Last year we had 2-cm accuracy. Now it's been improved to 1-in accuracy. Don't forget to thank Big Brother for raising the chocolate ration to twenty grammes a week !
To be fair, Slashdot's summary is not worse than the paper's summary.
There's a long list of issues with their methodology, and they make a fair assessment of these in the "Threats" part, which BTW should be discussed in the article, and not in the appendices.
As a whole, this paper reeks "We wanted to show how / how much women were discriminated against in Open Source. Our findings showed the opposite, so we kept making up criteria until one would exhibit (barely) the bias we wanted to denounce."
Of course when you're doing that, you're just begging to fall for this.
Non-exhaustive list of other issues I noticed: - Weighing issues: for example, how many commits from outsiders vs insiders. Given that, overall, women get better acceptance, I can conclude than insiders commit more than outsiders (in their dataset) - Missing stats (for example, we get gendered stats on whether a pull request is linked to an issue, but no insider / outsider distinction) - Plain old lies in the summary ("when a woman’s gender is identifiable, they are rejected more often" vs "Women have lower acceptance rates as outsiders when they are identifiable as women.") - Failure to mention that the error bars are for the strict dataset. I suppose this is standard practice, but the dataset error bars are probably swamped by the non-representativity of the dataset in the first place, and the methodology shortcomings, which means that they're misleading (nobody cares about their dataset). They don't make any effort to evaluate these errors (obviously that would be the hard part), and leave us with some hand-waving like "we are somewhat confident that robots are not substantially influencing the results". - Graphs that start at 60% to exaggerate differences (without using broken axis) - Using "theory" for "hypothesis"
They’re sending people that have lots of problems, and they’re bringing those problems with us. They’re bringing drugs. They’re bringing crime. They’re rapists, and some, I assume, are good people.
An accurate representation of what he said was "some illegal immigrants are criminals", which is true simply by the law of averages, with the implication that if they went through a vetting procedure we could perhaps filter out the criminal portion.
That's plain wrong and apologetic. A more accurate representation of what he said would be "some illegal immigrants might not be criminals". He only "assumes" that some are not criminals, but states as a fact that "by default" they ARE criminal (clearer with more context, like "they’re not sending their best"). It's a run-of-the-mill weasel word technique, which won't impress any seasoned journalist.
Yet I would agree with you that the WaPo title is awful, as is he's clearly talking about Mexican immigrants, not Mexican people in general.
And the whole idea of a country "sending" its problematic people across the border is stupid anyway. Poorest people that have nothing to lose leave the country in the hope of a better life across the border, of course if they were the brightest they wouldn't have a reason to leave (especially illegally), but that doesn't mean they're criminal by default.
I think that the English spelling has failed the English language from even before keyboards were invented, with the substitution of letters like thorn (not printed by Slashdot, though it is part of ISO-8859-1) with th that already had other pronunciations associated to it.
Look at orthographic depth. French is just as hard as English when you want to spell a word for which you have the pronunciation, however, when reading a word in French, apart from a few exceptions, you can reconstruct the pronunciation from the rules, where English is quite irregular (so, one-to-many vs many-to-many mappings).
Ironically, the Ars Technica article spells cafe instead of café.
In lower case, only ae (æ) and oe ligatures are missing from the French keyboard. Moreover, they're a PITA to type on Windows (alt code only) while there are sequences on OS X and Linux. Due to Bull SAS French engineers' stupidity (link in French), (oe) ligature is not in ISO-8859-1, which means Slashdot won't print it. Danes ensured that æ was in ISO-8859-1 (but it is much less used than oe in French). These ligatures are not typographical stylistic conventions, (oe) ligature is pronounced as é (like in, erm, café) when followed by a consonant where oe is pronounced o-e. For example, with words that are spelled and pronounced mostly the same in French and English: coexist, coelacanth. English lost the ligature, while it is present in French on the second word, which helps with the pronunciation.
In upper case, there are dead keys for ^, " and `, so Ô, Ï and È are not problematic. However, É and Ç again are a PITA on Windows. On Unix and OS X, one can use caps lock and the lower case letter. Upper case ÿ is also missing from ISO-8859-1, as it technically does not exist (no word begins with it , because " indicates that the letter is not pronounced together with the preceding letter), but ÿ is only used by old proper nouns, which sometimes have to be printed in all upper case.
To start, I'm not doing interviews, so my answer is more about how I would like the interviews to be, so keep that in mind:)
Asking how they would solve a problem (as long as it's relevant to your business) or work with a client, real-world puzzles relevant to your domain are good practices.
