Re:Review Ruby for the perl enthusiast please
on
Ruby 2.0.0 Released
·
· Score: 1
Although some very clever Ruby runtime implementers have come along to pick up the slack left by the language's founder (who still pretends the global interpreter lock is a virtue, or so I am told)
You probably shouldn't treat gossip as reliable. The Ruby GVL (the Ruby equivalent of the Python GIL) is an artifact of the conversion to the native-threading YARV VM (originally, one of the alternative runtimes) from the old original green-threaded implementation, Matz has never (at least that I've seen, and it would be out of line with a lot of what I have seen him say) described it as a "virtue". YARV was just the most-alround-ready-for-mainstream-use native-threaded alternative runtime, and was chosen because better-than-what-we-have now is better than perfect-sometime-later. There's already quite a bit of talk in the community about moving the mainline to Rubinius once it is ready-enough to be the mainstream.
The gossip you heard might have evolved from something Matz has actually said, which is that *given* the rest of the current mainline implementation, replacing the GVL with fine-grained locks (which has been *done* by the mainline maintainer, though not released) proved to degrade performance of single-threaded code in a way which would negatively impact lots of current users. This isn't saying the GVL is a virtue, its saying that the current efforts to replace it in the context of the current mainline implementation have proven to not meet Matz's gain-for-pain tradeoff preferences.
This is another failing of Science Channel styled science shows*. They neglect to tell you that light doesn't escape because the gravity well created by a black hole warps space, not because photons are pulled on by gravity.
Most of the Science Channel-style science shows I've seen that cover the issue not only cover that light doesn't escape the gravity well because the gravity of the black hole warps space, but also covered that that's how all gravity works, not just a special variation related to "black holes" as the source or "light" as the affected entity. (As do the more technical, less Science-Channel-ish, works that I've seen addressing the same subject matter.)
It may sound like I'm splitting hairs since the overall end result is the same but a lot of people mistake it as meaning that light is sucked in to the black hole because particles with mass are also sucked in. This also doubtlessly leaves people scratching their head over the misconception that maybe the gravity is forceful enough to actually attact the light.
The "sucked in" analogy is exactly as accurate (or inaccurate) applied to light as it is to "particles with mass".
It takes a special kind of terrorist to deploy a bioweapon, because bioweapons don't distinguish based on religion (although you could theoretically make one that distinguishes on race, it's a bit tricky).
Race is a social construct that has only a loose connection to biology, so it would only be even theoretically possible to have a bioweapon that distinguishes by genetic (or other biological) characteristics that loosely correspond to race, rather than race itself.
Really? I've been using Windows 8 for a while, and I "never" use any Windows 8 Metro (new UI) apps.
I don't remember the exact workflow, but on the one Win8 machine in my family that I've gotten called in to help with, just dealing with Microsoft updates required interacting with both New UI and desktop components.
It's not clear to me whether this ability to execute code is intended solely for Chrome OS, or whether it is intended for all versions of the Chrome browser.
Native Client has been in all versions of Chrome (well, except Chrome for iOS) and enabled-by-default for apps from the Chrome Web Store since Chrome 14; there are a number of apps that leverage it in the store.
If the intent is the latter, this has a good chance of driving users en masse away from Chrome as Google's security nightmare is probably just beginning.
The reality, not merely the intent, has been the latter for quite some time, and it hasn't driven users away from Chrome.
You really have no idea what you are talking about here.
While I think anyone has to be impressed by how extensible the browser and HTML has been and how far it's all been able to go, are we going to at some point face the fact that we're using the browser for something it was never intended for?
No, because what browsers are intended for has changed.
We want a browser experience that feels like a native app, but we shun things like flash and silverlight (and even java!).
This is simply equivocation: the "we" isn't the same group in both parts of this sentence? The people who I've seen who shun those kinds of technologies very often state that they prefer apps that "feel like native apps" to actually be native apps, not part of the "browser experience".
Don't we need to eventually concede the possibility that something like Silverlight wouldn't be that bad?
The people that want pure native apps and don't want native-like web apps certainly don't need to concede anything of the sort. The people that want native-quality web-delivered apps obviously need some kind of platform that can provide that, though there is no reason the appropriate low-level functionality built into browsers and exposed through HTML elements and/or JavaScript APIs and tied together with JavaScript code can't do virtually all of that; the main problem is defining what the base functionality needed is, getting the appropriate HTML elements and JavaScript APIs widely supported, and the fact that it means building new apps from the ground up to migrate from existing native apps.
