The late great Rick Riolo had a story about this, involving genetic algorithms.
The Air Force gave his team a contract to develop the most fuel efficient drone flight algorithms they could. So they got access to the Air Force's best simulation environment, and set up a genetic algorithm optimization that maximized fuel conservation. A few months later they came back, and discovered that every surviving algorithm had more fuel than it started with. The optimization had found a flaw in the simulation environment.
We're a long way from machine learning of affordances, much less meaning.
My cousin needs a netbook factor with a real keyboard (and no camera, which kiboshes a lot of options) for use in archival research and was considering one. I never asked her whether she bought one though.
You miss the point. Every citizen petitioner has the right to access to the same avenues of petition regardless of their message. You can disallow protests at 4am inside the White House so long as it's a blanket ban. Letting the Society in Favor of Puppies and Kittens have an advocacy event the Green Room before dawn means that any other organization, regardless of how controversial its opinions are, must be given similar consideration (see e.g.: De Jonge v. Oregon or Edwards v. South Carolina).
Indeed, that is also consistent with many theories of Bonobo/Chimpanzee species split and the attendant bifurcation of their social evolution (ie. why the Bonobos fuck and the Chimps fight). The Bonobos, being south of the Congo river (newly formed circa 2m years ago), had little climatic distress to deal with, while the Chimps to the north did, which is why they developed a much more aggressive social organization.
I'm a professional neuroscientist that specializes in vision research with a computational bent. They used all the main stream, state of the art, openly available object recognition algorithms currently in use. Computer vision is a huge market, with many applications, from the DoD to self-driving cars to image-based searches. I doubt some 5-figure prize is going to out perform the best algorithms several distinct industries and academia have managed to create while being funded to the tune of over a billion a year for the last 10 years or so.
These are serious researchers. If you think you think you can get any type of computer vision that significantly outperforms humans on this type of task, there is a unicorn startup and multiple ultra-high profile publications awaiting you.
And just FYI based on your further post: they used two types of convolutional neural nets (see Methods: Model Versions and Parameters).
No, you can't. Stalin's daughter defected from the USSR, and she had a good reason. None of Scalia's descendants will be excluded from political speech, much less imprisoned due to their beliefs.
The best memory I have of Scalia is that when Stephen Colbert gave his infamous White House Correspondents Dinner address, Scalia was laughing his ass off when he was lampooned. I might not agree with the man, but he had a great life lived on his own terms.
With regard to that, I'd highly recommend Fatal Passage by Ken McGoogan, which is a biography of arctic explorer John Rae, who conducted the search for the Franklin expedition (and probably could have rescued them if things had worked out just a little bit differently) and discovered the final link in the northwest passage that Amundsen used 50 years later.
My experience tells me absolutely nothing about whether Canada was being diligent in border screening.
In every single case I crossed a border as a minor it was with free will and full paperwork, so I can't say if the screening was diligent in a way that would uncover a coercive crossing situation.
It also doesn't tell me anything about their screening criteria. Making too many type 2 errors, like I experienced with my grandparents, means that resources are not being effectively directed. And that has costs as well, since after that experience neither my family nor I returned to Canada until after I was an adult.
We had proper documentation (notarized permission letter, notarized copy of my birth certificate, passport)... The reason I say it's a Canadian neurosis (albeit semi-seriously) is that I made many other (25 or so) border crossings as a minor without parental accompaniment (many more with one parent), and never had an issue anywhere else (including Mexico, France, Italy, England, Austria, US, Switzerland, Cayman Islands).
I made at least 25 border crossings (mostly foreign-foreign) while a minor and not under the care of my parents or with a cousin who was not under the care of their parents (including a solo one to Mexico to meet up with some family friends when I was 10, although IIRC that was pretty paperwork heavy with the airline)... Canada was the only time there was any trouble.
Many years ago, when I was 6 or 7, my grandparents drove me up to Prince Edward Island for a vacation. Getting through the border into Canada took about 45 minutes, with my grandparents getting grilled and the agents asking me about a dozen times whether I wanted to be with them and whether my parents knew where I was etc etc.. Getting back into the US took about 2 minutes. It seems like it's a Canadian neurosis.
