Tweenbots are human-dependent robots that navigate the city with the help of pedestrians they encounter. Rolling at a constant speed, in a straight line, Tweenbots have a destination displayed on a flag, and rely on people they meet to read this flag and to aim them in the right direction to reach their goal.

Given their extreme vulnerability, the vastness of city space, the dangers posed by traffic, suspicion of terrorism, and the possibility that no one would be interested in helping a lost little robot, I initially conceived the Tweenbots as disposable creatures which were more likely to struggle and die in the city than to reach their destination. Because I built them with minimal technology, I had no way of tracking the Tweenbot’s progress, and so I set out on the first test with a video camera hidden in my purse. I placed the Tweenbot down on the sidewalk, and walked far enough away that I would not be observed as the Tweenbot—a smiling 10-inch tall cardboard missionary—bumped along towards his inevitable fate.

The results were unexpected. Over the course of the following months, throughout numerous missions, the Tweenbots were successful in rolling from their start point to their far-away destination assisted only by strangers. Every time the robot got caught under a park bench, ground futilely against a curb, or became trapped in a pothole, some passerby would always rescue it and send it toward its goal. Never once was a Tweenbot lost or damaged. Often, people would ignore the instructions to aim the Tweenbot in the “right” direction, if that direction meant sending the robot into a perilous situation. One man turned the robot back in the direction from which it had just come, saying out loud to the Tweenbot, “You can’t go that way, it’s toward the road.”

It’s a measure of our restored sanity that no one called the TSA. Or maybe it’s just that no one has tried this in Boston yet. Or maybe it’s a lesson for terrorists: paint smiley faces on your bombs.

Posted on April 13, 2009 at 6:14 AM51 Comments


jorj April 13, 2009 7:28 AM

Reminds me of Paddington, whose survival depended on a cardboard sign saying “Please look after this bear”

Roger April 13, 2009 7:52 AM

Hmm. It’s nice to bask in the reflected glow of human kindness, but this nasty old cynic is a little sceptical. Kinzer says there were “numerous” missions with 100% success, but only a single experiment with a single device is actually documented here. Apart from the “mission” in this case being quite easy (point machine along correct flat, vehicle-free path), the results may have been somewhat influenced by a certain biasing factor: this park is a major student hang-out (being almost completely surrounded by NYU), and a frequent venue for street performers. We have to wonder how it would fair after dark in the seedy end of town.

A few other jaundiced thoughts:
1. These devices are not robots, as they are unable to react to environmental stimuli in any way. In fact, even the ability to move forward may be largely superfluous; it travelled approximately 400 m through the assistance of 29 persons, or about 14 m per person. Since a couple of people actually carried it several metres, I have to wonder how it would have fared had it just been a motionless box with the same message? (Indeed, Kinzer may have wondered the same thing, judging from a single still in the “NewBots” section.) Somehow, I don’t think a simple box would do as well, and the difference is the novelty. If a random stranger asked me to deliver his parcel for him, regardless of any cute smiley face I would:
a) wonder what was in the box, and possibly peek if I was not so honest as I am;
b) point out that many commercial organisations offer box delivery services for a small fee; and
c) tell him to deliver his own darn box.
But turn it into a novelty and it becomes a game.

  1. The overwhelming majority of people in Kinzer’s video in fact ignored the device. The curiosity is not so much that less than 1% of passer’s by “helped” the machine, but that no-one at all acted maliciously to it (if one can speak of malice toward a cardboard automaton.) I wonder if this was influenced by the facts that:
    a) this was done in broad daylight, with a large number of witnesses;
    b) being made of a cheap disposable material, there was nothing to gain by stealing it; and
    c) although Kinzer concealed the video camera, it surely would have been obvious to anyone of even modest intellect that whoever built the thing would be watching it.

Anonymous April 13, 2009 8:00 AM

It was a war or three ago that folks put bombs into stuffed toys. The folks who did that found out that they lost the respect of their core constituency, and thus it wasn’t worth it. Now it’s more likely that a group will unjustly accuse their opponent of using this tactic, than use it themselves.