Where technologies that let you use something very similar to the desktop code for web apps with native performance (Java, Silverlight, and Native Client all fit this description to a certain degree, though they aren't necessarily deeply similar) helps this isn't so much in making native-app-like performance practical as in facilitating porting existing native apps, or in easing the process of developing native and web apps in parallel. This is useful (and helps address a chicken-and-egg problem in establishing native-quality web-delivered applications), but isn't really all that fundamentally essential to building them.
There are some things TNG predicted well, but a few glaringly funny missteps in retrospect. My two favorite are:
1) Piles of PADDs. There's a few scenes where someone is "doing a lot of reading" or "has a lot of reports to file" and so they have a bunch of PADDs strewn about their desk. Little did I know I needed a separate Kindle for each ebook I read.
Lots of the time, they are cross-referencing things in parallel, which is inconvenient on a single screen of that size. With replicators, PADDs are presumably literally as cheap as dirt, rather than luxury gadgets, so there's no real reason not to have one for each document when you need to do that.
Star Trek is actually a great illustration of this, there were times in the original series where the actors had their hands on controls but attention focused on the action for dramatic effect, they didn't need to constantly look down as in the Next Generation.
They didn't actually constantly look down in Next Generation, nor did they need to (either actually or in-fiction.) Actually, because the controls weren't on the props but were digitally added in post production, and in-fiction because they used fancy adaptive interfaces that were customized to the individual user so that your controls would be exactly where they were most comfortable for them to be for you (my Trek lore is not good enough to recall whether there were also supposed to be haptics involved or not.)
This is what you get with proprietary software. Under that kind of software, you don't get the tools and legal permissions to do that kind of stuff.
Yes, not being able to do basic support yourself or contract it out to an independent third-party is a limitation that comes with closed-source software. That's why it is reasonable to demand greater support (both in terms of scope and lifetime) from closed-source vendors.
You got exactly what you bought, nobody promised you would get any sort of benefit once they stop providing updates for that product.
The question wasn't about whether or not Microsoft broke the purchase agreement with their EOL decisions, the question was about why people on Slashdot held Microsoft to a different standard as to when they should EOL an operating system than is applied, by the same people, to open-source OS vendors.
Even the hated "start screen/new UI/Metro" performancs exactly the same as the Win 8 start menu for the majority of people that use it
Assuming you mean Win7 start menu (there is no Win8 start menu), this is just wrong. It doesn't work at all the same.
1. Click the lower left of your screen (Windows 7 and 8) 2. Type the first 3 or 4 letters of the program you want to run (Windows 7 and 8)
The vast majority of people I know that use the Start Menu don't use click-and-type; the majority use click-and-click, and the minority that type use Start-key-and-type. But, yes, on the level you describe, there are similarities.
The difference is that the Win8 UI mechanism is visually different in a distracting way (and it goes beyond the menu, since for many tasks you end up using a mix of Win8 UI and traditional desktop UI apps, including for basic configuration.) I was a fan of lots of the principles of the Metro UI design language when it was published, but the actual Win8 implementation -- especially the way the desktop and Win8 UIs are combined on desktop systems -- is a big loss from the more consistent UI of Win7. Its possible that its a waystation on the road to an improved consistent UI in Win9 or later, but that remains to be seen.
Is Debian 2.2 still receiving updates? (It's not) Then why are you expecting Microsoft to still provide updates to XP?
Because they won't give me (in the general sense: I don't personally have any interest in doing it myself, personally) the tools and legal permissions to do it myself, that's why.
I don't know whether this has been mentioned before, but the big problem with Google's approach is that it won't allow me to define a formal grammar as the "set of things the user might reasonably say".
AFAICT from reading the API spec and surrounding information is that's not a problem with "Google's approach", its a problem with the fact that the W3C Speech API Community Group couldn't come to a consensus on the grammar format(s) to support in the Web Speech API, so that while the API adopted in the group's final report specifies containers for grammars and how to attach them to recognition requests, it doesn't specify any actual formats for grammars.
Since the mission of the group was to come up with a consensus limited-subset specification as a step on the road to a specification that would meet the full set of use cases set out by the Final Report of the Speech Incubator Group's final report, it makes sense for an implementation of the Web Speech API not to adopt an approach on grammars that would fail to be forward-compatible with the anticipated future specification, since that would encourage building applications that would be broken under the expected follow-on full-featured speech API, which can be expected to retain compatibility with the limited-subset API in the areas covered by that API, but is less likely to do so in areas not covered by the limited-subset API.