That's mentioned in the IEEE Spectrum article (which by the way is about the most clearly written article on an early prototype technology that I've ever read). The problems are: -Too high voltage; can be mitigated by better geometry (probably). -Insufficient simulations at present for improving the geometry, with the caveat that getting better performance (voltage-wise) might compromise durability. -Because of the above, they don't have a good set of design rules to produce an integrated circuit. They're hopeful about this step, since the technique uses well established CMOS technology and there are many tools available.
Their next targets are things like gyroscopes and accelerometers. I'd say on the whole this strikes me as realistic and non-sensational. But if anybody knows better, I'd like to hear why.
This is no different from trying to come up with ways of measuring scholars' intellectual impact using citation metrics, like the h-factor or the many recent successors to it, which try to repair the weaknesses in a fatally flawed idea. It makes no distinction between positive and negative citation, and it ignores the raw fact of historical precedence, while preserving every historical bias a culture may have.
The most influential people in world history, at least the very top-tier, isn't particularly debatable, but yet this list failed to capture it. In alphabetical order (and assuming they all existed):
Aristotle Buddha Confucius Homer Jesus Lao Tzu Muhammad Plato Ved Vyasa
Then there's the next tier, which include people like Al-Hazan, Alexander, Augustine, Einstein, Genghis, Hammurabi, Imhotep, Newton, Linnaeus, Peter (of Russia), Shakespeare, Suleiman, Zeami Motokiyo etc etc, since I'm sure the further I try to extend the list, the more it would converge with my cultural history.
While unsupervised algorithms can often find interesting things in high-dimensional data, they aren't interpret-able without some expert knowledge.. and if you don't have the 9 entries I mentioned above in your top 20 at least, you can toss the method.
This is a well known weakness with back-propagation based learning algorithms. In the learning stage it's called Catastrophic interference, in the testing stage it manifests itself by mis-classifying similar inputs.
Yikes, that sounds like a terrible experience. My sympathies to your sister in law and the whole family.
There are several methods available, most prominently implanting arrays of electrode over pre-motor cortex, which can then be decoded online and used to control a computer pointer.
You might want to contact Frank Guenther at BU. Who has worked on this for several years, and has started the Unlock Project particularly for people in your sister in law's situation.
There's a huge bias towards using exclusively male mice in many types of research, and the issue of higher variance in female rodent behavior (due to estrous cycle issues, among others) is well known (see eg: pdf).
There are also related problems more generally with stress and over-training in neuroscience. Experienced investigators are able to produce a much less stressful working environment for animals, so they tend to get different results from neophyte investigators even when following the same protocol. This shows up a lot when a different lab tries to replicate the work of an experienced post-doc and gets null results for the first 6 months then suddenly is able to replicate everything. Thus often is attributed to 'correcting' the protocol (often with extensive communication with the previous lab) when often I think the change is attributable to the investigator in the replicating lab becoming experienced enough to relieve stress (I don't have a great link for this, mostly just an observation from having been around quite a few labs).
Over-training is also a problem, since it often takes thousands (sometimes well into the hundreds of) to train animals in complex cognitive tasks, and it's well known from experiments in humans (and a few in non-human primate and rodent) that neural responses shift profoundly between 'trained' and 'over-trained' states, say between amateur and professional ballerinas watching videos of ballet.
However, these issues are a much bigger problem in pre-clinical research than in basic research. Our understanding of the brain is sufficiently limited that the effects we're used to seeing in basic research questions swamp the potential modulation from gender, stress and training factors (unless you're talking about stress research specifically, but they're pretty careful about controlling for these types of effect). The issue with pre-clinical research is that often the difference between the current treatment and proposed treatment is only a few percent (note: if valid, this can mean thousands of lives saved or hugely improved), and so failing to identify and control for factors such as researcher or mouse gender can overwhelm the supposed primary result.
To destroy the world's carrying capacity for humanity we have to opt-in to global thermonuclear war. To destroy that same capacity through climate change simply requires that a modest proportion of the world's population does not opt-out of mitigating carbon release (the Pareto-optimal level of GDP is pretty small, actually, around 2% of global GDP).
Actually, since neurons have functional homeostatic pruning and nonlinear membrane responses, there are quite large values of zero when we're recording firing rate.
With regard to question 2) No. Question 1 is an ongoing field of research. Some of the work that I've found helpful in approaching the question: -The Computational Beauty of Nature (Gary William Flake) -Barriers and Bounds to Rationality (Peter Albin; there are free pdf copies available online). -A New Kind of Science (Stephen Wolfram; also available free online).