As to not calling authorities, a park ranger is one of the folks who helped the little guy. New Yorkers, in the current decade, seem to have a good sense of balance between tolerance and intervention. How long will it last? Who knows, but I’m not terribly optimistic given the current declining economic environment there.

Mace Moneta April 13, 2009 8:08 AM

Such serious replies to such a light piece.

Doesn’t anyone remember fun anymore?

Andre LePlume April 13, 2009 8:40 AM

Given that “tween” is also used to refer to a pre-adolescent human (age 9-12 or so), a story about “tweenbots” could have been decidedly more annoying or less wholesome.

Alan Porter April 13, 2009 8:50 AM

That works because the ‘bot is unusual. Once they are commonplace, they will be seen as an annoyance… or a source for spare parts and perhaps a “free payload”.


Vicki April 13, 2009 9:22 AM

New Yorkers also have a deep-seated impulse to help strangers and tourists get where they’re going: people will stop on the corner if you are looking at a map or directions and look puzzled, and offer to carry a complete stranger’s suitcase up the subway steps. I’m not surprised some of that extends to obviously non-human objects.

bob April 13, 2009 9:32 AM

If these things were human-sized and had an angry face painted on them I suspect they would not fare as well. These things are like an electric puppy.

@Josh O: Exactly. Reminds me of a military operational readiness inspection we had (a major event where your whole units stops doing what it does in order to pretend to be doing it while someone else watches) where each of the three geographically separated sub-units was tested with a simulated bomb attack. At my unit they went up to a locked backdoor (which they had authorized us to have a ‘simulated’ guard at) and left a briefcase. Apparently our simulated guard was a moron because he didn’t notice it and it blew us up. However at the sister unit they put a stack of bright red cardboard tubes and an alarm clock in the middle of the parking lot and painted “La Bomba” on it. They noticed theirs and survived. Good training for Afghanistan.

FP April 13, 2009 9:54 AM

The bot is cute. I agree with Bob, the puppy analogy is accurate.

It’s also an interesting outcome that such a trivial device only needed the occasional attention from 1% of passersby to accomplish its mission. Makes you believe that a robot with just a little more intelligence can accomplish so much more (delivering pizza)? At least until, as Alan notes, such robots would be considered commonplace or even annoying.

Somehow it makes me think of the lost child scenario. The “child” made it safely from one end of the park to the other. It was assisted by total strangers for no benefit to them without being abducted or raped.

Peter Pearson April 13, 2009 9:57 AM

“Tweenbots are human-dependent robots that navigate the city with the help of pedestrians they encounter.”

I thought this article was going to poke fun at pedestrians bustling along the sidewalk while absorbed in their phones, blackberries, et cetera.

A nonny bunny April 13, 2009 10:59 AM

@ Roger

  1. These devices are not robots, as they are
    unable to react to environmental stimuli in
    any way.

Isn’t that the case for many industrial robots as well?
Besides, if you consider humans a component of the system (since the system was designed that way), then they do respond to the environment, through that component. It’s a bit like the robot equivalent of the extended mind.

Annie Nomous April 13, 2009 11:01 AM

@Roger: Thousands of industrial robots would disagree with you. Carrying one’s controller around with one isn’t necessary to be a robot. As this experiment demonstrates, you don’t even need you own intelligence, you can just borrow some when you need it. There are also experiments that demonstrate that you can pretty reliably get a letter delivered (sometimes faster than the postal service) by addressing it and leaving it in a public place. I’m also reminded of the “hitchhiker” cardboard cutouts from a few years ago. Sorry I don’t have links handy.

BF Skinner April 13, 2009 11:43 AM

There are a few people here inclined to puppy kicking.

“Kick the baby, Ike!”
“doan kick da baby”

Davi Ottenheimer April 13, 2009 12:01 PM

Extreme vulnerability of minimal technology? Vastness of city space? Struggle and die in the city?

The author sounds like a copy writer for sci-fi comics.