Eliminating waste -- costs that have been incurred (e.g., by investing programmer-hours in development) that are not delivering value (e.g., by being incorporated in a shipping product) are a form of waste. Basic application of Lean principles.
What's with the bloat?
One user's bloat is lots of other users' value.
Why don't they address speed and stability bugs that have been open for two years?
Probably because for those particular issues, the expected value of the fix (given the severity and the conditions which trigger them) divided by the expected cost is lower than for other fixes and features, and Google is focussed on doing the work that matters most (highest value for cost) first.
Google is the only one still on 32-bit. I'm going to put that down to laziness. This is the company after all that relied on wine on Linux for some of their software. For such a bunch of intelligent people they're lazy
Not doing extra work beyond that which is required for the goal you are attempting to achieve is efficient, not lazy (well, it isn't lazy-as-in-indolent, it is lazy-as-in-Haskell.)
Rates yes, revenue no. Every time the president states that revenue must be part of the deal, he is clouding the issue. The government is certainly able to legislate tax rates, but has no control over tax revenue.
The only way this can not be completely self-contradictory is if you assume that tax rates have no effect whatsoever on revenue. This is, to say the least, and extraordinary claim for which some evidence should be provided, and...
Smart tax policy will increase revenue if it can be supported. Stupid tax policy will lead to lower revenue.
...there is no way that this pair of sentences is consistent with "The government [...] has no control over tax revenue." If there are polciies that will increase revenue and policies that will reduce revenue, and government has the power to choose between policy courses, then government has control over tax revenue.
When the federal reserve is engaged in "quantitative easing" ie. flooding the bond market with cash, the government is in effect cratering the price
Cratering the price of treasury securities would be increasing effective rates. Which, even if it was true (which I don't see how you get from QE to that effect), would make the low rates (=high prices) of treasury securities more significant, not less.
You're operating under the misapprehension that the Free Market (peace be upon it) is the only thing that sets interest rates in the US.
The market is the thing that actually sets the interest rates that are relevant to the government's ability to raise funds through borrowing (those on Treasury securities).
It's not; the Federal Reserve board sets at least one key rate.
The Federal Reserve board sets the Federal funds rate, which is a rate for interbank loans, and has no direct bearing on the issue here. It is an interest rate, its just the interest rate that matters to the issue.
Wrong. The interest rates are low to stimulate the economy.
The Federal funds rates (which has knock-on effects on various lending and account interest rates) is low for that reason.
The low effective interest rate on US treasury securities (which includes the discount rate on T-bills, which technically aren't interest but amount to the same thing), which is what is relevant to the US governments ability to raise money, is because US treasury securities are widely seen as a very secure investment, and investors demand a low risk premium.
So, someone is rude, arrogant, flaunting, etc., and that is legal grounds for property seizure?
Whether or not they are, if someone refuses to sign a document which must, by law, be signed before an article can be released from government custody, whether because they dispute the accuracy of the document or for whatever other reason, then there is no legal grounds for releasing the article from government custody.
If the complaint was correct and the agent was authorized to make or accept alterations to the form -- and Arrington's doesn't provide much support for anything on either point more than Arrington's belief that the form was incorrect because the valuation on the form in USD matched the original purchase price in CAD, which may or may not be wrong -- then there was something done wrong in not correcting the form.
Customs didn't impound the shipment until the paperwork gets straightened out.
Yes, they did.
The only part that needs straightened out is his signature.
when the guy you're so angry at for having made more money than you refuses to authenticate paperwork listing the value of imported goods with a price that's likely in excess of $10,000 away from being correct, the customs agent didn't say "Well, until this paperwork error is corrected, we can't let you have your boat", she said "The paperwork is correct! Since you won't sign it, hand over the keys and get off the boat"
Assuming that the CBP has a standard method of calculating valuation (which is always approximate: there is a difference between "price" and "value") of the import based on purchase records and it was correctly applied (which may well use the exchange rate and the time of import, for purchases that are recent enough that the purchase price is even relevant to assessing the valuation), then the form was correct, Arrington's assumption that the form was incorrect because the valuation in USD happened to match the CAD purchase price was wrong, and the agent ws correct in what she said.
Dude, seriously, it's in the article... they took the Canadian dollar value, turned it into American dollars (incorrectly), and asked him to sign a form under oath that what the form said was true.
