I understand that there are all kinds of open source projects out there, and some better than others. But based on my personal experience with some of the more prominent ones, I seriously believe a government run by open source types would be as terrible as what we have now. The following thoughts are based just on those projects.
[...]
The Chewbacca defense will actually work.
[...]
No it wouldn't, since open source types will remember that Chewie is able to pilot a starship and is hence able to live wherever he wants.
Internet radio is usually implemented as a TCP stream, because the order of arriving packets usually matters in a streaming radio setting.
That's true for podcasts, but for live stuff the savvy internet radio implementations will be using RTP over UDP (a dropped packet is irritating, but better to keep the stream going rather than stopping until the missing stuff is resent). When it's a big event, it can be better still to use multicast (assuming you can make it work; some routers still suck in this area IIRC).
This all applies much more so to live internet video streaming.
actually, it's just the gstreamer guys saying that, and only because they don't understand why KDE developers would want to use a Qt/KDE style API for their sound, when they could use a horrible_gnome_style_api_instead.
Nobody other than Qt/KDE developers ought to care what the Qt/KDE API for sound is, just as GTK/Gnome devs are the only folks who ought to care about that sound API. What sucks is that there are too many sound system APIs; entire separated sound stacks. There should be a common base API (at the "send this sort of message over that channel" level) so that the type of hardware and the codecs that you use are independent of the GUI system that you use.
But the author's complaint is pretty much BS -- they're assuming that this is a problem for commercial vendors.
Of course it's not a problem for commercial vendors. They can just not support Linux. For many, that's not even going to cost them a particularly large fraction of their customers.
Sometimes you really do want to run commercial software, e.g. because it does things that aren't available to the same quality in OSS. The classic example is Skype; yes, there are open alternatives, but if there's a choice of using the closed code and being able to communicate with others or using the open code and being stuck on your own, well, that's not much of a choice. Anyway, my point is let's not be too hostile to commercial code. (Of course, positively supporting open code is good; it's trying to stop closed code that is a problem.)
If copyright terms were pulled back to something sane, such as 10 or 15 years [...]
That'd be really good, as it would allow those of us who don't like the FSF's political positions to contribute to (old versions of) Emacs without compromising our principles or our ability to take our contributions closed-source once we've got people hooked!
(Tongue in cheek, but only partially. Whenever you hack with the legal system in a deep way, changes can be far reaching.)
That's only part of it. Different apps have different modes of use. Some apps shouldn't be learned because people don't use them for long periods: they should just be usable straight away. Others will be used daily for years by trained experts, and the learning phase will be comparatively short: the real value there is in making the users productive, and the UI can be deep and complex. Those definitely aren't the only possibilities either; there is a sophisticated spectrum, and since the reason for that is "people", that isn't changing.
If you want to appear less ignorant, try finding a copy of About Face by Alan Cooper. I read the first edition long ago, and it was great then (though dated). The third edition (which ought to be using much more relevant examples if nothing else) was published last year, and so should be in your local good tech library.
At one place I worked, the guy who wrote up the coding standard explicitly prohibited jokes in comments and humorous variable names.
Add a comment to the code against the most business-critical part of the code something like "Hmm, I hope SOX-compliance never looks at this." and sign it with his name. The beauty is it's both not at all funny and hilarious, depending on your point of view.
It's the same kind of security you run into in large corporations. You know, you have to use a randomly generated 18 character alphanumeric password and it changes every 90 days... which is great except that when you go to do your timesheets you have to enter your LAN password... which goes over the wire plaintext encapsulated in an HTTP POST query. Oops. Also, because not everybody's memory is so great, it becomes common practice to keep the 18 character passwords written on sticky notes.
Yikes! That's insecure! Better require that at least a third of the characters are symbols, that both upper and lower case are used. Increase the of changes frequency to every 10 seconds, and require that people never reuse any of their passwords or their account will be locked until they've satisfied the gestapo^Wcorporate security team that they're not trying to hack the system!
And don't forget to provide everyone with post-its...
