I was at MIT for 5 years (bachelor's and master's) and, while I would agree that some of your criticisms have truth, by and large it's an excellent institution.
Some of the professors barely try to teach. But the majority I encountered (in the physics and electrical engineering departments) were good to excellent. In the physics department, especially, most of the recitation instructors are professors, not TAs. The TAs do grading, but professors lead the discussions and show up for office hours. If you bother to show up for those, most are very happy to help you with anything that's giving you trouble.
Class size? A few intro classes were large, but beyond that it's mostly small (25-30 range) and for larger classes, they break down into smaller recitations. As I mentioned, in physics, these are usually lead by professors. This is also often true in the EE department.
MIT students, even undergrads, do get to work on cutting-edge research. That is not a lie. However, it's not automatic. You have to have the ability to contribute to the project, and many undergrads need to learn through classes and/or experience to be able to contribute. Sometimes that learning involves grunge work. That's not a problem with the school, it is a fact about research. Every professor, every grad student, every researcher has gone through their share of the grunge work.
If you have the experience/ability coming in, you can find a professor who'll give you a job doing cutting edge stuff right away. If not, you can find one who will find a way for you to work in his lab while you learn what you need to know to make meaningful contributions. The opportunity is there in a way that it simply is not at some schools.
As for the Media Lab, I worked there for several years. If you really think the best that came out of there is the alarm clock, you should realize that headlines don't tell the whole story. A lot of good work was and is done there. A large number of start-up companies came out of the Lab. Some very interesting work on wearable and embedded computing was done there. A lot of work on spreading modern technology to the developing world was and is going on there. It's not all silly consumer products.
Finally, the high-school team (I thought they were from CA, not TX, but whatever) who won the underwater vehicle competition was awesome. They beat a lot of teams through ingenuity and hard work. MIT wasn't the only team these guys beat -- your example says a lot about the high-schoolers and very little about MIT. MIT has had enough success at various competitions (including their Orca underwater autonomous vehicle team -- also a submarine but run entirely by onboard software) that pointing to a single case where they didn't win doesn't say much.
MIT, like any school, has its problems and ultimately the success of an institution really depends more on the people who attend than on the institution itself. A bright person will do great things whatever school they go to. Thanks in part to reputation and in part to an "installed base" of bright people, MIT tends to attract a pretty large percentage of bright people.
Yep, it was not long after that that Covad went under. Speakeasy managed to hang on without them though, although I don't remember the details of that transition.
One other interesting thing is that while I was waiting the 3 months to get online, Speakeasy gave me free dial-up. Not great, but it was enough to get me by until I could get the high-speed line. It probably cost them very little (just using otherwise unused resources) and has helped them get a long-term customer.
Sorry -- no offense was intended, I'm sure there are many tech support people capable of giving better support than their organization will let them. I really meant to get at the fact that Speakeasy as an organization lets them do their jobs to the best of their ability.
There's some risk, but I think the techs are pretty wise about their suggestions -- I haven't encountered anything like swapping out parts, but I am pretty confident that if I did so on my own and told them the results, they would be able to use these in their diagnosis of a problem.
I've been with Speakeasy for almost 5 years and would not consider another provider. When I first signed up, it was a bit rocky getting online (took about 3 months while they coordinated with the local phone company and the Covad middle layer) but since then I have moved from coast to coast a couple times and had no trouble bringing my service with me. Other than at the very beginning, I've had virtually NO downtime -- ran into a little trouble when my DSL modem started failing, but they can hardly be held responsible for that.
Furthermore, they have eminently reasonable policies. You are allowed to use your DSL connection as a full and proper connection to the internet -- they have no arbitrary restrictions on services you can run. It's not a download-only pipe like the Telco and cable companies want to sell you. They do their best to support you running any OS you want and the techs I've spoken with are actually sharp enough to help you outside of a script. Not only that, but they have some authority to do what it takes to get the job done. All the while, you have access to the communications logs between the Speakeasy techs and the local telco and other parties involved in providing the line.
The existence of a company like this, IMO, indicates that there is demand for services the telcos are unwilling or unable to provide. They footed part of the bill to run the wires to your house, so they should get some return. That's why Speakeasy *rents* the line from them and adds their markup on top. There's no reason that the telco needs to bundle ISP services with the telco line. If I think the telco has a good pipe but offers crappy ISP service, it makes sense that I can opt out of their ISP offerings. The architecture is already in place to let me do this.