The linked article was horribly written. I'll give a shot at trying to explain it (or rather, a really, really simplified version).
Two of the fundamental problems that neural circuits must solve are the noise-saturation dilemma and the stability-plasticity dilemma. The first is best explained in the context of vision. Our visual system is capable of detecting contrast (ie. edges) over a massive range of brightness, spanning a space of about 10^10. Given that neurons have limited firing rates (typically between 0 and 200hz), there needs to be some normalization criteria that allows useful contrast processing over massive variations in absolute input (more on this later). The stability-plasticity dilemma is that the brain needs to be sufficiently flexible to learn based on a single event (let's say, touching a hot stove is a bad idea), but once learned memories have to be sufficiently stable to last the rest of a creatures' life span.
The stability-plasticity dilemma implies that neural circuits must operate in at least two (as I said, very simplified) distinct states, a "resting" or "maintenance" state, and a "learning" state, and that there is a phase-transition point in between them. Furthermore, these states need to have the following properties regarding stability: 1) the learning state must collapse into the maintenance state in the absence of input (otherwise you get epilepsy). 2) reasonable stimulation (input) during the resting state must be able to trigger a phase change into the learning state (or you become catatonic).
Many circuits/mechanisms have been proposed to explain how the brain solves these dilemmas. Most of them involve the definition of a recurrent neural network using some combination of gated-diffusion and oscillatory dynamics to fit well known oscillatory and wave-based dynamics that have been recorded in neural circuits. Some of these models employ intrinsic learning using a learning-rule (ie. self-organized maps) while others are fit by the researcher. One key point about this class of models (as opposed to the TFA approach) is that they have a macro-circuit architecture specified by the modeler. Typically these models are at least somewhat sensitive to parametric perturbation.
TFA describes another approach, which comes out of research on cellular automata done by Ulam, von Neumann, Conway and Wolfram. This approach posits that parametric stability and macro-circuit organization is only loosely important so long as the system obeys a certain set of rules regarding local interaction (could also be through of as micro-circuit) because it will self-organize to a point of 'critical stability'. In the the two-state model described above, this approach predicts that neural circuits are always at a state of 'critical stability' where maintenance occurs through frequent small perturbations or avalanches, and any new input will trigger a large avalanche, causing learning. Bak has proposed this as a general model of neural circuit organization. One trademark of these type of models is that they show 'scale free' or 'power law' behavior, where the size of an event is inversely proportional to its frequency by some exponential function. Some recent data has shown power-law dynamics in neural populations (a lot of other data doesn't show power-law dynamics).
One big problem with the critical stability hypothesis is that it doesn't deal well with the noise-saturation dilemma: it needs to cause the same general size of avalanche whether it's hit by one grain of sand, or 10^10 grains of sand.
None of this is particularly new, neural-avalanches (albeit in a different context) were postulated in the early 70s. Could some systems in the brain exploit self-organized criticality? Sure, but there is a lot of data out there that's inconsistent with it being the primary method of neural organization.
Having recently been in Spain (with my unlocked iphone 4 in tow), I can tell you that the support for iphones (at least in Barcelona) is terrible. It took trips to 4 different stores to find an iphone 4 compatible prepaid mini-sim (if I had the iphone 5, I would have been SOL and had to pay for roaming data from my US plan). None of those stores prominently placed iphones (although they were available, at least through vodaphone, even the 5 new, but you couldn't use a prepaid sim in it).
I tend to think that the issue is that Spain has a really fractured retail environment, both with a lot of providers (vodaphone/movistar/orange/yoigo and lots of 3rd party options) and with a lot of kiosk type stores. Vodaphone has their own retail outlets, but most of the others seemed to be based in malls, and the malls in turn seemed to have one 'basket' of stores, depending on who owns the mall. During my search for a mini-sim for example, I was sent on a goose-chase from store to store with directions that turned out to be pretty approximate (wrong address, but within about 300 meters of the correct address).
Given that retail environment, I think it's pretty natural that android, with its myriad of slightly customized, provider branded phones etc, fares a lot better than iOS at the moment... People want something that can be supported by their local mall/kiosk.
Slashdot Mirror
The late great Rick Riolo had a story about this, involving genetic algorithms.
The Air Force gave his team a contract to develop the most fuel efficient drone flight algorithms they could. So they got access to the Air Force's best simulation environment, and set up a genetic algorithm optimization that maximized fuel conservation. A few months later they came back, and discovered that every surviving algorithm had more fuel than it started with. The optimization had found a flaw in the simulation environment.
