each worker bee seems to select what work they are going to do based on their age and individual perception about what the hive needs at this time. In the article, "The oldest most experienced bees..." become the scouts. I don't believe this is because they're the most experienced, but because looking in random holes in stuff is dangerous work and loosing the oldest bees is safer than loosing young bees.In some ways, the bee is to the hive like a cell is to a dog. Hives reproduce hives. Individual bees can't make new bees.

Interestingly the queen isn't actually in charge of the hive. She does release pheromones that influence the hive to support her, but just like how the scouts present many options and the bees choose, the worker bees can choose to rear a new queen and replace the current one once she is old.

The bee is to the hive as the cell is to a dog. The hives make new hives. The hives bread with each other. A bee by itself, even a queen, can't make more bees.

The final paragraph says it doesn't see the bees' behavior in humans. To me it is what we call "market effects." What should my house look like? It should look like everyone elses I see. Humans at the market level seem very similar to bees.

Expand full comment

So while scouting is not necessarily about learning, recent studies have shown that honeybees can learn, and also do math.

Expand full comment

I've long liked this comparison to range voting. In section 5 it spells out rather clearly how bees reach consensus.https://rangevoting.org/ApisMellifera.html

Expand full comment

Scouts are among the oldest bees, experienced foragers. Individual honeybees, like other animals, have personalities, e.g. some are more novelty-seeking (and being scouts and foragers), others are fastidious (and spend more time as undertakers), and others are more anti-other (and spend more time as guards). Bees self-select, and (to my knowledge) are never selected. They respond to various cues about what roles most need more resources, and they move from one role to the next as they mature ("age polyethism").

Expand full comment

I don't recall hearing before of long term learning in insects but that Is obviously suggested by the scout selection criterion. Is there explicit evidence that the scouts are better at evaluation than non-scouts? That the initial search scouts are better at search than the other scouts and the non-scouts?

Expand full comment