The form almost certainly cares about the value at the time of import (since customs duty is based on value of the import, not the USD-value-of-the-foreign-purchase-price-of-the-import-based-on-exchange-rate-at-the-time-payments-were-made), and requires the value to be declared in US dollars, not foreign currency, and Canadian dollars currently trade at very close to 1:1 with US dollars (which Arrington acknowledges, though he states that wasn't true when he made most of the payments on the boat, but doesn't explain why he thinks that is relevant.) Given that the currency exchange rates are fluctuating and, at any given point in time, there is a range of plausible valuations though the current price on any particular exchange will be fixed, the exact CAD purchase price is as accurate as any other valuation in USD.
If Arrington refuses to sign the form because he disputes the CBP's valuation (and, therefore, the amount of import duty that must be paid prior to releasing the boat) then, whatever the basis for the dispute of the valuation, it needs to be resolved before the boat can be released.
Now, maybe if more details are available there is information which supports this being something done wrong bu Customs and Border Protection, but right now it looks like Arrington was presented an accurate form.
The DHS had the authority to hold the boat the minute it arrived on US soil, for customs clearance.
Where in the Constitution does it say that?
DHS's power is specified by statute adopted by Congress. The authority for Congress to adopt the statute giving the authority to DHS to do this is found in Article I, Section 8, Clause 3 (popularly know as the commerce clause), which states that Congress shall have the power "[t]o regulate commerce with foreign nations, and among the several states, and with the Indian tribes;"
To be honest, I'm having trouble finding any reference in that document that would authorize a Department of Homeland Security at all...
The Constitution isn't particularly concerned with how Congress names the particular agencies it creates in the exercise of its enumerated powers, its more concerned with the substance of what government does.
Not very relevant, since I clearly wrote "most online courses," which are not "traditional" or for credit.
Granted, it may not apply directly to TFA, it was a generalization. Are you asserting there is something wrong with that?
There is a difference between a generalization which merely doesn't apply to TFA and one which is refuted by TFA. Claiming, as you did, that the source of high drop out rates in online courses is that the incentive of credit isn't there and that the solution to that is "Put the incentive back in, and they will start keeping students" falls into the latter category, given that the actual studies addressed by TFA that provide the empirical evidence that there is a high drop-out rate in online classes are specific to online college courses that feature the incentive of regular college credit.
Slashdot Mirror
You probably shouldn't treat gossip as reliable. The Ruby GVL (the Ruby equivalent of the Python GIL) is an artifact of the conversion to the native-threading YARV VM (originally, one of the alternative runtimes) from the old original green-threaded implementation, Matz has never (at least that I've seen, and it would be out of line with a lot of what I have seen him say) described it as a "virtue". YARV was just the most-alround-ready-for-mainstream-use native-threaded alternative runtime, and was chosen because better-than-what-we-have now is better than perfect-sometime-later. There's already quite a bit of talk in the community about moving the mainline to Rubinius once it is ready-enough to be the mainstream.
The gossip you heard might have evolved from something Matz has actually said, which is that *given* the rest of the current mainline implementation, replacing the GVL with fine-grained locks (which has been *done* by the mainline maintainer, though not released) proved to degrade performance of single-threaded code in a way which would negatively impact lots of current users. This isn't saying the GVL is a virtue, its saying that the current efforts to replace it in the context of the current mainline implementation have proven to not meet Matz's gain-for-pain tradeoff preferences.
Most of the Science Channel-style science shows I've seen that cover the issue not only cover that light doesn't escape the gravity well because the gravity of the black hole warps space, but also covered that that's how all gravity works, not just a special variation related to "black holes" as the source or "light" as the affected entity. (As do the more technical, less Science-Channel-ish, works that I've seen addressing the same subject matter.)
The "sucked in" analogy is exactly as accurate (or inaccurate) applied to light as it is to "particles with mass".
I'm not sure you are familiar with the distinction between an example and an exhaustive description of the scope of a problem.
Race is a social construct that has only a loose connection to biology, so it would only be even theoretically possible to have a bioweapon that distinguishes by genetic (or other biological) characteristics that loosely correspond to race, rather than race itself.
I don't remember the exact workflow, but on the one Win8 machine in my family that I've gotten called in to help with, just dealing with Microsoft updates required interacting with both New UI and desktop components.
Native Client has been in all versions of Chrome (well, except Chrome for iOS) and enabled-by-default for apps from the Chrome Web Store since Chrome 14; there are a number of apps that leverage it in the store.
