Open-ended questions are great, but in order to answer you should have some idea of the domain of the desired answer. Am I meant to give the purely mathematical response, "There is not enough information"? Or to say "a non-negative number" because we know a living person's age must be a positive value? Should I specify more closely as "a non-negative real number" because I know an age can have fractional units and has no imaginary component? Am I meant to use my knowledge of the world to give a possible range -- "I know that you need to be 18 to be licensed as a ship's captain and the mandatory retirement age is 65, so the answer is bounded by those two values"? Is a probability wanted -- "Assuming a gaussian distribution of ages between 18 and 65..."? Or if it's a creativity exercise should we go a different direction and write a backstory for the captain, giving his age and motivations for captaining a ship full of livestock?
Hopefully the students had some guidelines regarding the sort of answer the exam was looking for.
For me the money quote is this: "Using a crowdsourced database of impersonator accounts, carefully curated by us to avoid any false positives [...]" Carefully curated how, exactly? What vetting did they do to make sure they actually targeted sock puppets and not real users' accounts? The article gives no further information. What they're doing is the very definition of harassment, so if their careful curation fails in any way they end up harassing legitimate users. Legitimate users with admittedly distasteful views, but legitimate users nonetheless.
"Punch a Nazi in the face" is a great sentiment, but you'd better make for damned sure it's really a Nazi. False positives are unacceptable.
And if you're fortunate enough to have a car with Autopilot, the road ahead will suddenly turn icy.
As someone who spent four hours yesterday driving at half-speed on legitimately icy roads, I feel obliged to say, fsck you and the reindeer you rode in on, Tesla.
Yup. Assuming that your plan still existed, you could keep it. Obama didn't lie about that. And it was a pretty fair assumption, since there was nothing in the ACA that required insurers to stop offering the existing plans. The big problem is that the insurance companies decided that this was the perfect opportunity to re-organize their offerings and incidentally jack up prices, crying "It's all Obamacare's fault!" the whole time. They'd have you believe the changes were required by the ACA, but the truth is they simply weren't prohibited by it. And insurance companies are run by greedy bastards.
I about choked on my coffee the other day when one of our IT guys was called to fix someone's computer in the middle of a meeting. He said, without the slightest trace of humor or irony, "Have you tried turning it off and on again?" Which, of course, fixed it.
So, what you're saying is that when you rush out a patch, the development and QA processes suffer? The hell you say. No one could have predicted *that*.
Sometimes you have to say "Make it work for the most common case *now* and we'll pick up anything we missed later.
The FCC has a published API for bulk submission of comments. Why is anyone surprised that someone took the phone book and posted a bunch of bogus comments from everyone from Aaron Aaronson to Zebediah Zywiec?
https://www.fcc.gov/ecfs/public-api-docs.html
I agree with Thelasko: We don't need to see the source code, we just need to see the results. Evaluate it as a black box -- feed it known samples and see if it produces the correct results. If so, it's reasonable to admit the machine's evaluation.
What if there's some flaw that affects a small fraction of cases? It's possible, but it's also possible that code inspection wouldn't find the problem either. There's always another test, another set of eyes, another *something* that could be done. At some point you have to just say that yeah, it's probably okay. For this purpose I think it's reasonable to accept the results of the black box testing. White-box testing and code inspection is beneficial but in this case I don't think it's necessary.
In a court of law, everything comes down to testimony. Any technological test is only as good as the expert presenting it. They're telling the court that in their opinion, this is what the results of the test say. Opposing witnesses can question it, of course. The science is never really examined; look at all the half-assed forensic tests that have been presented and are still being presented. A rigorous black box test of any new technology would be infinitely better than what we have now.
Frankly I think it's likely that the reason they want to keep the code secret is that there's no actual science to back it up; that even if the code is 100% correct, it's correctly implementing a flawed premise. Think of a polygraph. It doesn't matter if it's a digital system or good old-fashioned needle traces on paper. Even perfectly implemented, it's still a shit test. Fortunately a well designed black box test should be able to determine this in short order.