However I notice you let developers choose any language they want to solve some programming tasks. It seems too generic.
You're hiring someone for a business in a specific domain, so it's good to ask generic questions about the domain, but you're also hiring someone at a more-or-less specific position (rather less in smaller firms), so you have to ask questions relevant to the actual position. If every developer you hire decides to use another language for the new tools he develops, you will end up with a mess.
At least, give him a list of the languages you're actually using in-house to choose from. You're a small firm, so probably he will have to be fluent in most of the languages you're actually using, so give exercises in the most important ones, not only solving problems, but also "spot-the-bug", "tell us what this piece of code is doing", etc.
Languages usually have big libraries coming with them, don't expect the developer to know all of them by heart, unless you're explicitly hiring an expert in some language with some framework, so answer if he's asking about some library calls, and accept sensible made-up library calls.
Your first paragraph is completely wrong. You've been moderated accordingly and Dog-Cow told you as much, but I will provide an explanation.
The application is mostly not responsible for cleaning up up after itself, except for some shared resources (for example, POSIX shared memory). In your "test", you don't actually get a zombie process. Ironically, a zombie process would mean that resources have been cleaned.
A zombie process is a process that has/was terminated, and only the in-kernel process structure containing some statuses is kept until the parent process acknowledges that the child process has terminated. Other resources (memory, file handles, etc.) have already been reclaimed.
In your test nothing specific is likely to happen.The program will continue to run and will get a write error if he's trying to write to the opened file. It can be killed normally. What you probably were thinking about, is about what happens if you're actively writing to the file when you're removing the drive. In that case, the program will end up in uninterruptible sleep state (shown as the 'D' state in Linux's ps). That's nothing like zombie (that would be shown as <defunct>). In D state, the program has made a (synchronous) system call, and is sleeping while the system is performing the request. As the disk has been removed, the write request never returns. In that state, the OS will refuse to kill the process, because doing so would risk to crash the system if the syscall ever returns to a now non-existent process.
Your second paragraph is correct.
I suspect parts of your third paragraph are incorrect (specifically, the "need" to map 1:1), but I'm not an expert in the domain. AFAIK, there are framebuffers that are 1:1, and then you manipulate graphics objects through handles that are actually managed by the graphics driver (and don't need to be mapped).
It's you that don't understand how probabilities work.
You play heads or tails, and get heads 8 first times.
Gambler fallacy says that as heads and tails have overall the same probability, tails should happen next. Pulzar guesses that the coin is most probably a biased coin, and heads should happen next.
I was surprised but that one hods true: In 2008, 1.2M police officers vs. 2.4M incarcerated people. To put in perspective, USA represents 4.4% of the world's population and 24.7% of the world's incarcerated population.
Now you closed it a second time ! Let's match opening and closing tags !
(cue heavy Russian accent)
Voilà ! Much better now.
A physicist, a biologist and a mathematician are sitting in a street café watching people entering and leaving the house on the other side of the street. First they see two people entering the house. Time passes. After a while they notice three people leaving the house. The physicist says, "The measurement wasn't accurate." The biologist says, "They must have reproduced." The mathematician says, "If one more person enters the house then it will be empty.
They can just grind the remaining prescriptions for homeopathic remedies into dust, and present a grain of that dust to the pharmacist, who then gives them a glass of water.
But that would put them at risk of an overdose, as more diluted substances have higher potency !
And when Chinese tech companies are forming partnerships with American firms that have ties to the American military, critics complain that they will get crucial technology and sneak in backdoors, which could inadvertently improve the fundamental technology capabilities of the Chinese military — or worse, harm United States national security.
You guys didn't post one article about the new Star Wars trailer yesterday, and you are barely late to the game on this [snip] story (It's been weeks in the making).
I have a feeling of déjà-vu. Did you time travel from 1999 ? I can't wait for the dupe later this week. By the way, today is 'Back to The Future' Day on this side of the Atlantic.
Hint: If you're going to argue that fracking is safe or even actually good for health, you can pick plenty of arguments from yesterday's discussion.
I mean, if the arguments dismissing your favorite source of cheap energy as being the cause of 20x-50x higher rates in some health issue are working, they certainly should work even better for 30%-40% increases, right ?
If you don't use full-screen-as-a-separate-space, it only requires a little tweaking of default parameters to get something usable, even if not to your liking. (By the way, is there some setting so that the default action is zoom and the alt/option one is full screen ?)
In settings -> mission control, deactivate automatically rearrange spaces, which is probably why you think that some applications appear on multiple spaces: actually, it's only on a single space (by default), but the spaces keep being rearranged.