How about this: small boxes on wheels — that look like a common child’s toy — are put into a city park on a sunny day, with a flag explaining an experiment. Some percentage of people notice the box when it’s stuck in a spot, and nudge it in a different direction. Some read the instructions and point it in the right direction.


peri April 13, 2009 12:13 PM

I find this interesting on so many levels. First of all, the programming language is English, which is kind of brilliant. Secondly, repeated tests would shed interesting light on paths people take through the park. Finally, I am curious to see how participation changes over time; do people get bored of it or do they go out of their way to try to interact with it again?

bob April 13, 2009 12:36 PM

I am surprised no one called the cops to come blow it up (for that matter, in NY I am suprised no one stole it). They really SHOULD try this in Boston. They just need to make very sure the bots dont have any way of being traced back to them. Or maybe the best test would be to set one loose in NY with Boston as the destination (probably need solar cells to help it along. Or maybe have a sign that lights up “Please recharge me”). Although you would of course need to disable the lights near Boston because naturally the more LEDs a device has the more likely it is to be a bomb.

Back in ~2003 I was unable to enter airspace near Ft Bragg, NC due to administrative issues and was in a holding pattern for 20 mins or so between the base and a nuclear power plant (in a 4-seat plane) I have often wondered how many people saw my plane going over every two minutes and called it in as a possible terrorist attack.

Jason April 13, 2009 12:36 PM

I would love to see some good statistics on this. Run, say, 50 trials. How many robots got to within 10 m of their destination? How many were destroyed/lost/stolen? How rapidly did the robots make progress (trip distance * wheelspeed / time elapsed: would equal 1.0 if the robot were perfectly guided, less for inefficient guidance)?

Michael Dwyer April 13, 2009 1:15 PM

paint smiley faces on your bombs.

You know, I really wish that Bruce wouldn’t have said this.

The Tweenbots are a heart-meltingly nice example of how so called Good People still exist, depite all the media stories telling us that everyone in the park is a rapist, murderer, or paedophile.

I know we were all thinking it, but… somehow, it destroys the story when someone mentions the giant elephant in the room.

Sometimes I’d like to stay naive.

eve April 13, 2009 1:31 PM

@Davi Ottenheimer

RTFA again.

The author (and experimenter) is an artist, enrolled at ITP:

“ITP is a two-year graduate program located in the Tisch School of the Arts whose mission is to explore the imaginative use of communications technologies — how they might augment, improve, and bring delight and art into people’s lives. Perhaps the best way to describe us is as a Center for the Recently Possible”

Davi Ottenheimer April 13, 2009 4:10 PM

“nice example of how so called Good People still exist, depite all the media stories telling us that everyone in the park is a rapist, murderer, or paedophile”

the sentiment of this comment is very attractive, but it doesn’t account for the simple fact that it is impossible to rape or murder these little test boxes with wheels.

the litmus for “Good People” should probably be higher than whether we can nudge a small stuck toy with instructions into a different direction. i’m no psychologist but it seems to me the study does not include a representative choice or plausible inference to inter-human risk/safety.

Limnologist April 13, 2009 4:26 PM

Oh, and on topic:

I guess if you put a smiley face on something people don’t see it as a potential vector for terrorism. If I was a terrorist, perhaps I should give serious though to putting my weapons inside of “Hello Kitty” dolls.

Sam Greenfield April 13, 2009 5:28 PM

I like to think that we New Yorkers are marginally more sane than folks in other cities. (Not that we are perfectly sane, of course.)

@Limnologist: “If I was a terrorist, perhaps I should give serious though to putting my weapons inside of ‘Hello Kitty’ dolls.”
Not really; if you’re a terrorist your attack will probably be vicious, violent, and memorable regardless of the packaging. To me, this relates to the problem in Boston: people obviously overreacted to an innocuous installations because they feared the unknown. Terrorists do not need to wrap their bombs, guns, or weapons in either cute or frightening disguises. Let’s face it: if I wanted to plant an explosive device in Union Square, I would be better off just leaving it in a trash can.

Interestingly, truly random searches of people on the subway by the NYPD can prevent random explosive devices from being left in a trash can in the subway. Whether or not we should perform random searches is a question of costs (of freedom and money) versus security.

