Cool swarm simulations
I find identifying 'swarm' behaviors as "intelligent" to be a mis-use of the term "intelligent". Doesn't mean it's wrong, just means it makes me suspicious and skeptical.
Okay, now I'll read the article...
July 3rd, 2007 1:22pm
"That's how swarm intelligence works: simple creatures following simple rules, each one acting on local information. No ant sees the big picture."
One could the same about our neurons.
son of parnas
July 3rd, 2007 1:39pm
Fascinating article, thanks for the link.
The 'swarm behaviors' called out -- how an ant-hill 'decides' how many workers to send out, for instance, is all driven by 'simple' rules followed by 'simple' nodes (an ant, in this case).
The 'rules' are driven by scents, or concentrations of odors.
The remarkable thing is that a collection of simple units, following a small set of simple rules, can drive such apparently 'intelligent' behaviors.
But I would submit there's no "intelligence" involved here. We, from our 'high-level' perspective, percieve that it LOOKS 'intelligent' to send out a certain number of workers given one set of conditions, and a larger set should more food be available. But in fact 'intelligence' has nothing to do with it -- it's an 'emergent behavior' of the simple rule set applied by simple nodes.
"That's the wonderful appeal of swarm intelligence. Whether we're talking about ants, bees, pigeons, or caribou, the ingredients of smart group behavior -- decentralized control, response to local cues, simple rules of thumb -- add up to a shrewd strategy to cope with
Aha! Coping with complexity -- now THERE'S something that resonates with me.
"Such thoughts underline an important truth about collective intelligence: Crowds tend to be wise only if individual members act responsibly and make their own decisions. A group won't be smart if its members imitate one another, slavishly follow fads, or wait for someone to tell them what to do.
When a group is being intelligent, whether it's made up of ants or attorneys, it relies on its members to do their own part. For those of us who sometimes wonder if it's really worth recycling that extra bottle to lighten our impact on the planet, the bottom line is that our actions matter, even if we don't see how. "
Okay, so we CAN apply this to human systems. If we can figure out what the good 'rules of thumb' to apply are, and then convince people to apply these, then we can get groups of people to ALSO reflect useful "emergent behaviors".
And we can also apply this to software. If our modules are created with simple rules-of-thumb (coupling and cohesion, anyone?) then they too can reflect emergent properties which are useful (more easily maintained, reliable in the face of change).
July 3rd, 2007 1:57pm
I think bee hives and traffic jams have emergent properties not exhibited by their constituent parts. I just don't think human-like intelligence (or self-consciousness, as per other thread) is one of them.
July 3rd, 2007 2:02pm
> But I would submit there's no "intelligence" involved here.
Isn't intelligence in the eating? Your brain does not have any "intelligence" either, yet intelligence emerges.
son of parnas
July 3rd, 2007 2:07pm
See, that was my concern at the beginning, that we'd be defining and redefining "intelligence".
But the point of the article is that a 'swarm', of simple nodes following the RIGHT simple rules, SEEMS to exhibit behavior that an 'intelligent' being would recognize as being effective. It works. The ant-hill is fed, the hive finds the best nest, the Caribou avoid the wolf.
It's been widely recognized in anti-pattern group dynamics, that often the group is only as intelligent as its least intelligent person. Sometimes less so.
But if we can apply these approaches toward finding what the 'right' simple rules are for people to follow in group dynamics, then the group can exhibit more 'intelligence' than any one member.
July 3rd, 2007 2:17pm
> "I just don't think human-like intelligence (or self-consciousness, as per other thread) is one of them."
How would you know the difference? The 'emergent behavior' has varying levels of complexity. At some point it will become so complex and sophisticated that you can't tell it from what you think is 'real' intelligence.
> See, that was my concern at the beginning, that we'd be defining and redefining "intelligence".
It's going to happen when intelligent behavior is said not to be intelligent when it's ants but is said to be intelligent when it's humans. Neurons work by simple rules and don't have a global view. As do other cells. And we think we are intelligent. Does it matter if what arises doesn't come from neurons?
son of parnas
July 3rd, 2007 2:21pm
Yes, because calling both situations "intelligence" confuses the issue. An ant-hill making 'good decisions' based on a swarm algorithm is quite different from how a human being would make a 'good decision'. Entirely different systems.
And ant-hill uses simple nodes following simple (instinct driven) rules-of-thumb. A human can use many different algorithms to make a good decision -- write out plusses and minuses, look up what others have done on Google, write about it on CoT, toss a coin, throw a dart, write a simulation program to collect more data.
These are two extremely different systems. An ant-hill is not a collection of interacting neurons holding memory and associations -- which your brain is. An ant-hill's purpose is not to think -- which hopefully some part of your brain's purpose IS to think.
And the article was NOT about how an ant-hill acts like a group of neurons. The article WAS about how, even THOUGH an ant-hill ISN'T a group of neurons, an ant-hill can STILL have these "emergent properties" that result in 'good decisions'.
I still like that insight that if ONLY people in groups would follow a small set of rules-of-thumb, people in groups could generate better decisions than they do now.
July 3rd, 2007 2:34pm
And I agree, it's entirely likely that human intelligence is itself an emergent property of having sufficient "simple nodes" of neurons each connected only to its neighborw, following "simple rules" as to how these nodes interact to store 'information' and 'think' about stuff.