(I do real security. I work with people who do real security. Password anal retentiveness never ever helps; it just pisses users off. Using properly set up SSL/SSH for everything really does help. Network traffic monitoring - of TCP and UDP headers, not content - really does help, especially with detection of what happened. Firewalls/NAT do help, if used carefully; reduction of the attack surface to things that are vetted and monitored for intrusion really is a very strong technique.)
On the other hand, it's pretty easy to guess that you couldn't fix a light in another building from your phone. And that a Citroen C4 doesn't transform into a dancing robot
What is the deal with Starbucks and their imitators burning their beans?
It's also important what type of drink you're making from it. You need a different type of roast to make a good filter coffee from a good espresso (e.g. Italian roasts tend to be great for espresso, but don't drink filter coffee in Rome as they really don't understand it; the Nordic countries do far better at filter coffee). You may need a different roast if you're adulterating the coffee with milk or sugar or even just hot water.
The other thing is whether the coffee is burnt during or after final making. If it's made too hot, it burns and tastes nasty, just as if it is kept too hot it's horrible too.
IME at one *specific* Starbucks, their filter coffee is great neat black at 07:20 in the morning, but nowhere near as good 30 minutes later. I make no claims for this result being extensible to any other part of the chain, and nor am I telling you which store; I don't want people getting in my way at all before I get my coffee...
At this point it seems obvious that you've assumed a really simplistic model of a neuron that can compute a synaptic value in a single floating point operation. These simple neuron models don't behave like a real brain, and scaling up simulations of them doesn't produce anything interesting. Real neurons are capable of computing much more complex functions than these models. The throughput on the interconnect is going to be a major factor, and simulating each neuron will require from 10s to 1000000s of operations depending on the level of biological realism that is required.
The real question from an AI perspective is whether all that detail is necessary. Do we need to simulate individual signaling molecules? Do we need to simulate individual synapses? Can we simulate things at a higher level than neurons and still get a functionally similar model?
Obviously, for some things we definitely can't abstract (such as the effect of certain kinds of small molecules on consciousness). But nobody really has any idea whether general AI needs that level of detail. My hunch is that it doesn't, just as you don't need to know how transistors really work to write a computer program or even to simulate a computer with a pretty high degree of accuracy. But that is just a hunch...
Bingo.:) Idealism is a luxury for the rich. We don't have the privilege of questioning our 'superiors' on their ethics.
Next up you'll be suggesting that we should touch our forelocks, drop our pants and grin happily when we take it in the ass from our "superiors". Well that's BS. Hold them to ethics at all times, and tell other people that you're doing this. There's never any reason to accept your bosses being total scum.
This sounds like a good idea on the surface -- it'll never happen, of course, because too many companies and individuals have too much invested in the.com,.net, etc. without the country codes... but still, I like the consistency it all brings.
If you've got e-mobsters smacking companies around a lot with DNS cracks because.com, etc. aren't signed, you will see migration to signed domain trees. And masses of bitching. (You'll always get that.) Or possibly people stopping the arguments over root domain signing and getting on with it by appointing someone to do it. But still with the bitching.
The biggest slip-up in DNS was how the.us domain was managed for many years. And it was a policy mistake, not a technical one.
Pretty much every other DNS server besides MS supports DNSSEC
Can you see someone with deeeeeeeep pockets doing what they can to keep DNSSEC from becoming popular?
But why would they? Funnily enough, leaving their customers wide open in this area isn't actually in MS's best interest, and they've for sure got the developers to be able to fix this. I'm a cynic, but I expect that we'll see updates in this area.
Revenue is not taxed. Profit is taxed. A company tries to maximize profit no matter what. The amount of tax on their profit has no impact on prices, except for stock prices.
Actually no. Profit is what is left after tax. Before it is taxed it is called "earnings" (or "ebitda", which is an acronym for "earnings before interest, tax, depreciation and amortization"). Companies try to maximize profit, as that gives the best ability to both reinvest in growing the business and returning money to shareholders as dividends. From the perspective of shareholders (i.e. the people who appoint the board) there's no point in increasing earnings if that increase is more than swallowed up by the tax man!