Furthermore, the telcos did not foot the entire bill for running the wires. Government assistance and tax dollars helped set up the network. They're part of the public infrastructure and they knew that when they got in the business. They therefore have responsibilities not only to their shareholders, but to the society that they bargained with to get their business in the firstplace.
Yeah -- New Egg has always been good when I have used them. The last time I was there, they were very clear about their review policy. The highly visible disclaimer basically said what people are saying here: "We are a commercial site selling products. We are not an independent review site. If you want independent review, look somewhere else." Even if their reviews are all more or less positive, they still often have useful information.
Uh, nice of you to correct his correct statement, but perhaps you would like to re-read his statement and ask yourself whether you understood it the first time through.:-)
In California, auto insurance is not specifically required. You can opt to post a bond that exceeds the minimum required insurance level, as you suggest.
However, the minimum insurance is pretty low. With the 100/300/100 insurance I buy, the maximum possible insurance payout will pretty much always exceed what I've put into it. So even if I had the funds for the minimum, it still probably makes sense to buy the insurance.
Insurance is not really a waste if the mean net cost isn't break even. It's protection against standard deviations. $100k+ accidents are not common, so the average expected cost is low -- however you need to be able to cover the whole cost in the rare case where your number gets called. Insurance helps to smooth the cost over a large number of drivers, and it takes work to manage this process.
I don't think it's fair to call it a ripoff based on the justification that profit is involved. Profit for a service is fair -- the insurance agents have to put food on their families too.
2) (and this is the tenuous one) you can get out of a speeding ticket if you truely are 'going with the flow of traffic', but I don't think that's ever worked [...]
Are there really states where this is supposed to be a defense? In California they are very specific that the "flow of traffic" is not a legal defense for speeding. I think this is similar in Indiana and Massachusetts and I'd assumed it was the case for other states as well.
So the light turns green. (my light. the turning light. of a 10 lane highway +2 turning lanes)
This just makes me think of highway moderation. e.g., SR-137 (Score: +2, turning lanes)
Woah there, QuantumG, I think that horse you're riding is a little high for you.
I think I asked some relevant questions about a statement you made and you respond by asking me to STFU. Now you're attacking me (and others) as someone who "[does]n't know shit talking out of [his] ass"?
That's great. You've convinced me that you're an expert on the topic. Well, either that or your HCI class prof mentioned that ctrl-c and ctrl-v are monstrosities and you're just parroting that on slashdot.
Seriously, I would like to know what it is about a keyboard shortcut that makes it an HCI monstrosity. I honestly don't see how you can defend that position, given that it is there as an accelerator for a process that can be done in other ways.
And don't try to pull academic rank here, pal, you don't know anything about who you're talking to. Furthermore, HCI is not a topic that can or should be preached from on high, it's just not that kind of a field.
Personally, I have worked very closely with a lot of researchers in this area and while they have a lot of neat ideas, ultimately the measure you're after is does it make interaction easier. What's the measure of that? You go and ask one of these "people who don't know shit" to use it in the real world and see how it turns out. Theories are great, but ultimately it comes down to a very practical result. The HCI researcher has his theory, but these "people who don't shit" are the only ones who actually "know" whether the HCI theory is right.
Also, I hope you understand that there is a place for _transparent_ usability and there is a place for _efficient_ usability. These are distinct goals that are important for different applications. If I'm going to use a kiosk to check in at the airport, I want the interface to tell me how to use it without having to learn anything, even if an experienced user could operate faster. However, if I'm in a factory controlling an industrial machine to assemble parts of a car, I'm willing to tolerate a much higher learning curve if the end result is I'm 20% more efficient thanks to a pared-down interface.
In the first scenario, the total time I spend on the task is short -- it would be silly to spend 10 minutes learning how to use a system to perform a task I can do in 5 minutes. In the second scenario, I'll be spending 8 hours a day every day performing this task. Spending 2 weeks training on the system is a good investment -- if I am 20% more efficient, I'll recover that 2 week training investment after 10 weeks of work and after that I'm permanently better off.
Control-C and Control-V fall into this latter category. I spend a LOT of time editing text on a computer, and I am MUCH faster as a result of these keystrokes. It is worth my time to learn the tool. It increases the _efficiency_, even if it's not _transparent_ (but note that it doesn't even reduce the transparency -- if you're new to the computer, there are more transparent alternatives that let you perform the same operation).
Anyway, I really am interested in why you might label ctrl-C and ctrl-V as monstrosities and/or why you disagree with what I've said above. Additionally, I think you owe me and the others an apology for your obnoxious response, but I won't hold my breath.
Why? Because it's not the analog of a real-world operation? Would you feel better if you had a little dedicated copy and paste button on your keyboard, or does using the mouse just make things automatically more usable?