The reality, not merely the intent, has been the latter for quite some time, and it hasn't driven users away from Chrome.
You really have no idea what you are talking about here.
No, because what browsers are intended for has changed.
This is simply equivocation: the "we" isn't the same group in both parts of this sentence? The people who I've seen who shun those kinds of technologies very often state that they prefer apps that "feel like native apps" to actually be native apps, not part of the "browser experience".
The people that want pure native apps and don't want native-like web apps certainly don't need to concede anything of the sort. The people that want native-quality web-delivered apps obviously need some kind of platform that can provide that, though there is no reason the appropriate low-level functionality built into browsers and exposed through HTML elements and/or JavaScript APIs and tied together with JavaScript code can't do virtually all of that; the main problem is defining what the base functionality needed is, getting the appropriate HTML elements and JavaScript APIs widely supported, and the fact that it means building new apps from the ground up to migrate from existing native apps.
Where technologies that let you use something very similar to the desktop code for web apps with native performance (Java, Silverlight, and Native Client all fit this description to a certain degree, though they aren't necessarily deeply similar) helps this isn't so much in making native-app-like performance practical as in facilitating porting existing native apps, or in easing the process of developing native and web apps in parallel. This is useful (and helps address a chicken-and-egg problem in establishing native-quality web-delivered applications), but isn't really all that fundamentally essential to building them.
Lots of the time, they are cross-referencing things in parallel, which is inconvenient on a single screen of that size. With replicators, PADDs are presumably literally as cheap as dirt, rather than luxury gadgets, so there's no real reason not to have one for each document when you need to do that.
They didn't actually constantly look down in Next Generation, nor did they need to (either actually or in-fiction.) Actually, because the controls weren't on the props but were digitally added in post production, and in-fiction because they used fancy adaptive interfaces that were customized to the individual user so that your controls would be exactly where they were most comfortable for them to be for you (my Trek lore is not good enough to recall whether there were also supposed to be haptics involved or not.)
Yes, not being able to do basic support yourself or contract it out to an independent third-party is a limitation that comes with closed-source software. That's why it is reasonable to demand greater support (both in terms of scope and lifetime) from closed-source vendors.
The question wasn't about whether or not Microsoft broke the purchase agreement with their EOL decisions, the question was about why people on Slashdot held Microsoft to a different standard as to when they should EOL an operating system than is applied, by the same people, to open-source OS vendors.
Yes.
No, its not.
Assuming you mean Win7 start menu (there is no Win8 start menu), this is just wrong. It doesn't work at all the same.
The vast majority of people I know that use the Start Menu don't use click-and-type; the majority use click-and-click, and the minority that type use Start-key-and-type. But, yes, on the level you describe, there are similarities.
The difference is that the Win8 UI mechanism is visually different in a distracting way (and it goes beyond the menu, since for many tasks you end up using a mix of Win8 UI and traditional desktop UI apps, including for basic configuration.) I was a fan of lots of the principles of the Metro UI design language when it was published, but the actual Win8 implementation -- especially the way the desktop and Win8 UIs are combined on desktop systems -- is a big loss from the more consistent UI of Win7. Its possible that its a waystation on the road to an improved consistent UI in Win9 or later, but that remains to be seen.
Because they won't give me (in the general sense: I don't personally have any interest in doing it myself, personally) the tools and legal permissions to do it myself, that's why.
AFAICT from reading the API spec and surrounding information is that's not a problem with "Google's approach", its a problem with the fact that the W3C Speech API Community Group couldn't come to a consensus on the grammar format(s) to support in the Web Speech API, so that while the API adopted in the group's final report specifies containers for grammars and how to attach them to recognition requests, it doesn't specify any actual formats for grammars.
Since the mission of the group was to come up with a consensus limited-subset specification as a step on the road to a specification that would meet the full set of use cases set out by the Final Report of the Speech Incubator Group's final report, it makes sense for an implementation of the Web Speech API not to adopt an approach on grammars that would fail to be forward-compatible with the anticipated future specification, since that would encourage building applications that would be broken under the expected follow-on full-featured speech API, which can be expected to retain compatibility with the limited-subset API in the areas covered by that API, but is less likely to do so in areas not covered by the limited-subset API.
Eliminating waste -- costs that have been incurred (e.g., by investing programmer-hours in development) that are not delivering value (e.g., by being incorporated in a shipping product) are a form of waste. Basic application of Lean principles.
One user's bloat is lots of other users' value.