So, a tech company puts a software feature into a high-end model but doesn't enable it in the low-end model even though the low-end is technically capable of it.
Isn't that the way software has worked since, like, forever? They were doing this back in the mainframe days. How and why is this a surprise to anyone?
Next up: Ford could have put leather seats into the base-model Mustang but didn't. They mysteriously only show up in the "performance" trim, even though they have nothing to do with performance! Why they did it will shock you!
The first thing you have to keep in mind about Q: John de Lancie is fucking awesome and any excuse to have him play a recurring character is a good one.
Meanwhile, his kid / ex / obnoxious coworker is giggling in delight over the fun they've caused with a simple "spacebar produces two spaces" keyboard mapping when the author wasn't looking.
Is it even possible to say "The Year of the Linux Desktop" non-sarcastically any more? It's like saying nuclear fusion is five years away. It's been said for so long nobody could possibly take the phrase seriously anymore.
Maybe the title of the presentation was intended to be a joke, with the punch line that it's being presented from a Mac?
I just want to know why, when there are just two data points, the OP chose to extrapolate with an exponential curve instead of a line. You could just as well say "it's hyperbolically cheaper to produce". That'd probably be better anyway, given that the phrase itself is meant as simple hyperbole.
I'm curious how many programming jobs are code monkey jobs, and how many are fairly unique cases.
I suspect a lot of programming jobs are fairly unique. In my case, the first 15 years of my career were in embedded systems. None of that stuff was straightforward. Lots of custom hardware, no off-the-shelf solutions available. One job was practically a research position. We were making a product, but even the scientist who invented the technology didn't know why it worked. There was a ton of experimental work to figure out the properties of the system to take it from lab project to saleable product.
For the past 10 years I've been involved in network packet analysis. Again, not straightforward, lots of interesting problems to solve. I've never had to work in a "code monkey" situation where I was doing straightforward implementation of someone else's design. Everything I've done professionally has been unique and complex. And fun. Now if you'll excuse me, I have to get back to figuring out why this particular event results in three seconds of network saturation.
You do realize that I spent an entire post explaining why I thought the conclusions in the report were not justified based on the data given in the report, right?
The report can be found at the Royal Society for Public Health site, but it's really just a summary and conclusions. I have not been able to find the actual data, or examples of the survey.
I'd like to see the raw data. Based on what's in the report I'd have to say I disagree with the conclusions. Twitter, Facebook, Snapchat, and Instagram all show small net negative results, but YouTube shows a large net positive. "Social media", taken as a whole, has a small net positive. This is hardly a "harm to young people's mental health" and a need to take action as the RSPH is demanding.
Of course, the devil's in the details, which are sadly missing. The survey called out 14 factors and asked participants to rate how much each social media platform affected them. The two biggest negative factors across all platforms are "sleep" and "fear of missing out". These don't sound like things that are going to destroy western civilization. On the other hand, all platforms were rated a medium-to-large net positive on "emotional support", "self-expression", and "self-identity". Which do you value more? There's no discussion about how the categories were weighted.
I'd also like to know more about the methodology. How were the 1479 people selected? It makes a difference if, say, 1500 surveys were sent out and they got a 98% response rate, or if 150000 surveys were sent and fewer than 1% could be arsed to send them back. In the former case the data is likely to be representative of the population as a whole, but not so in the latter.
So I'm suspicious that the data were selected to support a pre-determined conclusion. But even taken at face value, the data as shown in the report show only a slight net effect, and a definite positive effect in several important areas. I just don't see any justification for the report's own calls to action.