We're a long way from machine learning of affordances, much less meaning.
Have you looked at:
the GPD Pocket 7.0? It's an atom that can run win10, 8/128gb, 7 in.
https://www.windowscentral.com...
My cousin needs a netbook factor with a real keyboard (and no camera, which kiboshes a lot of options) for use in archival research and was considering one. I never asked her whether she bought one though.
You miss the point. Every citizen petitioner has the right to access to the same avenues of petition regardless of their message. You can disallow protests at 4am inside the White House so long as it's a blanket ban. Letting the Society in Favor of Puppies and Kittens have an advocacy event the Green Room before dawn means that any other organization, regardless of how controversial its opinions are, must be given similar consideration (see e.g.: De Jonge v. Oregon or Edwards v. South Carolina).
Indeed, that is also consistent with many theories of Bonobo/Chimpanzee species split and the attendant bifurcation of their social evolution (ie. why the Bonobos fuck and the Chimps fight). The Bonobos, being south of the Congo river (newly formed circa 2m years ago), had little climatic distress to deal with, while the Chimps to the north did, which is why they developed a much more aggressive social organization.
I'm a professional neuroscientist that specializes in vision research with a computational bent. They used all the main stream, state of the art, openly available object recognition algorithms currently in use. Computer vision is a huge market, with many applications, from the DoD to self-driving cars to image-based searches. I doubt some 5-figure prize is going to out perform the best algorithms several distinct industries and academia have managed to create while being funded to the tune of over a billion a year for the last 10 years or so.
These are serious researchers. If you think you think you can get any type of computer vision that significantly outperforms humans on this type of task, there is a unicorn startup and multiple ultra-high profile publications awaiting you.
And just FYI based on your further post: they used two types of convolutional neural nets (see Methods: Model Versions and Parameters).
They did consider false positives.. They had a catch category (see page 4, last par) and human's did extremely well at it (see 3rd par, page 5).
No, you can't. Stalin's daughter defected from the USSR, and she had a good reason. None of Scalia's descendants will be excluded from political speech, much less imprisoned due to their beliefs.
The best memory I have of Scalia is that when Stephen Colbert gave his infamous White House Correspondents Dinner address, Scalia was laughing his ass off when he was lampooned. I might not agree with the man, but he had a great life lived on his own terms.
With regard to that, I'd highly recommend Fatal Passage by Ken McGoogan, which is a biography of arctic explorer John Rae, who conducted the search for the Franklin expedition (and probably could have rescued them if things had worked out just a little bit differently) and discovered the final link in the northwest passage that Amundsen used 50 years later.
My experience tells me absolutely nothing about whether Canada was being diligent in border screening.
In every single case I crossed a border as a minor it was with free will and full paperwork, so I can't say if the screening was diligent in a way that would uncover a coercive crossing situation.
It also doesn't tell me anything about their screening criteria. Making too many type 2 errors, like I experienced with my grandparents, means that resources are not being effectively directed. And that has costs as well, since after that experience neither my family nor I returned to Canada until after I was an adult.
We had proper documentation (notarized permission letter, notarized copy of my birth certificate, passport)... The reason I say it's a Canadian neurosis (albeit semi-seriously) is that I made many other (25 or so) border crossings as a minor without parental accompaniment (many more with one parent), and never had an issue anywhere else (including Mexico, France, Italy, England, Austria, US, Switzerland, Cayman Islands).
I made at least 25 border crossings (mostly foreign-foreign) while a minor and not under the care of my parents or with a cousin who was not under the care of their parents (including a solo one to Mexico to meet up with some family friends when I was 10, although IIRC that was pretty paperwork heavy with the airline)... Canada was the only time there was any trouble.
Many years ago, when I was 6 or 7, my grandparents drove me up to Prince Edward Island for a vacation. Getting through the border into Canada took about 45 minutes, with my grandparents getting grilled and the agents asking me about a dozen times whether I wanted to be with them and whether my parents knew where I was etc etc.. Getting back into the US took about 2 minutes. It seems like it's a Canadian neurosis.
That's mentioned in the IEEE Spectrum article (which by the way is about the most clearly written article on an early prototype technology that I've ever read).
The problems are:
-Too high voltage; can be mitigated by better geometry (probably).