If you want a space to span all monitors, I guess you can do it here too, but I'm a fan of distinct spaces on distinct screens, a feature I had 15 years ago on X11 (but which implied the inability to move windows between screens). Note that GNOME 3 does (did?) something "interesting" by default, which is a single desktop on the secondary screen, and virtual desktops on the primary one. Probably useful on laptops in presentation mode. (Is the default over-ridable) ?
In settings -> keyboard -> shortcuts, activate Mission control per-space bindings, so that you can swap between spaces with ctrl-(number). The catch is that you have to organize your spaces by task. You can then bind applications to some spaces. You can do that by alt-clicking in the dock and looking in options. Older versions of Mac OS X with Spaces (instead of Mission Control) had a list of assignations.
The space selection widget is gone though, if you won't use shortcuts, it's F3. Each space is rendered as a thumbnail on the top, you could choose a different background for each space if it helps you. The bottom part is all of the windows in the currently selected space. What is not very obvious is finding the currently selected space, depending on your color scheme: the white border is not very visible on white background.
You forgot to profess your love for Visual Studio, and how you used to program with vi or emacs. Because apart telling us than Windows 10 now has virtual desktops and how bad a job you're doing at organizing them, I don't know what was your point. If you can't keep your stuff organized, try invoking $MOTHER, because your issues are not technical in nature.
Slashdot Mirror
Last year we had 2-cm accuracy. Now it's been improved to 1-in accuracy.
Don't forget to thank Big Brother for raising the chocolate ration to twenty grammes a week !
To be fair, Slashdot's summary is not worse than the paper's summary.
There's a long list of issues with their methodology, and they make a fair assessment of these in the "Threats" part, which BTW should be discussed in the article, and not in the appendices.
As a whole, this paper reeks "We wanted to show how / how much women were discriminated against in Open Source. Our findings showed the opposite, so we kept making up criteria until one would exhibit (barely) the bias we wanted to denounce."
Of course when you're doing that, you're just begging to fall for this.
Non-exhaustive list of other issues I noticed:
- Weighing issues: for example, how many commits from outsiders vs insiders. Given that, overall, women get better acceptance, I can conclude than insiders commit more than outsiders (in their dataset)
- Missing stats (for example, we get gendered stats on whether a pull request is linked to an issue, but no insider / outsider distinction)
- Plain old lies in the summary ("when a woman’s gender is identifiable, they are rejected more often" vs "Women have lower acceptance rates as outsiders when they are identifiable as women.")
- Failure to mention that the error bars are for the strict dataset. I suppose this is standard practice, but the dataset error bars are probably swamped by the non-representativity of the dataset in the first place, and the methodology shortcomings, which means that they're misleading (nobody cares about their dataset). They don't make any effort to evaluate these errors (obviously that would be the hard part), and leave us with some hand-waving like "we are somewhat confident that robots are not substantially influencing the results".
- Graphs that start at 60% to exaggerate differences (without using broken axis)
- Using "theory" for "hypothesis"
Well, you see the problem: it doesn't fit in the subject line.
Next time you reference Him, I suggest you use the proper enneagrammaton for The Man Known As The Real Life Tony Stark: TMKATRLTS.
I wonder who could be that anonymous reader ?
The 100% oxygen environment was the perfect fuel
Yeah, sure. I wonder what could have been the oxidant ?
startswithabang doesn't submit any more ?
What he actually said:
They’re sending people that have lots of problems, and they’re bringing those problems with us. They’re bringing drugs. They’re bringing crime. They’re rapists, and some, I assume, are good people.
An accurate representation of what he said was "some illegal immigrants are criminals", which is true simply by the law of averages, with the implication that if they went through a vetting procedure we could perhaps filter out the criminal portion.
That's plain wrong and apologetic. A more accurate representation of what he said would be "some illegal immigrants might not be criminals". He only "assumes" that some are not criminals, but states as a fact that "by default" they ARE criminal (clearer with more context, like "they’re not sending their best"). It's a run-of-the-mill weasel word technique, which won't impress any seasoned journalist.
Yet I would agree with you that the WaPo title is awful, as is he's clearly talking about Mexican immigrants, not Mexican people in general.
And the whole idea of a country "sending" its problematic people across the border is stupid anyway. Poorest people that have nothing to lose leave the country in the hope of a better life across the border, of course if they were the brightest they wouldn't have a reason to leave (especially illegally), but that doesn't mean they're criminal by default.
I think that the English spelling has failed the English language from even before keyboards were invented, with the substitution of letters like thorn (not printed by Slashdot, though it is part of ISO-8859-1) with th that already had other pronunciations associated to it.
