> Ants have been learning how to deal with congestion for millions of years. They might just show us the way out.
But the research here doesn't involve studying ants to learn their successful evacuation strategies. It involves noticing that ants don't seem to have any special strategies, studying which room layouts seem to make things easier for their strategyless approach, and then using those same layouts for human rooms, since humans, just like ants, have no particular evacuation strategies.
I think this is right. In my boredom, I have studied ants and my observation was that they seem to send a small number of ants in multiple directions, carving a room up into sections. If an ant finds something of value, they change their pheromone to something that attracts the other ants. With time, they all eventually lock onto that pheromone trail, especially if it involves food. On a side note, they especially like vodka and I suspect they are disallowed from drinking it, as the bigger ants come along and chop off their heads. I was taken aback and never gave them alcohol again. Anyway, it seems this behavior is how they manage to find food and then transport it back to their nest. If you get this pheromone on your hands, the ants will swarm you and run around in different directions very chaotically. It seems to be a chemical based form of group think. Perhaps their intelligence should be measured as a group.
“In Australia when it gets very hot, the nectar in some flowers ferments and turns into alcohol. Bees that get drunk from the nectar are not allowed back in their hive—guard bees keep them out to prevent them from making the nectar into alcohol honey.”
> In a room with six exits, it seems like the most logical course of action would be for the crowd to divide evenly among all six. Instead, we stampede to just one.
Something to watch out for if ever in that situation.
In high school I got caught in the crowd once following a release from some kind of pep rally, I was effectively crushed against a wall while it passed. The gymnasium had plenty of exits, but the field-side wasn't where anyone wanted to go. Classes and the cafeteria were through three double-door exits on the non-field-side. After that one mistake, I started going toward the field and then walked around the exterior. I've kept that pattern up whenever I'm in any crowded place with a lot of people being released at once. I look specifically for the reasonable exits that are away from most people's interest, even if it takes me a bit longer (often it's faster because there is no crowd jamming up the exit).
Movie theaters (remember those?) usually have a door or two up by the screen that no one uses. Most of the time I have seen them lead out some weird hallway that goes directly outside. They usually aren't emergency exits either. All around pretty great.
And many, in the US, don't alarm their emergency exits. I've used those to good effect when I've opted to see a particularly popular new release. You can usually park behind/beside the theater, not just in the main lot, so these exits are actually faster for accessing your car and then you can use access roads behind the theater to exit the parking lot as well.
Often they go directly outside. I once was seated near such an exit when someone who had apparently left (and left something behind) started hammering on the exit door.
This was years before the Aurora shooting, and after looking around at everyone else who didn’t want to make the decision either, I let him in instead of letting him continue to distract from the movie.
How many Aurora shootings have there been? How many people have used those doors as entrances? I very much suspect the odds suggest it was not a bad decision.
Ditto old theatre buildings in London's theatreland. These alt routes help deconflict the people exiting the current show from those coming in to see the next one.
This gave me the mental image of an HNer patting himself on the back for not following the crowd after finding himself burning alive in the nightclub's broom closet as it burns down.
Yeah exactly this. If you know all the exits are safe, then evenly splitting up is best. But I reckon in situations like this, most of the time you don't know and then following the rest seems like the best strategy to me.
I attended some emergency training a couple of years ago, and the instructor drove home one point in particular: always identify the exits when you’re in an unfamiliar building.
It seems like the best strategy because you figure other people know something you don't know. But since they always pick just one exit, that's not a valid conclusion.
Since the crowd behaves the same way whether they picked the best exit or a random one, you have no idea whether it's the best way out, but you do know it has added risk of delay and trampling.
The mechanisms for ants and humans clumping behaviours are, I expect, quite different. My understanding is that we are looking and listening while ants are smelling and feeling. This means that the actual architectural insights you get from ants - put a column here and a door there - might not generalize to humans.
In very crowded situations you end up with something closer to the ant-like behavior. Your visibility is reduced to your immediate neighbors, and theirs is similarly limited. This prevents the deliberate movement you'd see in lower crowd situations, or where people are very familiar with their surroundings.
