I was anticipating such a reaction. I never suggested using tables for layout normally, although the end user certainly would not notice the difference, but given a severe time constraint, I suggested it as an alternative to fiddling around with CSS to get things working in IE and other browsers.
Also, both tables and framesets are tools. They do have legitimate uses, and to dogmatically reject them in all situations is foolish. It's like saying you'll never use a hex key because you don't want to accept that hex sockets exist.
"Best practices" is a misnomer anyway. It implies that the aforementioned software development practices are somehow optimal. If this is as good as software engineering can ever get, we're in trouble!
I'm not sure what you mean, but my post has nothing to do with the ACID test; it was a reply to the parent mentioning that certain pages broke in IE7 that worked in IE6. It was largely due to people exploiting weird IE6-specific CSS parsing bugs to create hacks, which were then fixed when IE7 rolled around, causing their pages to break.
As I recall, the reason many pages broke between the IE6 / IE7 transition was because people were hacking up their CSS and the parsing bugs underlying their hacks failed.
Ironically, your best bet in such a situation is probably to use tables. No, it isn't "proper", but they tend to render more consistently across browsers, likely because they're older and less complex than most of the CSS that replaces them. Also, certain things are unnecessarily complicated in CSS but trivial in tabular layouts, like vertically centering the content of block elements (without doing kludgey things like setting the line height to 100%, vertical-align won't work for this purpose... except on a table cell).
My own answer is to insist on projects that have deadlines sufficient for their scope rather than compromising on the quality or compatibility of the site, but that's a luxury only available to those who freelance, of course.
I think we need to define what the long-term goal of the space program is - not just "let's go to Mars!", which, cool as that would be, is not the sort of goal we should be striving for in and of itself. If our goal is to get off of earth in such a way that any random asteroid can't kill us all off, setting up a permanent base on the moon would seem to be a practical step in that direction.
In biological evolution, the difference between natural and sexual selection is a bit more clear-cut than it is in optimization. Things that are naturally fit survive; things that are sexually fit are selected first for reproduction based on the perceived ability of their offspring to survive. Like you said, barring weird things like Fisherian runaway, allowing sexual selection based on perceived fitness is a good selection strategy for improving fitness not only of each individual, but of the population as a whole.
However, how would you advertise the correctness of an optimization solution? Perhaps by the degree of convergence (MSE?), but this is part of any good fitness function already anyway. You don't need to advertise that you weren't eaten by a leopard; the fact that you're still here makes it redundant. Maybe selecting traits that are disadvantageous but survive anyway? (Handicap principle). Looking at the functional form is promising if you can recognize one that looks like it would converge well. Of course, we'd like to avoid getting stuck in local minima as well, and Fisherian runaway would probably take the form here of some sort of functional overfitting which we'd also like to avoid if possible (we can probably get rid of it, as in biological evolution, using the standard fitness function). One possibility is to first naturally select (standard GA) a solution, then to compare forms to the ones that most closely converge (in essence, to look for "healthy" solutions). Another might be to evolve the criteria with the GAs, which might give a better result.
Anyway, this is getting tangential. You shouldn't have gotten me started:)
I never said it wasn't, though perhaps the term "voluntary" didn't quite convey the meaning that I wanted. Blogging is not a method that most people use to make a living, whereas scientist is an actual occupation. If the scientists haven't unionized yet, I don't see why the bloggers consider it necessary to.
Blogging is a voluntary activity generally conducted solely by the individual doing the blogging. Whether to charge or not is an individual choice.
Also, scientists generally contribute far more intellectual energy to submitting their publications and they aren't paid for it either (although it is considered somewhat of a job expectation). As for protecting their IP, their articles generally cease being their own IP once a journal gets ahold of it, upon which it controls distribution and very often ransoms access to the public, making a profit for the journals - but not the scientists who wrote the paper. I think researchers may need to unionize earlier than bloggers if abuse of IP is what you're concerned about.
My guess is that GA in this context means "genetic algorithms". However, I can't see how such an experiment would be conducted using GAs, considering that they're merely a sort of optimization/machine learning algorithm (and machine learning is one of my research interests). No one expects GAs to exhibit human-level reasoning.
Therefore, I think the GP should cite these experiments so we can judge for ourselves.
No, I am not a recruiter. I'm not even in industry at the moment; I'm one of those poor sops chasing a Ph. D. and going into science:)
However, even though I'm not searching for jobs, I am a recruitee more often than not, and from companies including Google, AT&T, and Exxon-Mobil to boot. Since this article deals with entry-level jobs fresh out of college, I'm not yet that far out of college, and I've received offers that are far better than those my peers received in terms of frequency, employers, and compensation despite not even applying, I think I'm doing something right and wanted to share my strategy. (Granted, having an MS probably has something to do with it too, but that's why I think it's worth the extra year to get one).