Shoot, the entire field of "Neural Nets" is predicated on this idea, and it can work pretty well.
I guess I'm splitting hairs when it gets to the term "Intelligence", and people start talking about "hive minds" of ant-hills. At that point, I think they leave reality behind.
July 3rd, 2007 2:39pm
> Entirely different systems.
Which doesn't mean one is not intelligent. Just like life that isn't carbon based is no less life.
> An ant-hill's purpose is not to think
The purpose of your brain isn't to think either. Your brain is a survival adaptation. It reacts, controls, monitors, and responds. The thinking is a just a way to learn stylized responses for quicker activation. The going to the stars stuff is merely incidental.
son of parnas
July 3rd, 2007 2:43pm
Oh, dear. This has the danger of going down the slippery slope of "define self-awareness, I think therefore I am", and "If you're not you then why are you typing this?".
I wish you luck in your future debate on this, but I'm done on this one.
July 3rd, 2007 2:46pm
> but I'm done on this one.
Sure, we'll just accept the default assumption that neurons == good intelligence and anything else == bad.
son of parnas
July 3rd, 2007 2:47pm
Not what I said, not what I meant to say. Are you feeling all right?
July 3rd, 2007 2:50pm
locutus > At some point it will become so complex and sophisticated that you can't tell it from what you think is 'real' intelligence.
Yes, at some point. Perhaps, under more rules, different rules, more degrees of organization. Perhaps. Perhaps not. Have traffic jams achieved intelligence or the sort of stimulus-response coordination seen in frogs?
STH > A human can use many different algorithms to make a good decision -- write out plusses and minuses, look up what others have done on Google, write about it on CoT, toss a coin, throw a dart, write a simulation program to collect more data.
Usually though, they discard all this and go with their gut. On issues of good and bad. Only philosophers pontificate and get confused.
sop > Neurons work by simple rules and don't have a global view. As do other cells. And we think we are intelligent.
Funny. I didn't think neurons are all that there was. To intelligence.
July 3rd, 2007 2:52pm
> Are you feeling all right?
Is disagreement a sign of sickness?
son of parnas
July 3rd, 2007 2:52pm
Not disagreement, no. Deliberate mis-statement of another person's position is not something I associate with you. Nor is reducing an argument to a one sentence "You say you good, ants bad" message.
I certainly did not mean to hurt your feelings in any way. In fact I hugely appreciate your posting that article, I had never thought about the issue in that way.
July 3rd, 2007 2:56pm
> Deliberate mis-statement of another person's position is not something I associate with you.
I didn't deliberately do that. I just boiled it down to my impression of your argument. You make a lot of statements about not wanting to define intelligence etc yet you are very sure that the ants aren't. That doesn't really work.
son of parnas
July 3rd, 2007 3:04pm
I've made several arguments as to why I think ants are not intelligent. An ant-hill's ability to make what look like "good decisions" is an emergent behavior based on a swarm algorithm. Which is a brilliant model and a valuable insight for me.
Since you've accepted none of my arguments, I'll conclude we have different definitions of intelligence and leave it at that.
July 3rd, 2007 3:11pm
The behavior of the colony is "emergent." Something that doesn't exist in the DNA of the ant, or bee; yet emerges from simple actions and a simple vocabulary that *is* encoded in the DNA.
My book review on slashdot:
The point is that it *appears* to be intelligent, and so people look for that central controller which leads to the Myth of the Ant Queen. And this hunt for the Ant Queen is why the wargasm in Iraq is guaranteed to fail - there is no central planning committee, there is no single leader that one can shoot to stop the violence. It is hundreds of little groups, who get smarter by sharing information.
>But if we can apply these approaches toward finding what the 'right' simple rules are for people to follow in group dynamics, then the group can exhibit more 'intelligence' than any one member.
Part of why the US can't possibly win in Iraq is that our feedback loops were contaminated and perverted by the neocon menace. As a result, the country is very much divided over even simple issues like "legal" vs "illegal." The general consensus from the 20th Century is that torture is useless from an "intelligence" aspect, yet we have the insane lunatics demanding that they be allowed to do what has been shown to not-work.
With the republican't tribe perverting feedback loops, they prevent "the group" from exhibiting more intelligence of any kind. That tribe tends to form a pack to insulate themselves from any outside influence, and stampede together. While this has been a success mechanism in the past, it will result in their downfall, probably due to them starting a civil war. I do not believe that there will be an election in 2008, and that they will create some crises that prevent the elections from happening. Oh, like maybe an H5N1 outbreak. Followed by, oh, some "attacks" on infrastructure, like power substations, a brdige or two. Just enough kristallnacht to make it so they have to declare a national emergency and cheney remains dictator for life (he'll probably have bush shot).
July 3rd, 2007 3:11pm
Wow, Peter. For a minute there, I thought you were talking about the Iraqi 2008 elections, then it turned out you were talking about America.
It sure would be convenient for Cheney if there was no 2008 election. But I don't think even HIS disregard for the Constitution would take him that far.
But I've been wrong before.
July 3rd, 2007 3:18pm
What if it turns out that the costly but necessary illusion that evolution has burdened us with is not that of free will or consciousness as has been suggested, but that of the I, the Ant Queen of Peter's.
If you'll excuse me, I'll go and google for cheap flights to Tibet now.
July 3rd, 2007 7:58pm