-Insufficient simulations at present for improving the geometry, with the caveat that getting better performance (voltage-wise) might compromise durability.
-Because of the above, they don't have a good set of design rules to produce an integrated circuit. They're hopeful about this step, since the technique uses well established CMOS technology and there are many tools available.
Their next targets are things like gyroscopes and accelerometers. I'd say on the whole this strikes me as realistic and non-sensational. But if anybody knows better, I'd like to hear why.
This is no different from trying to come up with ways of measuring scholars' intellectual impact using citation metrics, like the h-factor or the many recent successors to it, which try to repair the weaknesses in a fatally flawed idea. It makes no distinction between positive and negative citation, and it ignores the raw fact of historical precedence, while preserving every historical bias a culture may have.
The most influential people in world history, at least the very top-tier, isn't particularly debatable, but yet this list failed to capture it. In alphabetical order (and assuming they all existed):
Aristotle
Buddha
Confucius
Homer
Jesus
Lao Tzu
Muhammad
Plato
Ved Vyasa
Then there's the next tier, which include people like Al-Hazan, Alexander, Augustine, Einstein, Genghis, Hammurabi, Imhotep, Newton, Linnaeus, Peter (of Russia), Shakespeare, Suleiman, Zeami Motokiyo etc etc, since I'm sure the further I try to extend the list, the more it would converge with my cultural history.
While unsupervised algorithms can often find interesting things in high-dimensional data, they aren't interpret-able without some expert knowledge.. and if you don't have the 9 entries I mentioned above in your top 20 at least, you can toss the method.
This is a well known weakness with back-propagation based learning algorithms. In the learning stage it's called Catastrophic interference, in the testing stage it manifests itself by mis-classifying similar inputs.
Shenoy's group is also working with patients these days, but I think they're focused on ALS rather than locked-in.
Yikes, that sounds like a terrible experience. My sympathies to your sister in law and the whole family.
There are several methods available, most prominently implanting arrays of electrode over pre-motor cortex, which can then be decoded online and used to control a computer pointer.
See for example:
http://www.youtube.com/watch?v...
You might want to contact Frank Guenther at BU. Who has worked on this for several years, and has started the Unlock Project particularly for people in your sister in law's situation.
There's a huge bias towards using exclusively male mice in many types of research, and the issue of higher variance in female rodent behavior (due to estrous cycle issues, among others) is well known (see eg: pdf).
There are also related problems more generally with stress and over-training in neuroscience. Experienced investigators are able to produce a much less stressful working environment for animals, so they tend to get different results from neophyte investigators even when following the same protocol. This shows up a lot when a different lab tries to replicate the work of an experienced post-doc and gets null results for the first 6 months then suddenly is able to replicate everything. Thus often is attributed to 'correcting' the protocol (often with extensive communication with the previous lab) when often I think the change is attributable to the investigator in the replicating lab becoming experienced enough to relieve stress (I don't have a great link for this, mostly just an observation from having been around quite a few labs).
Over-training is also a problem, since it often takes thousands (sometimes well into the hundreds of) to train animals in complex cognitive tasks, and it's well known from experiments in humans (and a few in non-human primate and rodent) that neural responses shift profoundly between 'trained' and 'over-trained' states, say between amateur and professional ballerinas watching videos of ballet.
However, these issues are a much bigger problem in pre-clinical research than in basic research. Our understanding of the brain is sufficiently limited that the effects we're used to seeing in basic research questions swamp the potential modulation from gender, stress and training factors (unless you're talking about stress research specifically, but they're pretty careful about controlling for these types of effect). The issue with pre-clinical research is that often the difference between the current treatment and proposed treatment is only a few percent (note: if valid, this can mean thousands of lives saved or hugely improved), and so failing to identify and control for factors such as researcher or mouse gender can overwhelm the supposed primary result.
To destroy the world's carrying capacity for humanity we have to opt-in to global thermonuclear war. To destroy that same capacity through climate change simply requires that a modest proportion of the world's population does not opt-out of mitigating carbon release (the Pareto-optimal level of GDP is pretty small, actually, around 2% of global GDP).
Actually, since neurons have functional homeostatic pruning and nonlinear membrane responses, there are quite large values of zero when we're recording firing rate.
With regard to question 2) No.