Anyway, this isn't about me, and I think you misunderstood what I said: start gathering the experience before college (and do well in it) so that by the time you graduate, you have both a good GPA and experience. This will start you off with a considerable advantage - you're neither candidate A nor candidate B; you have the best of both. It worked for me, anyway.
As a programmer, the key is to start before college. By the time you're out, you have a degree and a good deal of experience. If you can land a job in the industry before college (I managed to), that looks even more impressive.
Also, as stupid as it is, having a high GPA will also greatly improve your job prospects. It's always the first thing that the recruiters' eyes seem to fly to on the resume, probably because they desire something quantitative and can't really use years of experience (or so they think).
Also, if you're going to grad. school for a Master's, you can obtain it in as little as one year (full time). It's the Ph. D. that takes a long time, and you wouldn't really want to go for that unless you planned on doing research at some point - it doesn't confer much of an advantage.
Ultimately, if you have the skills and make use of them when you're off the job as well as on, nothing can permanently hold you back. Sooner or later someone notices.
It's also not the sort of thing you normally plan multi-node trips around. If you have to pick someone up on the way to the airport, you can't visit the airport first then pick the person up - these sorts of things are just shortest-path problems, rather than TSPs.
Although this would be a nifty feature for the few times that the order doesn't matter.
How would censoring a potential Alzheimer's cure (and one that makes sense based on fairly well-established theory) on the basis of the methodology used to conduct the research improve the scenario?
If the research doesn't hold up, the community will figure it out. Now that the cat is essentially out of the bag, people are going to perform their own experiments to confirm or refute these claims. Even if this particular treatment doesn't work, the paper still contains the germs of ideas required to produce a more effective one.
So here are the possible "bad" scenarios:
False positive: The paper is accepted when the theory is really flawed. A few scientists spend perhaps a few months devising experiments to test the claims, and refute the paper based on further empirical evidence. If the theoretical aspects of the paper make sense, it may still be extended to produce a working cure. The only thing wasted is a small amount of effort, and even that is not entirely in vain.
False negative: A working Alzheimer's cure never sees the light of day.
It's clear which of these outcomes is better.
I've never quite understood the concept of scientific skepticism, at least as it's currently practiced in the field. It's ok to demand a large amount of empirical evidence before personally accepting a theory, but it is actually a very bad decision from the perspective of utility theory (or anything that utilizes expectation) to censor a plausible idea. Statistically, unless the probability that the theory is true is small enough to negate the consequences of the theory being true, it's a losing strategy, since it prevents others from even gaining exposure to the theory; a necessary precondition to devising experiments to gather additional evidence.
(Disclaimer: I do biomedical research on the human brain, but do not possess expert knowledge in the pathology of Alzheimer's disease; my field is closer to CAD. YMMV).
Re:The best tools stay out of the way...
on
Goodbye Cruel Word
·
· Score: 1
I have the whole thing under version control, of course:)
I've never written anything that long in Word before, so it's possible that I may run into this later. I plan on the whole thing being about 150-200 pages, so I could probably just split it up into two smaller documents, create PDFs from each at the end, and merge them together if that becomes an issue.
Re:The best tools stay out of the way...
on
Goodbye Cruel Word
·
· Score: 1
No need to get so defensive - this is a word processor you're arguing about, not a religion. It reminds me of the vi vs. emacs debate.
Any other LaTeX recommendations I've seen attached to this thread have centered around the ability to "simply write" in LaTeX, which is essentially what I'm doing in Word - yours is the only ad hominem I've seen thus far. I could do it in LaTeX as well (or in LyX if I really wanted a frontend), but since my department prefers reviewing Word documents and I'm equally comfortable with both systems, I've decided to use Word instead. Since I've written 60 pages conforming to my university's style guidelines and have not yet run into any serious formatting problems, I am going to conclude that, for my own purposes at least, this was not a poor decision and that, whatever I know, I know enough to keep my document under control. Because my workflow is not impaired, use of another system carries no additional performance benefit.
I will grant that I have not yet numbered my equations (I prefer to reiterate an equation from another section if I reference it in the future rather than directing the reader to jump across the document searching for it by number, as this makes my proofs easier to understand). However, this is the sort of thing that can be easily annotated after completion using a trivial macro should I require it.
Re:The best tools stay out of the way...
on
Goodbye Cruel Word
·
· Score: 1
The equations are converted to images upon saving to.doc format (and you can convert to PDF as well). Furthermore, if science journals reject my dissertation on nothing more than the basis of the editor I use to type them (and no one has objected to Word in any of my articles yet), they have fallen far from their original intent.