Neighborcat April 13, 2009 8:04 PM

It’s probably just me, but does that thing give anyone else the creeps? Adorably cute, totally out of place, and clearly up to no good. It’s a caricature of innocence, and I wouldn’t be surprised if it cheerfully incinerates and/or eats everyone who touches it.

Clowns are evil too.

denis bider April 13, 2009 9:26 PM

I find it interesting that no one before me mentioned how the mechanism by which the bot arrives at its destination is similar to how Wikipedia manages to thrive. Like with these “bots”, only a small percentage of people who encounter an article influence it, and the net balance of that is improvement.

Jason April 13, 2009 11:55 PM

“similar to how Wikipedia manages to thrive”

Very good point, Denis. But there’s a key difference difference between Wikipedia and real-world objects: on Wikipedia, damage done by malicious people can easily be reverted.

On the other hand, mangling a robot in a busy park is not as anonymous as mangling Wikipedia.

I’d be willing to bet that the lifetime of a Tweenbot is measured in minutes once the park gets empty enough that nobody nearby will see how you treat the robot.

Just in Case April 14, 2009 6:25 AM

Or maybe it’s a lesson for terrorists: paint smiley faces on your bombs.

Do not taunt happy fun ball?

bob April 14, 2009 7:13 AM

@Davi Ottenheimer: Rape, probably not. But murder? I suspect a post-pubescent male under the age of 50 could do to one of these botlets the equivalent of Manson v Tate in about 4 seconds with a hard-soled shoe; 2 seconds longer in sneakers. I’m kind of surprised that no one destroyed one by accident just by stepping on it.

Fred April 14, 2009 9:41 AM

Or maybe it’s a lesson for terrorists: paint smiley faces on your bombs.

Will you stop giving them ideas!

Paul April 14, 2009 9:56 AM

@Sam Greenfield

“Interestingly, truly random searches of people on the subway by the NYPD can prevent random explosive devices from being left in a trash can in the subway. Whether or not we should perform random searches is a question of costs (of freedom and money) versus security.”

Care to support that with evidence? “Random search” does not mean “a search which magically stops random events”. It means you pick a random sample and search it. Whether the set of actors randomly searched and the set of actors that are going to randomly place a bomb overlap is a minor complication you seem to be ignoring. The only way you would be able to keep people from randomly placing explosives is to search and test EVERYBODY, otherwise there is no way to guarantee your random screening is not missing the bad actors.

Anonymous April 14, 2009 6:38 PM

@A nonny bunny, Annie Nomous:

Some industrial automatons are casually referred to as “robots” simply because their sequence of actions is so complex it looks like the work of a robot. Such a usage is rather loose, and most roboticists exclude such machines from the category.

Having said that, nowadays it is actually quite rare for such machines to have no sensor processing. At an absolute minimum, nearly all of them have some kind of sensor to detect the presence of humans and shut down as a safety measure. (There is a nearby factory where a main warehousing floor is occupied by robot forklifts that noisily whiz about fetching and carrying, until a human comes in. Then they instantly stop and lower their loads, and all falls silent. They look, for all the world, like beasts squatting on their haunches and watching you. It’s quite spooky, actually.)

Most industrial robots also have things like force feedback sensors (so they squeeze just hard enough, not too hard), or even more sophisticated methods to optimise performance. At a nearby factory, pretty well all of the systems have a network of sensors feeding into PLCs or PID controllers, constantly regulating the machine’s operation to as close as possible to the optimum — because it produces better products cheaper, with less waste output, fewer defects, less energy, fewer resources, lower cost and less environmental impact.

Roger April 14, 2009 6:55 PM

@Davi Ottenheimer:

Yes, like what percentage of people simply ignored the boxes on wheels. 10% or 90%?

Before my first post, I made the following crude approximation: the path length indicated in the map is about 400 m (actually more, since I didn’t count along all the wiggles.) In the video, the machine seems to be doing somewhere on the order of a quarter of a metre per second. Thus, its journey took at least 1600 seconds, plus any time it spent stuck.