You have to learn how to use a tool. A computer is a tool. Copy and paste are things you do when you use it. Nearly every program that supports copy and paste uses ctrl-c and ctrl-v and many keyboards even print "copy" and "paste" as hints. Ok, the Mac goes and uses the "Apple" key instead of ctrl, but it's the same idea.
Furthermore, they are very convenient buttons to press with your left hand while mousing with your right. Not perfect, utterly transparent design, but eminently pragmatic and *consistent*. That sounds usable to me.
I am rather appalled at the attitude of an ISP that sharing your connection is somehow stealing from them. This is one of the many reasons I am so happy with SpeakEasy for my DSL. They not only allow you to share your connection freely, they will help you charge your neighbors if you provide that service. You provide the tech support and bandwidth, they will reduce your bill and manage billing those you share your connection with.
It makes sense -- they're selling you bandwidth, and how you use that is up to you.
I've had a lot of problems interoperating with MS Word. Documents frequently come out with formatting that's not quite right -- text that was clearly supposed to fit on one page spills onto the next page (e.g., a signature line for a contract is on the next, otherwise empty, page).
As an experiment, I recently tried writing a technical document in a combination of Word:Mac and OO.o on linux. It was a nightmare. Figure and table caption references appeared to be almost completely broken -- they would not update if numbering changed. I had a date field at the top to automatically date stamp the document -- OO.o would repeatedly turn this into junk if the document was saved as a.doc, so I had to change it to a manually-typed date.
Furthermore, the table editing features in OO.o were really complete crap. It was next to impossible to get it formatted as I wanted. Once I did, if I changed anything, it was a complicated process to get it to look nice.
I've long been a supporter of OO.o, and I probably would continue to use it for short documents. From a comparison with Word:Mac, though, its interface is really below par. In the end, I converted my document to LaTeX, and I expect that I'll use that for any non-trivial writing I do in the future.
This was done in one case in a system whose password check routine took a pointer to the password to check and then ran strcmp() or the equivalent against the password IN THE ORIGINAL LOCATION. This could be cracked by aligning the password to check against a page boundary. If the strcmp() ran past the page boundary, a large delay would be generated while the missing page was fetched.
This would work on a fast processor, but relied on being able to control what parts of memory were paged to disk, which is harder to do these days.
I use Firefox normally, but there are a few pages (the bungie.net halo forums, in particular) that I have to use Konqueror for. Firefox renders the funky animated crap they insist on adding to their forum excruciatingly slowly. Konqueror is snappy.
IMO, the radio series is by far the best of the group. You can get it on CDs from Amazon.co.uk. Here's a link to the first series ("Primary Phase") and you can follow to the second series ("Secondary Phase") from there. They seem to have a Tertiary Phase now, too, but it's not part of the original radio stuff.
By the way, there is a complete set available from amazon.com (the US version) but based on the reviews, the audio quality is not so great. I got the CDs from Amazon UK a few years ago (I am in the US) and they are great.
I don't know, I find jacket-and-tie-only occasions to be pretty annoying. Analogies between real life and the internet are always weak, but your analogy is a bit more apt. Still, I think the restaurant is within its rights to refuse service.
Unless you need the earplugs because of a legally recognized handicap -- then they'd have to accommodate you. So there should be an exemption for seeing-eye earplugs. Likewise, a web site operator should accommodate someone with a browser that blocks ads due to a handicap, such as a reader for the visually impaired.
Just this morning, I noticed that my previously favorite travel website now has a pop-up ad that circumvents the standard pop-up blocking in Firefox (not AdBlock or whatever the extension is). I sent them feedback letting them know that as long as they were not respecting the clear wish of their customer not to be bothered with a pop-up, I would no longer be using their services. I could (and may) install the AdBlock, but I won't return to their site until they fix the real problem which is disregard for their customers wishes. If they decide they'd rather have the ads than uptight customers like me, then I'll use other services.
Much as I can opt out of using their service if I am unwilling to put up with ads, the service provider should be able to opt out of providing me with a service if I am unwilling to accept their terms. It's no different than a a fancy restaurant with a dress code that requires a jacket and tie. If you want to wear a t-shirt and sandals while you eat, go somewhere else.
However, unless they notify me of these terms, there is and should be no rule that says I must view every bit of data they throw at my web browser. If they do notify me, then I should respect their wishes. I don't see it as a "social contract" so much as common sense and courtesy.
Um, the generosity you refer to was precisely the availability of the free license. It has everything to do with licenses, even if it doesn't use the word.