Re:The best tools stay out of the way...
on
Goodbye Cruel Word
·
· Score: 1
I almost used LyX, but my advisor prefers to receive Word drafts. LyX is a great frontend, though.
Re:The best tools stay out of the way...
on
Goodbye Cruel Word
·
· Score: 1
By far my worst complaint about the ribbon is that it makes keyboard navigation exceptionally difficult, since there is no clear indication of what shortcuts do what. Some things, like "insert equation", show the shortcuts in the tooltip (it's alt+=), but most others, such as "insert symbol" right next to it, are completely absent from view. Maybe these shortcuts exist. Maybe they don't. There's no way of easily finding out. Menus would show these to you right next to the menu itself.
Another issue is seemingly arbitrary placement of commands on the ribbon. Of course, this is an effect of cramming the full functionality of the application into around 8 categories. For example, "Insert Table of Figures" is under the References tab, when it should really be under the "Insert" tab, while "Cross-reference" appears in the Insert AND References tab, when it should probably only appear in the latter. "Macros" is in the View tab; it should probably appear only in the Developer tab (which isn't shown by default). Some commands, such as "insert horizontal line" and "insert column break", appear to be absent from the ribbon altogether.
Re:The best tools stay out of the way...
on
Goodbye Cruel Word
·
· Score: 1, Informative
I'm writing my dissertation (60 pages done so far) in Word 2007. The new equation editor makes it far better at this than Word 2003 and it accepts most LaTeX syntax as well. I'm actually finding it easier than LaTeX because of this - I type my type, I type my equations, and Word takes care of most of the other drudgery for me. I don't have to deal with issues of markup, as in LaTeX. Now if only they'd add it to PowerPoint too.
(In light of this, I find it odd that I still prefer to hand-code HTML, but that's probably because each page has different elements. It's not just a mountain of text with a few images and tables thrown in).
There really is no problem with Intel selling a competing product. No one is forcing the kids to buy a Classmate. If they can compete with a free OLPC and still sell their laptops, those laptops are probably superior in some way.
Expecting Intel to simply withdraw their laptops to avoid competing with OLPC is like something out of "Atlas Shrugged" (and I'm not even an objectivist). If OLPC is really as good as it's cracked up to be, they have no reason to block competition. Moreover, Intel was helping them. If they want to make demands of Intel that Intel isn't ready to meet in exchange for these benefits, Intel can simply pack up and leave and has every moral right to do so.
Slashdot Mirror
I was anticipating such a reaction. I never suggested using tables for layout normally, although the end user certainly would not notice the difference, but given a severe time constraint, I suggested it as an alternative to fiddling around with CSS to get things working in IE and other browsers.
Also, both tables and framesets are tools. They do have legitimate uses, and to dogmatically reject them in all situations is foolish. It's like saying you'll never use a hex key because you don't want to accept that hex sockets exist.
"Best practices" is a misnomer anyway. It implies that the aforementioned software development practices are somehow optimal. If this is as good as software engineering can ever get, we're in trouble!
I'm not sure what you mean, but my post has nothing to do with the ACID test; it was a reply to the parent mentioning that certain pages broke in IE7 that worked in IE6. It was largely due to people exploiting weird IE6-specific CSS parsing bugs to create hacks, which were then fixed when IE7 rolled around, causing their pages to break.
As I recall, the reason many pages broke between the IE6 / IE7 transition was because people were hacking up their CSS and the parsing bugs underlying their hacks failed.
Ironically, your best bet in such a situation is probably to use tables. No, it isn't "proper", but they tend to render more consistently across browsers, likely because they're older and less complex than most of the CSS that replaces them. Also, certain things are unnecessarily complicated in CSS but trivial in tabular layouts, like vertically centering the content of block elements (without doing kludgey things like setting the line height to 100%, vertical-align won't work for this purpose... except on a table cell).
My own answer is to insist on projects that have deadlines sufficient for their scope rather than compromising on the quality or compatibility of the site, but that's a luxury only available to those who freelance, of course.
I think we need to define what the long-term goal of the space program is - not just "let's go to Mars!", which, cool as that would be, is not the sort of goal we should be striving for in and of itself. If our goal is to get off of earth in such a way that any random asteroid can't kill us all off, setting up a permanent base on the moon would seem to be a practical step in that direction.
In biological evolution, the difference between natural and sexual selection is a bit more clear-cut than it is in optimization. Things that are naturally fit survive; things that are sexually fit are selected first for reproduction based on the perceived ability of their offspring to survive. Like you said, barring weird things like Fisherian runaway, allowing sexual selection based on perceived fitness is a good selection strategy for improving fitness not only of each individual, but of the population as a whole.