In various scenes, people are passing it at a rate of between 5 per second and none. I took 2 per second as a ballpark guess. That gives us 3200 passers-by, of whom 29 stopped to help, i.e. 0.9%.

Since there are a lot of uncertainties in this estimate, of course I rounded off to 1%; however due to not counting the wiggles, and not allowing for “stuck” time, I think that is conservative. I certainly don’t think anything like as many as 10% helped it.

Roger April 14, 2009 7:17 PM


First of all, the programming language is English, which is kind of brilliant.

Well, I suspect it would have worked much less well if the instructions were in a language that most New Yorkers do not understand. On the other hand it would be fun to repeat the experiment with the instructions written in Arabic!

Guy in NC April 14, 2009 11:15 PM


you were most likely diverted because of Pope AFB not Ft Bragg unless training was taking place.

The little guy looks just like the bots from the ps1 game megaman: legends.
they gave a feeling of evil and chilling feelings, turned out to be just minions for the villian!

still i thought of setting it loose in a warehouse see how long it takes to navigate with the help of busy package sorters 🙂

if the experiment had taken place near anything significant to the military, gov, or known organizations it would have been called in. i suspect that the artist had notified someone, maybe even spoke to the NPS before setting it loose.

MysticKnightoftheSea April 15, 2009 3:03 AM

@ eddie
Though unrelated to topic, I’d like to know how many children have tried to do their own version of “Paddle to the Sea”?

Just a thought.


Clive Robinson April 15, 2009 3:48 AM

@ Guy in NC,

“i suspect that the artist had notified someone, maybe even spoke to the NPS before setting it loose.”

In the buracratic CYA society many on this blog claim the US has turned into, I’d find it difficult to belive any peon of officaldom would take the risk to their pension and livelyhood…

If they did then there certainly is light at the end of the tunnel 8)

Clive Robinson April 15, 2009 4:53 AM

@ Paul,

“Care to support that with evidence? “Random search” does not mean “a search which magically stops random events”. It means you pick a random sample and search it.”

As you well know it cannot be supported with evidence, (other than the fact that a bomb has not yet happened).

@ Sam Greenfield,

The idea as far as I can tell comes from two areas (both statistical in nature and normaly applied to pasive systems).

The first most people come across in “goods inwards test”. A randomly selected sample of the goods (usually 10% or less) is tested. If no defects are found then you have a “reasonable” confidence that the defect rate on the goods is close to (but not) zero. However it assumes that to work the defect rate is going to be significant ie not of very low probability (and a bin bomer is a very very low probability).

The second is mathmatical in nature that sugests “on average” the solution to a large search will be faster if random than an ordered search. Therefor the resources required are on average less. However where as an ordered search is going to find the answer a random search may never find the answer no matter how long it runs…

There are a number of variations on this idea depending on if the item (defect/answer) sort is truly independent or affects others around it. One of these is the shifting comb filter where you test every Nth item where either N is less than the Nyquist point or you repeat the search using an ofset. In either case if you detect an “effected” item you then change the search to one more effective for a smaler more localised set.

Unfortunatly a number of people belive that you can combine these ideas and reduce the result to reduce “security costs”…

Unfortunatly it only “might” work where you have an ordered system like a que where people entering have no fore knowledge and either cannot move place or the sampling is “truly” random. Further that either there is only a single access point, or all access points are continuously randomly sampled.

If any of those constraints are broken then it just will not work.

And as we know that means that it is only deamed to be practical in transit systems where those being sampled will put up with it due to infrequent usage (hence the idea of “frequent traveler” schemes etc for air travel).

Try it on the New York Metro (or that of any other city) in anything like a sample size that is going to be effective and the result is likley to be “civil disobediance”.

Which brings us around to the “human” aspect of this idea. It implicitly assumes that the bomber wishes to survive anonymously (that is place the bomb and get away undetected).

If that constraint does not hold then random searching in areas with low occupancy (ie uncrowded) might just “arguably” reduce the effectivness of the bomber in terms of casualities.