Don't discount the necessity of knowing buzzwords if you want to succeed in a field. If you want to contribute to a complex field, you MUST know the buzzwords (i.e., jargon, the specialized language of the field) and you MUST use them correctly. Otherwise, you're wasting your readers' time -- time they should be using to understand your carefully crafted argument, not trying to figure out what basic concept you are talking about.
You're right, that's not ALL there is to it -- obviously you also need to learn to make the arguments and generally write well. However, it's not unreasonable to get a poor grade in a sociology course because you turned in a well-crafted paper that didn't use sociological vocabulary.
I'm not attempting to bring questions of higher powers into the question.
What I am saying is that interactions between neurons in a brain are much much much more complex and complicated than the "inspired cartoon" that is a neural network. The connection diagram of an NN is similar to that of neurons in the brain, and the notion that one node responds to a signal from others is also similar. However, the mechanisms going on in the brain are much more complicated.
I don't disagree with most of your post, but I think you overstated the biological basis for neural networks.
The article states that the software is not used for all the papers -- final papers are read and graded by hand. I'm not sure whether that means several versions of each paper are assigned (draft, edit, then final) or whether final project type papers are the only ones graded by hand, but that seems reasonable to me -- it allows more papers to be assigned. Even if they're not graded as insightfully as by a human, the way you get better at writing is to write a lot.
As someone else mentioned, it's very helpful to get detailed feedback about your writing. However, you don't NEED that feedback at every step to get better. Just going through the process, thinking through the arguments, and writing them down forces you to learn. If you are writing something, even just for yourself, every week, you will naturally improve. Plus, you will gain the ability to do so quickly -- a necessary skill for academia.
Sure, if you can't write AT ALL, then you are going to need a lot of feedback to even get your papers to a coherent level -- you're not going to be able to use self-directed feedback at all. But if you can't write coherently, you really shouldn't be learning that skill in a sociology class. You should be learning the particulars of writing pertinent to that discipline, but mostly using the papers as a way to learn the sociology material.
Overall, I think this seems reasonable if used judiciously, and it sounds to me like the professor does so.
You overstate the similarity between a neural net framework and the actual operation of a brain. A neural net is better characterized as a cartoon vaguely inspired by the structure of an animal brain rather than "based on the fundamental theory of how animal brains (including human brains) operate."
You are correct that a true counterexample in some sense disproves a theory, but in practical terms, you're mistaken on two counts.
First, the implication of an experiment is rarely so black and white. Not all experiments are as concrete as the examples you find in textbooks. Many experiments in astrophysics, for example, rely on enormous extrapolation from the data using a complex model. Even when the data comes in, it is difficult to determine conclusively whether it contradicts the theory. Often, individual experiments only rule out small regions of a large parameter space. A combination of several experiments may be enough to rule out that model if they share no common allowed region of paramter space.
Gravity in particular is notoriously difficult to test experimentally. In our region of the universe, Newtonian gravity is correct to the 10^-8 level. GR picks up those billionth-part corrections. GR has been tested to another 10^-4 or better, so by the correspondence principle, it is not "wrong" by very much. Compared to electrodynamics, weak, and strong interactions, gravity is so weak that it is very difficult to probe using local measurements. Thus, it is tested using astronomical observations, but as I mentioned, particular cosmological models and other complications often interfere with the clarity you have of an experiment with a ball and a wall.
So, while your (and Hawking's) logic is obviously correct, it is an enormous simplification to imagine that, in practice, a single experiment could possibly unseat GR. Practically, what would happen is that a body of unexplained evidence begins to build up. With all the successful tests of the theory that have occurred, until quite a few failures occur, the experiments themselves are more suspect than the theory. This is not a failure of the ideal of the scientific method, but rather a reflection that experiments have error bars and experimentalists make mistakes.
Second, a disproven theory is not "bunk" -- it may be incomplete, but if it was a good theory to begin with, it has a wide domain of applicability all the same. Remember how you start by learning Newtonian mechanics, Newtonian gravity, classical electromagnetism, etc? Those theories have all been "disproven." They are incomplete. However, for vast, huge, enormous parts of observation, they are more than accurate enough. The correspondence principle reflects the fact that even when a theory is disproven, the parts of it that had been tested had damn well better be matched by whatever theory replaces (or, more accurately, extends) it.