Here's an example... I can't go to the expressions window and edit an existing expression. I also can't copy/paste that expression somewhere else to edit it. I can give up and use gdb: print x and that's ok I guess, but it's not a persistent, updated value.
Relatedly, did you know that if you're running a terminal-based program inside the debugger inside Xcode, you can't cut-n-paste strings into it? That sucks, and yet it should be pretty easy for Apple to fix.
Slashdot Mirror
I understand that there are all kinds of open source projects out there, and some better than others. But based on my personal experience with some of the more prominent ones, I seriously believe a government run by open source types would be as terrible as what we have now. The following thoughts are based just on those projects.
[...]
The Chewbacca defense will actually work.
[...]
No it wouldn't, since open source types will remember that Chewie is able to pilot a starship and is hence able to live wherever he wants.
Internet radio is usually implemented as a TCP stream, because the order of arriving packets usually matters in a streaming radio setting.
That's true for podcasts, but for live stuff the savvy internet radio implementations will be using RTP over UDP (a dropped packet is irritating, but better to keep the stream going rather than stopping until the missing stuff is resent). When it's a big event, it can be better still to use multicast (assuming you can make it work; some routers still suck in this area IIRC).
This all applies much more so to live internet video streaming.
actually, it's just the gstreamer guys saying that, and only because they don't understand why KDE developers would want to use a Qt/KDE style API for their sound, when they could use a horrible_gnome_style_api_instead.
Nobody other than Qt/KDE developers ought to care what the Qt/KDE API for sound is, just as GTK/Gnome devs are the only folks who ought to care about that sound API. What sucks is that there are too many sound system APIs; entire separated sound stacks. There should be a common base API (at the "send this sort of message over that channel" level) so that the type of hardware and the codecs that you use are independent of the GUI system that you use.
But the author's complaint is pretty much BS -- they're assuming that this is a problem for commercial vendors.
Of course it's not a problem for commercial vendors. They can just not support Linux. For many, that's not even going to cost them a particularly large fraction of their customers.
Sometimes you really do want to run commercial software, e.g. because it does things that aren't available to the same quality in OSS. The classic example is Skype; yes, there are open alternatives, but if there's a choice of using the closed code and being able to communicate with others or using the open code and being stuck on your own, well, that's not much of a choice. Anyway, my point is let's not be too hostile to commercial code. (Of course, positively supporting open code is good; it's trying to stop closed code that is a problem.)
If copyright terms were pulled back to something sane, such as 10 or 15 years [...]
That'd be really good, as it would allow those of us who don't like the FSF's political positions to contribute to (old versions of) Emacs without compromising our principles or our ability to take our contributions closed-source once we've got people hooked!
(Tongue in cheek, but only partially. Whenever you hack with the legal system in a deep way, changes can be far reaching.)
Good UI is about making the app easily learnable.
That's only part of it. Different apps have different modes of use. Some apps shouldn't be learned because people don't use them for long periods: they should just be usable straight away. Others will be used daily for years by trained experts, and the learning phase will be comparatively short: the real value there is in making the users productive, and the UI can be deep and complex. Those definitely aren't the only possibilities either; there is a sophisticated spectrum, and since the reason for that is "people", that isn't changing.
If you want to appear less ignorant, try finding a copy of About Face by Alan Cooper. I read the first edition long ago, and it was great then (though dated). The third edition (which ought to be using much more relevant examples if nothing else) was published last year, and so should be in your local good tech library.
At one place I worked, the guy who wrote up the coding standard explicitly prohibited jokes in comments and humorous variable names.
Add a comment to the code against the most business-critical part of the code something like "Hmm, I hope SOX-compliance never looks at this." and sign it with his name. The beauty is it's both not at all funny and hilarious, depending on your point of view.