However, how would you advertise the correctness of an optimization solution? Perhaps by the degree of convergence (MSE?), but this is part of any good fitness function already anyway. You don't need to advertise that you weren't eaten by a leopard; the fact that you're still here makes it redundant. Maybe selecting traits that are disadvantageous but survive anyway? (Handicap principle). Looking at the functional form is promising if you can recognize one that looks like it would converge well. Of course, we'd like to avoid getting stuck in local minima as well, and Fisherian runaway would probably take the form here of some sort of functional overfitting which we'd also like to avoid if possible (we can probably get rid of it, as in biological evolution, using the standard fitness function). One possibility is to first naturally select (standard GA) a solution, then to compare forms to the ones that most closely converge (in essence, to look for "healthy" solutions). Another might be to evolve the criteria with the GAs, which might give a better result.
Anyway, this is getting tangential. You shouldn't have gotten me started :)
I never said it wasn't, though perhaps the term "voluntary" didn't quite convey the meaning that I wanted. Blogging is not a method that most people use to make a living, whereas scientist is an actual occupation. If the scientists haven't unionized yet, I don't see why the bloggers consider it necessary to.
Blogging is a voluntary activity generally conducted solely by the individual doing the blogging. Whether to charge or not is an individual choice.
Also, scientists generally contribute far more intellectual energy to submitting their publications and they aren't paid for it either (although it is considered somewhat of a job expectation). As for protecting their IP, their articles generally cease being their own IP once a journal gets ahold of it, upon which it controls distribution and very often ransoms access to the public, making a profit for the journals - but not the scientists who wrote the paper. I think researchers may need to unionize earlier than bloggers if abuse of IP is what you're concerned about.
It could have rolled over :)
My guess is that GA in this context means "genetic algorithms". However, I can't see how such an experiment would be conducted using GAs, considering that they're merely a sort of optimization/machine learning algorithm (and machine learning is one of my research interests). No one expects GAs to exhibit human-level reasoning.
Therefore, I think the GP should cite these experiments so we can judge for ourselves.
In our effort to Leave No Child Behind, we've Left The Entire Country Behind. There's a sort of irony in that.
Although as a researcher, I have to say that the very structure of academia interferes with the ability to perform groundbreaking research as well.
No, I am not a recruiter. I'm not even in industry at the moment; I'm one of those poor sops chasing a Ph. D. and going into science :)
However, even though I'm not searching for jobs, I am a recruitee more often than not, and from companies including Google, AT&T, and Exxon-Mobil to boot. Since this article deals with entry-level jobs fresh out of college, I'm not yet that far out of college, and I've received offers that are far better than those my peers received in terms of frequency, employers, and compensation despite not even applying, I think I'm doing something right and wanted to share my strategy. (Granted, having an MS probably has something to do with it too, but that's why I think it's worth the extra year to get one).
Anyway, this isn't about me, and I think you misunderstood what I said: start gathering the experience before college (and do well in it) so that by the time you graduate, you have both a good GPA and experience. This will start you off with a considerable advantage - you're neither candidate A nor candidate B; you have the best of both. It worked for me, anyway.
As a programmer, the key is to start before college. By the time you're out, you have a degree and a good deal of experience. If you can land a job in the industry before college (I managed to), that looks even more impressive.
Also, as stupid as it is, having a high GPA will also greatly improve your job prospects. It's always the first thing that the recruiters' eyes seem to fly to on the resume, probably because they desire something quantitative and can't really use years of experience (or so they think).
Also, if you're going to grad. school for a Master's, you can obtain it in as little as one year (full time). It's the Ph. D. that takes a long time, and you wouldn't really want to go for that unless you planned on doing research at some point - it doesn't confer much of an advantage.
Ultimately, if you have the skills and make use of them when you're off the job as well as on, nothing can permanently hold you back. Sooner or later someone notices.
Why are "substantial mental improvements" worthless? Anything that improves the condition of the patients is probably worthwhile.
Isn't the dominant theory that that's what the moon is? A chunk of the Earth that ended up in orbit when something big hit it?
It's also not the sort of thing you normally plan multi-node trips around. If you have to pick someone up on the way to the airport, you can't visit the airport first then pick the person up - these sorts of things are just shortest-path problems, rather than TSPs.
Although this would be a nifty feature for the few times that the order doesn't matter.
How would censoring a potential Alzheimer's cure (and one that makes sense based on fairly well-established theory) on the basis of the methodology used to conduct the research improve the scenario?