That is, a stadium exit is not like the exit of a thousand workers from a factory at the end of the shift. The factory workers (ok, not always) know their path and may have an action to perform on exit (punching out) which establishes a tempo. A stadium exit is more freeform and people are less familiar with their overall exit plan. So they jump into the stream and go where it goes. They have to move as fast as the crowd behind them pushes them or faster.
Stadiums have signs though? Clearly lit pathways, signs, and directions to all the exits. I'm pretty sure it's the law to have those (at least for fire codes). I would be highly surprised if people followed their nearest neighbor instead of moving in a direction denoted by the signs and pathways prescribed.
I visited Venkateswara Temple in India, which gets some 50~100k visitors daily. The main takeaway I got was that it doesn't really matter what the path is -- you don't get a say in the matter. The flow of the crowd moves you along, and that's all there is to it.
You have people pushing from all around you (in a vague general direction), and next to no space to actually influence your movement -- and the lack of choice in the matter is fairly incredible. The pushing stops coming from any particular place, it just becomes some indeterminate feeling of pressure forcing you along.
I'm quite positive you could easily be crushed by it, getting caught between a pole somewhere in the path, and with no external factor to influence the crowd (a wall is clearly visible, a pole is not), simply be pressed against it indefinitely.
A stadium probably isn't an intense, but likely similar -- once you're in the stream, the only thing that matters is stream.
Eventually, you're suddenly ejected from the stream and have to figure things out for yourself again.
> the lack of choice in the matter is fairly incredible. The pushing stops coming from any particular place, it just becomes some indeterminate feeling of pressure forcing you along.
You can experience this in total safety any day you want by riding a subway during rush hour in a populous city.
Yes, but when you're in a packed crowd even these may become difficult to see for any individual. Think no real room to move freely, maybe not even 1' in any direction other than "forward", the direction in which the crowd is clearing. Some people see them, hopefully, and those are the ones that determine the direction of the crowd's flow. Which may be towards a specific exit and not to all exits.
I wonder, is this somehow a latency-throughput tradeoff in disguise?
This seems roughly analogous to batch processing, where you structure the dataflow to be mechanically sympathetic to the machine doing the processing. Processing each item becomes more efficient, so you process more items per unit of time, but each item takes longer to process individually.
In the same way, it seems like we're structuring the crowd into columns to optimize the layout of individuals with respect to the floorspace. It takes less time to evacuate everybody, but each person spends more time in the building.
I'm not sure what to do with this knowledge, if it is indeed correct.
EDIT: I think the question I'm orbiting is whether or not this strategy can generalize to people. In the realm of ants, things like self-preservation and the value of the individual are very different from humans.
It is definitely not a latency-throughput tradeoff in disguise, since the metric is "time taken for everyone to leave the enclosure".
> it seems like we're structuring the crowd into columns to optimize the layout of individuals with respect to the floorspace. It takes less time to evacuate everybody, but each person spends more time in the building.
As stated, this is impossible. Everyone starts inside.[1] If the time taken to empty the building is lower, it cannot be the case that each person spends more time inside; at least one person takes the full evacuation time, or else the evacuation time would be lower.
[1] This is the core difference from the analogy you picked, batch processing, in which jobs arrive at different times. If all the jobs arrived at once, batch processing would be the only possible kind of processing.
Does anyone know what's up with this account? They submit nautil.us and phys.org articles in rigid alternation and they seem to hit the frontpage with unlikely regularity, especially for a religious organization like nautil.us.
That they give grants to something doesn't mean that thing is religious. For example, they give grants to PBS [1] [2] for Nova and I doubt many would argue that Nova is a religious program or PBS is a religious broadcaster.
And their submission history looks to me to be full of whiffs, rather than getting the frontpage with "unlikely regularity". The user could be a bot, but both of those sources are well in the curiosity-gratifying wheelhouse of the average HN reader.
> Ants have been learning how to deal with congestion for millions of years. They might just show us the way out.
But the research here doesn't involve studying ants to learn their successful evacuation strategies. It involves noticing that ants don't seem to have any special strategies, studying which room layouts seem to make things easier for their strategyless approach, and then using those same layouts for human rooms, since humans, just like ants, have no particular evacuation strategies.