But as we know from airports the search process of any significant size causes a bottle neck and the place becomes more crouded than it might otherwise be…

What random searching at very low levels does do however is “boost confidence” and often catch a lot of other crimes…

So from that asspect it does work…

DaveC April 15, 2009 5:40 PM

The use of Washington Sq park in particular is a variable that needs to be controlled for in future experiments …. maybe the reason the public helpers were so laid back was due to what they bought to smoke in the park 🙂

bob April 16, 2009 6:42 AM

@Guy in NC: Actually, Pope was my destination. But I had used a cellphone to file my flightplan after stopping for fuel (since Pope doesn’t have avgas) which routed it to my home state, not Fayetteville; once in the air I could not get it opened, you cant get into a military base without one and base ops only had UHF radios and I only had VHF. I wound up having to open it via cellphone (which is a challenge itself since at that altitude I must have been hitting 100,000 cellphone towers; plus the inside of a bugsmasher in the air is a similar acoustic environment to being inside a running clothesdryer full of ballbearings). Finally center cleared me to talk to the tower direct and they let me in. Then, since no one was looking, I made one of my prettiest landings ever.

bob April 16, 2009 7:08 AM

@Paul, Sam Greenfield: Well, I dont believe searching everyone would be required to prevent bombings. Merely searching enough people that potential bombers think they have a high likelihood of being detected and thwarted; and therefore go somewhere else/do something else rather than run the risk of wasting a bomb and the preparations that go with it.

@Clive: It has. In about ~50 years we went from being a risk-taking, aggressive, entrepreneurial, forward thinking society where everyone has a chance to excel to a nation of motionless crybabys waiting for someone else to wipe our butts for us because we are afraid to take ANY risks, and we’re too stupid to realize that without risk there can be no reward.

This is compounded by the fact that we have the highest density of Lawyers of any country in the world and lawyers have the effect on society that control rods have on nuclear reactions – they absorb energy causing everything to come to a halt. I have long advocated a law that the losing attorney in any lawsuit (and on BOTH sides of a class-action suit since both parties will lose) should be executed, not only to add some (any) motivation to their performance (which has always reminded me of underwater Tai Chi) but also to thin the ranks so we drop to a reasonable number.

Paul April 16, 2009 10:35 AM


And does that make a difference from a public policy standpoint? Home security systems are a good idea because in the end it doesn’t matter to you if your neighbor gets robbed instead. Increasing the effort when there are other easy targets lowers your risk, which is a good outcome.

Whereas, with the subway if your deterrent intent is simply “risky enough that they move to the next subway station/soft target” then what is society gaining from prohibitive screening costs? I’m not arguing against screening of any kind, I’m just interested in seriously studying what works.

As for your “searching enough people that potential bombers think they have a high likelihood of being detected and thwarted”, you’re assuming the location isn’t important. Sam Greenfield sounded like he thought it was a worthy target, which was the context my comment was framed in. It’s easy to say “x will help because they will figure it’s not worth it and move on”, but that doesn’t really speak to what I was saying. I was simply pointing out that if it was a serious target, a little random screening provides small respite. There are several ways to get around random screening, not the least of which would be sending a flunkie or someone whose family is being held hostage (if they fail, wait a week or two and send another. or just be content with the ensuing panic).

I just now read the rest of yout response to Clive. I do hope your last paragraph is tongue in cheek, otherwise I just wasted my time replying to a psychopath.

Clive Robinson April 16, 2009 9:11 PM

@ bob, Paul,

I think there is something about Lawyers that close contact brings out the psychopath in the best of us (usually when you get the bill 😉

However I don’t agree with outright execution, I’m thinking a little bit at a time, a finger here a toe there.

And you could make it usefull to society most organs we have two of.

So take a kidney etc for transplants and with the latest generation of face transplants it would definatly be a case of “ear today gone to Morow”.

(Yup for a joke like that…)

Clive Robinson April 16, 2009 9:17 PM

Agh I mucked it up…

It should be,

“Ear to pay gone to Morow”…

Ah well it is gone three in the morning here in the UK.

Leave a comment


Allowed HTML <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre> Markdown Extra syntax via

Sidebar photo of Bruce Schneier by Joe MacInnis.