If the research doesn't hold up, the community will figure it out. Now that the cat is essentially out of the bag, people are going to perform their own experiments to confirm or refute these claims. Even if this particular treatment doesn't work, the paper still contains the germs of ideas required to produce a more effective one.
So here are the possible "bad" scenarios:
It's clear which of these outcomes is better.
I've never quite understood the concept of scientific skepticism, at least as it's currently practiced in the field. It's ok to demand a large amount of empirical evidence before personally accepting a theory, but it is actually a very bad decision from the perspective of utility theory (or anything that utilizes expectation) to censor a plausible idea. Statistically, unless the probability that the theory is true is small enough to negate the consequences of the theory being true, it's a losing strategy, since it prevents others from even gaining exposure to the theory; a necessary precondition to devising experiments to gather additional evidence.
(Disclaimer: I do biomedical research on the human brain, but do not possess expert knowledge in the pathology of Alzheimer's disease; my field is closer to CAD. YMMV).
I have the whole thing under version control, of course :)
I've never written anything that long in Word before, so it's possible that I may run into this later. I plan on the whole thing being about 150-200 pages, so I could probably just split it up into two smaller documents, create PDFs from each at the end, and merge them together if that becomes an issue.
No need to get so defensive - this is a word processor you're arguing about, not a religion. It reminds me of the vi vs. emacs debate.
Any other LaTeX recommendations I've seen attached to this thread have centered around the ability to "simply write" in LaTeX, which is essentially what I'm doing in Word - yours is the only ad hominem I've seen thus far. I could do it in LaTeX as well (or in LyX if I really wanted a frontend), but since my department prefers reviewing Word documents and I'm equally comfortable with both systems, I've decided to use Word instead. Since I've written 60 pages conforming to my university's style guidelines and have not yet run into any serious formatting problems, I am going to conclude that, for my own purposes at least, this was not a poor decision and that, whatever I know, I know enough to keep my document under control. Because my workflow is not impaired, use of another system carries no additional performance benefit.
I will grant that I have not yet numbered my equations (I prefer to reiterate an equation from another section if I reference it in the future rather than directing the reader to jump across the document searching for it by number, as this makes my proofs easier to understand). However, this is the sort of thing that can be easily annotated after completion using a trivial macro should I require it.
The equations are converted to images upon saving to .doc format (and you can convert to PDF as well). Furthermore, if science journals reject my dissertation on nothing more than the basis of the editor I use to type them (and no one has objected to Word in any of my articles yet), they have fallen far from their original intent.
I almost used LyX, but my advisor prefers to receive Word drafts. LyX is a great frontend, though.
By far my worst complaint about the ribbon is that it makes keyboard navigation exceptionally difficult, since there is no clear indication of what shortcuts do what. Some things, like "insert equation", show the shortcuts in the tooltip (it's alt+=), but most others, such as "insert symbol" right next to it, are completely absent from view. Maybe these shortcuts exist. Maybe they don't. There's no way of easily finding out. Menus would show these to you right next to the menu itself.
Another issue is seemingly arbitrary placement of commands on the ribbon. Of course, this is an effect of cramming the full functionality of the application into around 8 categories. For example, "Insert Table of Figures" is under the References tab, when it should really be under the "Insert" tab, while "Cross-reference" appears in the Insert AND References tab, when it should probably only appear in the latter. "Macros" is in the View tab; it should probably appear only in the Developer tab (which isn't shown by default). Some commands, such as "insert horizontal line" and "insert column break", appear to be absent from the ribbon altogether.
I'm writing my dissertation (60 pages done so far) in Word 2007. The new equation editor makes it far better at this than Word 2003 and it accepts most LaTeX syntax as well. I'm actually finding it easier than LaTeX because of this - I type my type, I type my equations, and Word takes care of most of the other drudgery for me. I don't have to deal with issues of markup, as in LaTeX. Now if only they'd add it to PowerPoint too.
(In light of this, I find it odd that I still prefer to hand-code HTML, but that's probably because each page has different elements. It's not just a mountain of text with a few images and tables thrown in).
There really is no problem with Intel selling a competing product. No one is forcing the kids to buy a Classmate. If they can compete with a free OLPC and still sell their laptops, those laptops are probably superior in some way.
Expecting Intel to simply withdraw their laptops to avoid competing with OLPC is like something out of "Atlas Shrugged" (and I'm not even an objectivist). If OLPC is really as good as it's cracked up to be, they have no reason to block competition. Moreover, Intel was helping them. If they want to make demands of Intel that Intel isn't ready to meet in exchange for these benefits, Intel can simply pack up and leave and has every moral right to do so.