An interesting historical example -- before Einstein published GR, anomalies that were not explained by Newton's theory of gravitation were known and had been known for more than 40 years. For example, it was known that the orbit of Mercury had an anomalous drift. No one immediately tossed out Newton's gravity -- in fact, for much of that time, it was believed that an undiscovered planet existed between Mercury and the sun! It turned out that GR explained almost exactly the perturbation and no extra planet was necessary, but until his new theory came along, there was no definite need to assume that Newton had made an error based solely on the experimental observations of Mercury's orbit.
Slashdot Mirror
I was at MIT for 5 years (bachelor's and master's) and, while I would agree that some of your criticisms have truth, by and large it's an excellent institution.
Some of the professors barely try to teach. But the majority I encountered (in the physics and electrical engineering departments) were good to excellent. In the physics department, especially, most of the recitation instructors are professors, not TAs. The TAs do grading, but professors lead the discussions and show up for office hours. If you bother to show up for those, most are very happy to help you with anything that's giving you trouble.
Class size? A few intro classes were large, but beyond that it's mostly small (25-30 range) and for larger classes, they break down into smaller recitations. As I mentioned, in physics, these are usually lead by professors. This is also often true in the EE department.
MIT students, even undergrads, do get to work on cutting-edge research. That is not a lie. However, it's not automatic. You have to have the ability to contribute to the project, and many undergrads need to learn through classes and/or experience to be able to contribute. Sometimes that learning involves grunge work. That's not a problem with the school, it is a fact about research. Every professor, every grad student, every researcher has gone through their share of the grunge work.
If you have the experience/ability coming in, you can find a professor who'll give you a job doing cutting edge stuff right away. If not, you can find one who will find a way for you to work in his lab while you learn what you need to know to make meaningful contributions. The opportunity is there in a way that it simply is not at some schools.
As for the Media Lab, I worked there for several years. If you really think the best that came out of there is the alarm clock, you should realize that headlines don't tell the whole story. A lot of good work was and is done there. A large number of start-up companies came out of the Lab. Some very interesting work on wearable and embedded computing was done there. A lot of work on spreading modern technology to the developing world was and is going on there. It's not all silly consumer products.
Finally, the high-school team (I thought they were from CA, not TX, but whatever) who won the underwater vehicle competition was awesome. They beat a lot of teams through ingenuity and hard work. MIT wasn't the only team these guys beat -- your example says a lot about the high-schoolers and very little about MIT. MIT has had enough success at various competitions (including their Orca underwater autonomous vehicle team -- also a submarine but run entirely by onboard software) that pointing to a single case where they didn't win doesn't say much.
MIT, like any school, has its problems and ultimately the success of an institution really depends more on the people who attend than on the institution itself. A bright person will do great things whatever school they go to. Thanks in part to reputation and in part to an "installed base" of bright people, MIT tends to attract a pretty large percentage of bright people.
Yep, it was not long after that that Covad went under. Speakeasy managed to hang on without them though, although I don't remember the details of that transition.
One other interesting thing is that while I was waiting the 3 months to get online, Speakeasy gave me free dial-up. Not great, but it was enough to get me by until I could get the high-speed line. It probably cost them very little (just using otherwise unused resources) and has helped them get a long-term customer.
Sorry -- no offense was intended, I'm sure there are many tech support people capable of giving better support than their organization will let them. I really meant to get at the fact that Speakeasy as an organization lets them do their jobs to the best of their ability.
There's some risk, but I think the techs are pretty wise about their suggestions -- I haven't encountered anything like swapping out parts, but I am pretty confident that if I did so on my own and told them the results, they would be able to use these in their diagnosis of a problem.
I've been with Speakeasy for almost 5 years and would not consider another provider. When I first signed up, it was a bit rocky getting online (took about 3 months while they coordinated with the local phone company and the Covad middle layer) but since then I have moved from coast to coast a couple times and had no trouble bringing my service with me. Other than at the very beginning, I've had virtually NO downtime -- ran into a little trouble when my DSL modem started failing, but they can hardly be held responsible for that.
Furthermore, they have eminently reasonable policies. You are allowed to use your DSL connection as a full and proper connection to the internet -- they have no arbitrary restrictions on services you can run. It's not a download-only pipe like the Telco and cable companies want to sell you. They do their best to support you running any OS you want and the techs I've spoken with are actually sharp enough to help you outside of a script. Not only that, but they have some authority to do what it takes to get the job done. All the while, you have access to the communications logs between the Speakeasy techs and the local telco and other parties involved in providing the line.
The existence of a company like this, IMO, indicates that there is demand for services the telcos are unwilling or unable to provide. They footed part of the bill to run the wires to your house, so they should get some return. That's why Speakeasy *rents* the line from them and adds their markup on top. There's no reason that the telco needs to bundle ISP services with the telco line. If I think the telco has a good pipe but offers crappy ISP service, it makes sense that I can opt out of their ISP offerings. The architecture is already in place to let me do this.