Probably because for those particular issues, the expected value of the fix (given the severity and the conditions which trigger them) divided by the expected cost is lower than for other fixes and features, and Google is focussed on doing the work that matters most (highest value for cost) first.
Not doing extra work beyond that which is required for the goal you are attempting to achieve is efficient, not lazy (well, it isn't lazy-as-in-indolent, it is lazy-as-in-Haskell.)
The only way this can not be completely self-contradictory is if you assume that tax rates have no effect whatsoever on revenue. This is, to say the least, and extraordinary claim for which some evidence should be provided, and...
Are households sovereigns whose income is set largely by the family's choice of how much to tax?
If not, does it make sense to pretend that the budget of the government of a soverign nation-state differs only in scale from that of a household?
Cratering the price of treasury securities would be increasing effective rates. Which, even if it was true (which I don't see how you get from QE to that effect), would make the low rates (=high prices) of treasury securities more significant, not less.
The market is the thing that actually sets the interest rates that are relevant to the government's ability to raise funds through borrowing (those on Treasury securities).
The Federal Reserve board sets the Federal funds rate, which is a rate for interbank loans, and has no direct bearing on the issue here. It is an interest rate, its just the interest rate that matters to the issue.
The Federal funds rates (which has knock-on effects on various lending and account interest rates) is low for that reason.
The low effective interest rate on US treasury securities (which includes the discount rate on T-bills, which technically aren't interest but amount to the same thing), which is what is relevant to the US governments ability to raise money, is because US treasury securities are widely seen as a very secure investment, and investors demand a low risk premium.
Whether or not they are, if someone refuses to sign a document which must, by law, be signed before an article can be released from government custody, whether because they dispute the accuracy of the document or for whatever other reason, then there is no legal grounds for releasing the article from government custody.
If the complaint was correct and the agent was authorized to make or accept alterations to the form -- and Arrington's doesn't provide much support for anything on either point more than Arrington's belief that the form was incorrect because the valuation on the form in USD matched the original purchase price in CAD, which may or may not be wrong -- then there was something done wrong in not correcting the form.
Yes, they did.
The only part that needs straightened out is his signature.
Assuming that the CBP has a standard method of calculating valuation (which is always approximate: there is a difference between "price" and "value") of the import based on purchase records and it was correctly applied (which may well use the exchange rate and the time of import, for purchases that are recent enough that the purchase price is even relevant to assessing the valuation), then the form was correct, Arrington's assumption that the form was incorrect because the valuation in USD happened to match the CAD purchase price was wrong, and the agent ws correct in what she said.
The form almost certainly cares about the value at the time of import (since customs duty is based on value of the import, not the USD-value-of-the-foreign-purchase-price-of-the-import-based-on-exchange-rate-at-the-time-payments-were-made), and requires the value to be declared in US dollars, not foreign currency, and Canadian dollars currently trade at very close to 1:1 with US dollars (which Arrington acknowledges, though he states that wasn't true when he made most of the payments on the boat, but doesn't explain why he thinks that is relevant.) Given that the currency exchange rates are fluctuating and, at any given point in time, there is a range of plausible valuations though the current price on any particular exchange will be fixed, the exact CAD purchase price is as accurate as any other valuation in USD.
If Arrington refuses to sign the form because he disputes the CBP's valuation (and, therefore, the amount of import duty that must be paid prior to releasing the boat) then, whatever the basis for the dispute of the valuation, it needs to be resolved before the boat can be released.
Now, maybe if more details are available there is information which supports this being something done wrong bu Customs and Border Protection, but right now it looks like Arrington was presented an accurate form.
DHS's power is specified by statute adopted by Congress. The authority for Congress to adopt the statute giving the authority to DHS to do this is found in Article I, Section 8, Clause 3 (popularly know as the commerce clause), which states that Congress shall have the power "[t]o regulate commerce with foreign nations, and among the several states, and with the Indian tribes;"
The Constitution isn't particularly concerned with how Congress names the particular agencies it creates in the exercise of its enumerated powers, its more concerned with the substance of what government does.
There is a difference between a generalization which merely doesn't apply to TFA and one which is refuted by TFA. Claiming, as you did, that the source of high drop out rates in online courses is that the incentive of credit isn't there and that the solution to that is "Put the incentive back in, and they will start keeping students" falls into the latter category, given that the actual studies addressed by TFA that provide the empirical evidence that there is a high drop-out rate in online classes are specific to online college courses that feature the incentive of regular college credit.