Slashdot Mirror
Open-ended questions are great, but in order to answer you should have some idea of the domain of the desired answer. Am I meant to give the purely mathematical response, "There is not enough information"? Or to say "a non-negative number" because we know a living person's age must be a positive value? Should I specify more closely as "a non-negative real number" because I know an age can have fractional units and has no imaginary component? Am I meant to use my knowledge of the world to give a possible range -- "I know that you need to be 18 to be licensed as a ship's captain and the mandatory retirement age is 65, so the answer is bounded by those two values"? Is a probability wanted -- "Assuming a gaussian distribution of ages between 18 and 65..."? Or if it's a creativity exercise should we go a different direction and write a backstory for the captain, giving his age and motivations for captaining a ship full of livestock?
Hopefully the students had some guidelines regarding the sort of answer the exam was looking for.
For me the money quote is this: "Using a crowdsourced database of impersonator accounts, carefully curated by us to avoid any false positives [...]" Carefully curated how, exactly? What vetting did they do to make sure they actually targeted sock puppets and not real users' accounts? The article gives no further information. What they're doing is the very definition of harassment, so if their careful curation fails in any way they end up harassing legitimate users. Legitimate users with admittedly distasteful views, but legitimate users nonetheless.
"Punch a Nazi in the face" is a great sentiment, but you'd better make for damned sure it's really a Nazi. False positives are unacceptable.
Should regulators force tattoo parlors to use erasable ink? Things I expressed in 2011 may not represent me today!
As someone who spent four hours yesterday driving at half-speed on legitimately icy roads, I feel obliged to say, fsck you and the reindeer you rode in on, Tesla.
Yup. Assuming that your plan still existed, you could keep it. Obama didn't lie about that. And it was a pretty fair assumption, since there was nothing in the ACA that required insurers to stop offering the existing plans. The big problem is that the insurance companies decided that this was the perfect opportunity to re-organize their offerings and incidentally jack up prices, crying "It's all Obamacare's fault!" the whole time. They'd have you believe the changes were required by the ACA, but the truth is they simply weren't prohibited by it. And insurance companies are run by greedy bastards.
Next up on Slashdot, "Old Man Yells At Cloud". What happened next will astonish you!
I about choked on my coffee the other day when one of our IT guys was called to fix someone's computer in the middle of a meeting. He said, without the slightest trace of humor or irony, "Have you tried turning it off and on again?" Which, of course, fixed it.
Who's a good bot? You're a good bot! Yes you are. YES YOU ARE!
I'm waiting for #alt_binaries where people post ripped blu-rays to twitter one base64 line at a time.
So, what you're saying is that when you rush out a patch, the development and QA processes suffer? The hell you say. No one could have predicted *that*.
Sometimes you have to say "Make it work for the most common case *now* and we'll pick up anything we missed later.
I just want to see him drive it down from orbit, like in the opening of Heavy Metal.
The FCC has a published API for bulk submission of comments. Why is anyone surprised that someone took the phone book and posted a bunch of bogus comments from everyone from Aaron Aaronson to Zebediah Zywiec? https://www.fcc.gov/ecfs/public-api-docs.html
I agree with Thelasko: We don't need to see the source code, we just need to see the results. Evaluate it as a black box -- feed it known samples and see if it produces the correct results. If so, it's reasonable to admit the machine's evaluation.
What if there's some flaw that affects a small fraction of cases? It's possible, but it's also possible that code inspection wouldn't find the problem either. There's always another test, another set of eyes, another *something* that could be done. At some point you have to just say that yeah, it's probably okay. For this purpose I think it's reasonable to accept the results of the black box testing. White-box testing and code inspection is beneficial but in this case I don't think it's necessary.
In a court of law, everything comes down to testimony. Any technological test is only as good as the expert presenting it. They're telling the court that in their opinion, this is what the results of the test say. Opposing witnesses can question it, of course. The science is never really examined; look at all the half-assed forensic tests that have been presented and are still being presented. A rigorous black box test of any new technology would be infinitely better than what we have now.
Frankly I think it's likely that the reason they want to keep the code secret is that there's no actual science to back it up; that even if the code is 100% correct, it's correctly implementing a flawed premise. Think of a polygraph. It doesn't matter if it's a digital system or good old-fashioned needle traces on paper. Even perfectly implemented, it's still a shit test. Fortunately a well designed black box test should be able to determine this in short order.