Look at orthographic depth. French is just as hard as English when you want to spell a word for which you have the pronunciation, however, when reading a word in French, apart from a few exceptions, you can reconstruct the pronunciation from the rules, where English is quite irregular (so, one-to-many vs many-to-many mappings).
Ironically, the Ars Technica article spells cafe instead of café.
In lower case, only ae (æ) and oe ligatures are missing from the French keyboard. Moreover, they're a PITA to type on Windows (alt code only) while there are sequences on OS X and Linux. Due to Bull SAS French engineers' stupidity (link in French), (oe) ligature is not in ISO-8859-1, which means Slashdot won't print it. Danes ensured that æ was in ISO-8859-1 (but it is much less used than oe in French). These ligatures are not typographical stylistic conventions, (oe) ligature is pronounced as é (like in, erm, café) when followed by a consonant where oe is pronounced o-e. For example, with words that are spelled and pronounced mostly the same in French and English: coexist, coelacanth. English lost the ligature, while it is present in French on the second word, which helps with the pronunciation.
In upper case, there are dead keys for ^, " and `, so Ô, Ï and È are not problematic. However, É and Ç again are a PITA on Windows. On Unix and OS X, one can use caps lock and the lower case letter. Upper case ÿ is also missing from ISO-8859-1, as it technically does not exist (no word begins with it , because " indicates that the letter is not pronounced together with the preceding letter), but ÿ is only used by old proper nouns, which sometimes have to be printed in all upper case.
To start, I'm not doing interviews, so my answer is more about how I would like the interviews to be, so keep that in mind :)
Asking how they would solve a problem (as long as it's relevant to your business) or work with a client, real-world puzzles relevant to your domain are good practices.
However I notice you let developers choose any language they want to solve some programming tasks. It seems too generic.
You're hiring someone for a business in a specific domain, so it's good to ask generic questions about the domain, but you're also hiring someone at a more-or-less specific position (rather less in smaller firms), so you have to ask questions relevant to the actual position. If every developer you hire decides to use another language for the new tools he develops, you will end up with a mess.
At least, give him a list of the languages you're actually using in-house to choose from. You're a small firm, so probably he will have to be fluent in most of the languages you're actually using, so give exercises in the most important ones, not only solving problems, but also "spot-the-bug", "tell us what this piece of code is doing", etc.
Languages usually have big libraries coming with them, don't expect the developer to know all of them by heart, unless you're explicitly hiring an expert in some language with some framework, so answer if he's asking about some library calls, and accept sensible made-up library calls.
Your first paragraph is completely wrong. You've been moderated accordingly and Dog-Cow told you as much, but I will provide an explanation.
The application is mostly not responsible for cleaning up up after itself, except for some shared resources (for example, POSIX shared memory).
In your "test", you don't actually get a zombie process. Ironically, a zombie process would mean that resources have been cleaned.
A zombie process is a process that has/was terminated, and only the in-kernel process structure containing some statuses is kept until the parent process acknowledges that the child process has terminated. Other resources (memory, file handles, etc.) have already been reclaimed.
In your test nothing specific is likely to happen.The program will continue to run and will get a write error if he's trying to write to the opened file. It can be killed normally. What you probably were thinking about, is about what happens if you're actively writing to the file when you're removing the drive. In that case, the program will end up in uninterruptible sleep state (shown as the 'D' state in Linux's ps). That's nothing like zombie (that would be shown as <defunct>).
In D state, the program has made a (synchronous) system call, and is sleeping while the system is performing the request. As the disk has been removed, the write request never returns. In that state, the OS will refuse to kill the process, because doing so would risk to crash the system if the syscall ever returns to a now non-existent process.
Your second paragraph is correct.
I suspect parts of your third paragraph are incorrect (specifically, the "need" to map 1:1), but I'm not an expert in the domain. AFAIK, there are framebuffers that are 1:1, and then you manipulate graphics objects through handles that are actually managed by the graphics driver (and don't need to be mapped).
It's you that don't understand how probabilities work.
You play heads or tails, and get heads 8 first times.
Gambler fallacy says that as heads and tails have overall the same probability, tails should happen next.
Pulzar guesses that the coin is most probably a biased coin, and heads should happen next.
having been shown at film festivals for the past year or so,
Do you intend to complete the summary later ?
Why We Stopped Going To the Moon
NASA and White House shifting priorities. More demand for a space station (Skylab).
I'll skip the fallacies and go straight to the factual errors.
We don't need 250 million guns.