Furthermore, the telcos did not foot the entire bill for running the wires. Government assistance and tax dollars helped set up the network. They're part of the public infrastructure and they knew that when they got in the business. They therefore have responsibilities not only to their shareholders, but to the society that they bargained with to get their business in the firstplace.
Yeah -- New Egg has always been good when I have used them. The last time I was there, they were very clear about their review policy. The highly visible disclaimer basically said what people are saying here: "We are a commercial site selling products. We are not an independent review site. If you want independent review, look somewhere else." Even if their reviews are all more or less positive, they still often have useful information.
Uh, nice of you to correct his correct statement, but perhaps you would like to re-read his statement and ask yourself whether you understood it the first time through. :-)
In California, auto insurance is not specifically required. You can opt to post a bond that exceeds the minimum required insurance level, as you suggest.
However, the minimum insurance is pretty low. With the 100/300/100 insurance I buy, the maximum possible insurance payout will pretty much always exceed what I've put into it. So even if I had the funds for the minimum, it still probably makes sense to buy the insurance.
Insurance is not really a waste if the mean net cost isn't break even. It's protection against standard deviations. $100k+ accidents are not common, so the average expected cost is low -- however you need to be able to cover the whole cost in the rare case where your number gets called. Insurance helps to smooth the cost over a large number of drivers, and it takes work to manage this process.
I don't think it's fair to call it a ripoff based on the justification that profit is involved. Profit for a service is fair -- the insurance agents have to put food on their families too.
Are there really states where this is supposed to be a defense? In California they are very specific that the "flow of traffic" is not a legal defense for speeding. I think this is similar in Indiana and Massachusetts and I'd assumed it was the case for other states as well.
So the light turns green. (my light. the turning light. of a 10 lane highway +2 turning lanes)
This just makes me think of highway moderation. e.g., SR-137 (Score: +2, turning lanes)
Yep, I wasn't careful about the history, particularly since we were talking about ctrl- shortcuts to start with.
Woah there, QuantumG, I think that horse you're riding is a little high for you.
I think I asked some relevant questions about a statement you made and you respond by asking me to STFU. Now you're attacking me (and others) as someone who "[does]n't know shit talking out of [his] ass"?
That's great. You've convinced me that you're an expert on the topic. Well, either that or your HCI class prof mentioned that ctrl-c and ctrl-v are monstrosities and you're just parroting that on slashdot.
Seriously, I would like to know what it is about a keyboard shortcut that makes it an HCI monstrosity. I honestly don't see how you can defend that position, given that it is there as an accelerator for a process that can be done in other ways.
And don't try to pull academic rank here, pal, you don't know anything about who you're talking to. Furthermore, HCI is not a topic that can or should be preached from on high, it's just not that kind of a field.
Personally, I have worked very closely with a lot of researchers in this area and while they have a lot of neat ideas, ultimately the measure you're after is does it make interaction easier. What's the measure of that? You go and ask one of these "people who don't know shit" to use it in the real world and see how it turns out. Theories are great, but ultimately it comes down to a very practical result. The HCI researcher has his theory, but these "people who don't shit" are the only ones who actually "know" whether the HCI theory is right.
Also, I hope you understand that there is a place for _transparent_ usability and there is a place for _efficient_ usability. These are distinct goals that are important for different applications. If I'm going to use a kiosk to check in at the airport, I want the interface to tell me how to use it without having to learn anything, even if an experienced user could operate faster. However, if I'm in a factory controlling an industrial machine to assemble parts of a car, I'm willing to tolerate a much higher learning curve if the end result is I'm 20% more efficient thanks to a pared-down interface.
In the first scenario, the total time I spend on the task is short -- it would be silly to spend 10 minutes learning how to use a system to perform a task I can do in 5 minutes. In the second scenario, I'll be spending 8 hours a day every day performing this task. Spending 2 weeks training on the system is a good investment -- if I am 20% more efficient, I'll recover that 2 week training investment after 10 weeks of work and after that I'm permanently better off.
Control-C and Control-V fall into this latter category. I spend a LOT of time editing text on a computer, and I am MUCH faster as a result of these keystrokes. It is worth my time to learn the tool. It increases the _efficiency_, even if it's not _transparent_ (but note that it doesn't even reduce the transparency -- if you're new to the computer, there are more transparent alternatives that let you perform the same operation).