Question 1 is an ongoing field of research. Some of the work that I've found helpful in approaching the question:
-The Computational Beauty of Nature (Gary William Flake)
-Barriers and Bounds to Rationality (Peter Albin; there are free pdf copies available online).
-A New Kind of Science (Stephen Wolfram; also available free online).
The linked article was horribly written. I'll give a shot at trying to explain it (or rather, a really, really simplified version).
Two of the fundamental problems that neural circuits must solve are the noise-saturation dilemma and the stability-plasticity dilemma. The first is best explained in the context of vision. Our visual system is capable of detecting contrast (ie. edges) over a massive range of brightness, spanning a space of about 10^10. Given that neurons have limited firing rates (typically between 0 and 200hz), there needs to be some normalization criteria that allows useful contrast processing over massive variations in absolute input (more on this later). The stability-plasticity dilemma is that the brain needs to be sufficiently flexible to learn based on a single event (let's say, touching a hot stove is a bad idea), but once learned memories have to be sufficiently stable to last the rest of a creatures' life span.
The stability-plasticity dilemma implies that neural circuits must operate in at least two (as I said, very simplified) distinct states, a "resting" or "maintenance" state, and a "learning" state, and that there is a phase-transition point in between them. Furthermore, these states need to have the following properties regarding stability:
1) the learning state must collapse into the maintenance state in the absence of input (otherwise you get epilepsy).
2) reasonable stimulation (input) during the resting state must be able to trigger a phase change into the learning state (or you become catatonic).
Many circuits/mechanisms have been proposed to explain how the brain solves these dilemmas. Most of them involve the definition of a recurrent neural network using some combination of gated-diffusion and oscillatory dynamics to fit well known oscillatory and wave-based dynamics that have been recorded in neural circuits. Some of these models employ intrinsic learning using a learning-rule (ie. self-organized maps) while others are fit by the researcher. One key point about this class of models (as opposed to the TFA approach) is that they have a macro-circuit architecture specified by the modeler. Typically these models are at least somewhat sensitive to parametric perturbation.
TFA describes another approach, which comes out of research on cellular automata done by Ulam, von Neumann, Conway and Wolfram. This approach posits that parametric stability and macro-circuit organization is only loosely important so long as the system obeys a certain set of rules regarding local interaction (could also be through of as micro-circuit) because it will self-organize to a point of 'critical stability'. In the the two-state model described above, this approach predicts that neural circuits are always at a state of 'critical stability' where maintenance occurs through frequent small perturbations or avalanches, and any new input will trigger a large avalanche, causing learning. Bak has proposed this as a general model of neural circuit organization. One trademark of these type of models is that they show 'scale free' or 'power law' behavior, where the size of an event is inversely proportional to its frequency by some exponential function. Some recent data has shown power-law dynamics in neural populations (a lot of other data doesn't show power-law dynamics).
One big problem with the critical stability hypothesis is that it doesn't deal well with the noise-saturation dilemma: it needs to cause the same general size of avalanche whether it's hit by one grain of sand, or 10^10 grains of sand.
None of this is particularly new, neural-avalanches (albeit in a different context) were postulated in the early 70s. Could some systems in the brain exploit self-organized criticality? Sure, but there is a lot of data out there that's inconsistent with it being the primary method of neural organization.
Having recently been in Spain (with my unlocked iphone 4 in tow), I can tell you that the support for iphones (at least in Barcelona) is terrible. It took trips to 4 different stores to find an iphone 4 compatible prepaid mini-sim (if I had the iphone 5, I would have been SOL and had to pay for roaming data from my US plan). None of those stores prominently placed iphones (although they were available, at least through vodaphone, even the 5 new, but you couldn't use a prepaid sim in it).
I tend to think that the issue is that Spain has a really fractured retail environment, both with a lot of providers (vodaphone/movistar/orange/yoigo and lots of 3rd party options) and with a lot of kiosk type stores. Vodaphone has their own retail outlets, but most of the others seemed to be based in malls, and the malls in turn seemed to have one 'basket' of stores, depending on who owns the mall. During my search for a mini-sim for example, I was sent on a goose-chase from store to store with directions that turned out to be pretty approximate (wrong address, but within about 300 meters of the correct address).
Given that retail environment, I think it's pretty natural that android, with its myriad of slightly customized, provider branded phones etc, fares a lot better than iOS at the moment... People want something that can be supported by their local mall/kiosk.
What was the different solution? (I've also wrecked quite a few shirts in my time)