Excessive hyperbole.
310 million civil firearms in the USA in 2009
civilians who commit literally hundreds of mass shootings every year.
Excessive hyperbole.
355 mass shootings this year so far.
On the other hand
There are more criminals than cops
I was surprised but that one hods true: In 2008, 1.2M police officers vs. 2.4M incarcerated people. To put in perspective, USA represents 4.4% of the world's population and 24.7% of the world's incarcerated population.
Now you closed it a second time ! Let's match opening and closing tags !
(cue heavy Russian accent)
Voilà ! Much better now.
A physicist, a biologist and a mathematician are sitting in a street café watching people entering and leaving the house on the other side of the street. First they see two people entering the house. Time passes. After a while they notice three people leaving the house. The physicist says, "The measurement wasn't accurate." The biologist says, "They must have reproduced." The mathematician says, "If one more person enters the house then it will be empty.
Wizard-hunt ?
I put on my robe and wizard hat.
They can just grind the remaining prescriptions for homeopathic remedies into dust, and present a grain of that dust to the pharmacist, who then gives them a glass of water.
But that would put them at risk of an overdose, as more diluted substances have higher potency !
And when Chinese tech companies are forming partnerships with American firms that have ties to the American military, critics complain that they will get crucial technology and sneak in backdoors, which could inadvertently improve the fundamental technology capabilities of the Chinese military — or worse, harm United States national security.
You guys didn't post one article about the new Star Wars trailer yesterday, and you are barely late to the game on this [snip] story (It's been weeks in the making).
I have a feeling of déjà-vu. Did you time travel from 1999 ?
I can't wait for the dupe later this week.
By the way, today is 'Back to The Future' Day on this side of the Atlantic.
Hint: If you're going to argue that fracking is safe or even actually good for health, you can pick plenty of arguments from yesterday's discussion.
I mean, if the arguments dismissing your favorite source of cheap energy as being the cause of 20x-50x higher rates in some health issue are working, they certainly should work even better for 30%-40% increases, right ?
When will they fix the bug that's slowly turning Firefox into a crappy clone of Chrome?
As soon as they get their financial independence. They're working on it, by the way, there was a deal with Yahoo earlier this year.
Year after year, the infamous "Science: It's a Girl Thing!" ad looks more and more politically correct. They were misunderstood visionary geniuses :)
If you don't use full-screen-as-a-separate-space, it only requires a little tweaking of default parameters to get something usable, even if not to your liking. (By the way, is there some setting so that the default action is zoom and the alt/option one is full screen ?)
In settings -> mission control, deactivate automatically rearrange spaces, which is probably why you think that some applications appear on multiple spaces: actually, it's only on a single space (by default), but the spaces keep being rearranged.
If you want a space to span all monitors, I guess you can do it here too, but I'm a fan of distinct spaces on distinct screens, a feature I had 15 years ago on X11 (but which implied the inability to move windows between screens). Note that GNOME 3 does (did?) something "interesting" by default, which is a single desktop on the secondary screen, and virtual desktops on the primary one. Probably useful on laptops in presentation mode. (Is the default over-ridable) ?
In settings -> keyboard -> shortcuts, activate Mission control per-space bindings, so that you can swap between spaces with ctrl-(number). The catch is that you have to organize your spaces by task. You can then bind applications to some spaces. You can do that by alt-clicking in the dock and looking in options. Older versions of Mac OS X with Spaces (instead of Mission Control) had a list of assignations.
The space selection widget is gone though, if you won't use shortcuts, it's F3. Each space is rendered as a thumbnail on the top, you could choose a different background for each space if it helps you. The bottom part is all of the windows in the currently selected space. What is not very obvious is finding the currently selected space, depending on your color scheme: the white border is not very visible on white background.
You forgot to profess your love for Visual Studio, and how you used to program with vi or emacs.
Because apart telling us than Windows 10 now has virtual desktops and how bad a job you're doing at organizing them, I don't know what was your point.
If you can't keep your stuff organized, try invoking $MOTHER, because your issues are not technical in nature.
you won't have ads. You'll just pay directly for the services you want to access.
I thought that if American cable TV ever taught us anything, it is that you will end up paying for the service and have ads.
Just repeatedly ask "Have you spied on me yet?" Eventually, they'll snap.
Nah, they're an administration. They're experts. You will snap.
You mean that automakers are allowing the police to stop people's vehicles at any time for any reason, remotely.
That technology already exists. It's usually called a police roadblock.
In these trying times, one shall grant his contempt sparingly, for the needy are plentyfull -- François-René de Chateaubriand