Anyway, I really am interested in why you might label ctrl-C and ctrl-V as monstrosities and/or why you disagree with what I've said above. Additionally, I think you owe me and the others an apology for your obnoxious response, but I won't hold my breath.
Grow up.
Why? Because it's not the analog of a real-world operation? Would you feel better if you had a little dedicated copy and paste button on your keyboard, or does using the mouse just make things automatically more usable?
You have to learn how to use a tool. A computer is a tool. Copy and paste are things you do when you use it. Nearly every program that supports copy and paste uses ctrl-c and ctrl-v and many keyboards even print "copy" and "paste" as hints. Ok, the Mac goes and uses the "Apple" key instead of ctrl, but it's the same idea.
Furthermore, they are very convenient buttons to press with your left hand while mousing with your right. Not perfect, utterly transparent design, but eminently pragmatic and *consistent*. That sounds usable to me.
It makes sense -- they're selling you bandwidth, and how you use that is up to you.
I've had a lot of problems interoperating with MS Word. Documents frequently come out with formatting that's not quite right -- text that was clearly supposed to fit on one page spills onto the next page (e.g., a signature line for a contract is on the next, otherwise empty, page).
.doc, so I had to change it to a manually-typed date.
As an experiment, I recently tried writing a technical document in a combination of Word:Mac and OO.o on linux. It was a nightmare. Figure and table caption references appeared to be almost completely broken -- they would not update if numbering changed. I had a date field at the top to automatically date stamp the document -- OO.o would repeatedly turn this into junk if the document was saved as a
Furthermore, the table editing features in OO.o were really complete crap. It was next to impossible to get it formatted as I wanted. Once I did, if I changed anything, it was a complicated process to get it to look nice.
I've long been a supporter of OO.o, and I probably would continue to use it for short documents. From a comparison with Word:Mac, though, its interface is really below par. In the end, I converted my document to LaTeX, and I expect that I'll use that for any non-trivial writing I do in the future.
This was done in one case in a system whose password check routine took a pointer to the password to check and then ran strcmp() or the equivalent against the password IN THE ORIGINAL LOCATION. This could be cracked by aligning the password to check against a page boundary. If the strcmp() ran past the page boundary, a large delay would be generated while the missing page was fetched.
This would work on a fast processor, but relied on being able to control what parts of memory were paged to disk, which is harder to do these days.
I use Firefox normally, but there are a few pages (the bungie.net halo forums, in particular) that I have to use Konqueror for. Firefox renders the funky animated crap they insist on adding to their forum excruciatingly slowly. Konqueror is snappy.
Primary Phase
By the way, there is a complete set available from amazon.com (the US version) but based on the reviews, the audio quality is not so great. I got the CDs from Amazon UK a few years ago (I am in the US) and they are great.
I don't know, I find jacket-and-tie-only occasions to be pretty annoying. Analogies between real life and the internet are always weak, but your analogy is a bit more apt. Still, I think the restaurant is within its rights to refuse service.
Unless you need the earplugs because of a legally recognized handicap -- then they'd have to accommodate you. So there should be an exemption for seeing-eye earplugs. Likewise, a web site operator should accommodate someone with a browser that blocks ads due to a handicap, such as a reader for the visually impaired.
Just this morning, I noticed that my previously favorite travel website now has a pop-up ad that circumvents the standard pop-up blocking in Firefox (not AdBlock or whatever the extension is). I sent them feedback letting them know that as long as they were not respecting the clear wish of their customer not to be bothered with a pop-up, I would no longer be using their services. I could (and may) install the AdBlock, but I won't return to their site until they fix the real problem which is disregard for their customers wishes. If they decide they'd rather have the ads than uptight customers like me, then I'll use other services.
Much as I can opt out of using their service if I am unwilling to put up with ads, the service provider should be able to opt out of providing me with a service if I am unwilling to accept their terms. It's no different than a a fancy restaurant with a dress code that requires a jacket and tie. If you want to wear a t-shirt and sandals while you eat, go somewhere else.
However, unless they notify me of these terms, there is and should be no rule that says I must view every bit of data they throw at my web browser. If they do notify me, then I should respect their wishes. I don't see it as a "social contract" so much as common sense and courtesy.
Um, the generosity you refer to was precisely the availability of the free license. It has everything to do with licenses, even if it doesn't use the word.
Don't discount the necessity of knowing buzzwords if you want to succeed in a field. If you want to contribute to a complex field, you MUST know the buzzwords (i.e., jargon, the specialized language of the field) and you MUST use them correctly. Otherwise, you're wasting your readers' time -- time they should be using to understand your carefully crafted argument, not trying to figure out what basic concept you are talking about.