Pisses off 4chan? That by itself is reason enough for the change! +1, SHIPIT
So, a tech company puts a software feature into a high-end model but doesn't enable it in the low-end model even though the low-end is technically capable of it.
Isn't that the way software has worked since, like, forever? They were doing this back in the mainframe days. How and why is this a surprise to anyone?
Next up: Ford could have put leather seats into the base-model Mustang but didn't. They mysteriously only show up in the "performance" trim, even though they have nothing to do with performance! Why they did it will shock you!
The first thing you have to keep in mind about Q: John de Lancie is fucking awesome and any excuse to have him play a recurring character is a good one.
There is no second thing.
Meanwhile, his kid / ex / obnoxious coworker is giggling in delight over the fun they've caused with a simple "spacebar produces two spaces" keyboard mapping when the author wasn't looking.
Is it even possible to say "The Year of the Linux Desktop" non-sarcastically any more? It's like saying nuclear fusion is five years away. It's been said for so long nobody could possibly take the phrase seriously anymore. Maybe the title of the presentation was intended to be a joke, with the punch line that it's being presented from a Mac?
I just want to know why, when there are just two data points, the OP chose to extrapolate with an exponential curve instead of a line. You could just as well say "it's hyperbolically cheaper to produce". That'd probably be better anyway, given that the phrase itself is meant as simple hyperbole.
In other words, this exists because somebody wanted to find out if his girlfriend was cheating on him.
I suspect a lot of programming jobs are fairly unique. In my case, the first 15 years of my career were in embedded systems. None of that stuff was straightforward. Lots of custom hardware, no off-the-shelf solutions available. One job was practically a research position. We were making a product, but even the scientist who invented the technology didn't know why it worked. There was a ton of experimental work to figure out the properties of the system to take it from lab project to saleable product.
For the past 10 years I've been involved in network packet analysis. Again, not straightforward, lots of interesting problems to solve. I've never had to work in a "code monkey" situation where I was doing straightforward implementation of someone else's design. Everything I've done professionally has been unique and complex. And fun. Now if you'll excuse me, I have to get back to figuring out why this particular event results in three seconds of network saturation.
You do realize that I spent an entire post explaining why I thought the conclusions in the report were not justified based on the data given in the report, right?
Sorry, I linked the wrong page on the RSPH site. Here is the report: #StatusOfMind
The report can be found at the Royal Society for Public Health site, but it's really just a summary and conclusions. I have not been able to find the actual data, or examples of the survey.
I'd like to see the raw data. Based on what's in the report I'd have to say I disagree with the conclusions. Twitter, Facebook, Snapchat, and Instagram all show small net negative results, but YouTube shows a large net positive. "Social media", taken as a whole, has a small net positive. This is hardly a "harm to young people's mental health" and a need to take action as the RSPH is demanding.
Of course, the devil's in the details, which are sadly missing. The survey called out 14 factors and asked participants to rate how much each social media platform affected them. The two biggest negative factors across all platforms are "sleep" and "fear of missing out". These don't sound like things that are going to destroy western civilization. On the other hand, all platforms were rated a medium-to-large net positive on "emotional support", "self-expression", and "self-identity". Which do you value more? There's no discussion about how the categories were weighted.
I'd also like to know more about the methodology. How were the 1479 people selected? It makes a difference if, say, 1500 surveys were sent out and they got a 98% response rate, or if 150000 surveys were sent and fewer than 1% could be arsed to send them back. In the former case the data is likely to be representative of the population as a whole, but not so in the latter.
So I'm suspicious that the data were selected to support a pre-determined conclusion. But even taken at face value, the data as shown in the report show only a slight net effect, and a definite positive effect in several important areas. I just don't see any justification for the report's own calls to action.
Young Health Movement