You're right, that's not ALL there is to it -- obviously you also need to learn to make the arguments and generally write well. However, it's not unreasonable to get a poor grade in a sociology course because you turned in a well-crafted paper that didn't use sociological vocabulary.
I'm not attempting to bring questions of higher powers into the question.
What I am saying is that interactions between neurons in a brain are much much much more complex and complicated than the "inspired cartoon" that is a neural network. The connection diagram of an NN is similar to that of neurons in the brain, and the notion that one node responds to a signal from others is also similar. However, the mechanisms going on in the brain are much more complicated.
I don't disagree with most of your post, but I think you overstated the biological basis for neural networks.
The article states that the software is not used for all the papers -- final papers are read and graded by hand. I'm not sure whether that means several versions of each paper are assigned (draft, edit, then final) or whether final project type papers are the only ones graded by hand, but that seems reasonable to me -- it allows more papers to be assigned. Even if they're not graded as insightfully as by a human, the way you get better at writing is to write a lot.
As someone else mentioned, it's very helpful to get detailed feedback about your writing. However, you don't NEED that feedback at every step to get better. Just going through the process, thinking through the arguments, and writing them down forces you to learn. If you are writing something, even just for yourself, every week, you will naturally improve. Plus, you will gain the ability to do so quickly -- a necessary skill for academia.
Sure, if you can't write AT ALL, then you are going to need a lot of feedback to even get your papers to a coherent level -- you're not going to be able to use self-directed feedback at all. But if you can't write coherently, you really shouldn't be learning that skill in a sociology class. You should be learning the particulars of writing pertinent to that discipline, but mostly using the papers as a way to learn the sociology material.
Overall, I think this seems reasonable if used judiciously, and it sounds to me like the professor does so.
You overstate the similarity between a neural net framework and the actual operation of a brain. A neural net is better characterized as a cartoon vaguely inspired by the structure of an animal brain rather than "based on the fundamental theory of how animal brains (including human brains) operate."
You are correct that a true counterexample in some sense disproves a theory, but in practical terms, you're mistaken on two counts.
First, the implication of an experiment is rarely so black and white. Not all experiments are as concrete as the examples you find in textbooks. Many experiments in astrophysics, for example, rely on enormous extrapolation from the data using a complex model. Even when the data comes in, it is difficult to determine conclusively whether it contradicts the theory. Often, individual experiments only rule out small regions of a large parameter space. A combination of several experiments may be enough to rule out that model if they share no common allowed region of paramter space.
Gravity in particular is notoriously difficult to test experimentally. In our region of the universe, Newtonian gravity is correct to the 10^-8 level. GR picks up those billionth-part corrections. GR has been tested to another 10^-4 or better, so by the correspondence principle, it is not "wrong" by very much. Compared to electrodynamics, weak, and strong interactions, gravity is so weak that it is very difficult to probe using local measurements. Thus, it is tested using astronomical observations, but as I mentioned, particular cosmological models and other complications often interfere with the clarity you have of an experiment with a ball and a wall.
So, while your (and Hawking's) logic is obviously correct, it is an enormous simplification to imagine that, in practice, a single experiment could possibly unseat GR. Practically, what would happen is that a body of unexplained evidence begins to build up. With all the successful tests of the theory that have occurred, until quite a few failures occur, the experiments themselves are more suspect than the theory. This is not a failure of the ideal of the scientific method, but rather a reflection that experiments have error bars and experimentalists make mistakes.
Second, a disproven theory is not "bunk" -- it may be incomplete, but if it was a good theory to begin with, it has a wide domain of applicability all the same. Remember how you start by learning Newtonian mechanics, Newtonian gravity, classical electromagnetism, etc? Those theories have all been "disproven." They are incomplete. However, for vast, huge, enormous parts of observation, they are more than accurate enough. The correspondence principle reflects the fact that even when a theory is disproven, the parts of it that had been tested had damn well better be matched by whatever theory replaces (or, more accurately, extends) it.
An interesting historical example -- before Einstein published GR, anomalies that were not explained by Newton's theory of gravitation were known and had been known for more than 40 years. For example, it was known that the orbit of Mercury had an anomalous drift. No one immediately tossed out Newton's gravity -- in fact, for much of that time, it was believed that an undiscovered planet existed between Mercury and the sun! It turned out that GR explained almost exactly the perturbation and no extra planet was necessary, but until his new theory came along, there was no definite need to assume that Newton had made an error based solely on the experimental observations of Mercury's orbit.