It's the same kind of security you run into in large corporations. You know, you have to use a randomly generated 18 character alphanumeric password and it changes every 90 days... which is great except that when you go to do your timesheets you have to enter your LAN password... which goes over the wire plaintext encapsulated in an HTTP POST query. Oops. Also, because not everybody's memory is so great, it becomes common practice to keep the 18 character passwords written on sticky notes.
Yikes! That's insecure! Better require that at least a third of the characters are symbols, that both upper and lower case are used. Increase the of changes frequency to every 10 seconds, and require that people never reuse any of their passwords or their account will be locked until they've satisfied the gestapo^Wcorporate security team that they're not trying to hack the system!
And don't forget to provide everyone with post-its...
(I do real security. I work with people who do real security. Password anal retentiveness never ever helps; it just pisses users off. Using properly set up SSL/SSH for everything really does help. Network traffic monitoring - of TCP and UDP headers, not content - really does help, especially with detection of what happened. Firewalls/NAT do help, if used carefully; reduction of the attack surface to things that are vetted and monitored for intrusion really is a very strong technique.)
On the other hand, it's pretty easy to guess that you couldn't fix a light in another building from your phone. And that a Citroen C4 doesn't transform into a dancing robot
Noooooooooo!
There's no such thing as -20 K
Not so, though it's a very strange thing indeed.
Just about everything to do with computers is simpler and more regular than just about anything related to boating.
You ought to try running a production HPC service before saying that...
What is the deal with Starbucks and their imitators burning their beans?
It's also important what type of drink you're making from it. You need a different type of roast to make a good filter coffee from a good espresso (e.g. Italian roasts tend to be great for espresso, but don't drink filter coffee in Rome as they really don't understand it; the Nordic countries do far better at filter coffee). You may need a different roast if you're adulterating the coffee with milk or sugar or even just hot water.
The other thing is whether the coffee is burnt during or after final making. If it's made too hot, it burns and tastes nasty, just as if it is kept too hot it's horrible too.
IME at one *specific* Starbucks, their filter coffee is great neat black at 07:20 in the morning, but nowhere near as good 30 minutes later. I make no claims for this result being extensible to any other part of the chain, and nor am I telling you which store; I don't want people getting in my way at all before I get my coffee...
At this point it seems obvious that you've assumed a really simplistic model of a neuron that can compute a synaptic value in a single floating point operation. These simple neuron models don't behave like a real brain, and scaling up simulations of them doesn't produce anything interesting. Real neurons are capable of computing much more complex functions than these models. The throughput on the interconnect is going to be a major factor, and simulating each neuron will require from 10s to 1000000s of operations depending on the level of biological realism that is required.
The real question from an AI perspective is whether all that detail is necessary. Do we need to simulate individual signaling molecules? Do we need to simulate individual synapses? Can we simulate things at a higher level than neurons and still get a functionally similar model?
Obviously, for some things we definitely can't abstract (such as the effect of certain kinds of small molecules on consciousness). But nobody really has any idea whether general AI needs that level of detail. My hunch is that it doesn't, just as you don't need to know how transistors really work to write a computer program or even to simulate a computer with a pretty high degree of accuracy. But that is just a hunch...
The Internet turns all the knobs to 11.
Except for the ones it turns to 0.
Suppose "what you want" is to name your project "The".
Reminds me when I was googling for an eatery called "This and That" a few years ago...
Bingo. :) Idealism is a luxury for the rich. We don't have the privilege of questioning our 'superiors' on their ethics.
Next up you'll be suggesting that we should touch our forelocks, drop our pants and grin happily when we take it in the ass from our "superiors". Well that's BS. Hold them to ethics at all times, and tell other people that you're doing this. There's never any reason to accept your bosses being total scum.
This sounds like a good idea on the surface -- it'll never happen, of course, because too many companies and individuals have too much invested in the .com, .net, etc. without the country codes... but still, I like the consistency it all brings.
If you've got e-mobsters smacking companies around a lot with DNS cracks because .com, etc. aren't signed, you will see migration to signed domain trees. And masses of bitching. (You'll always get that.) Or possibly people stopping the arguments over root domain signing and getting on with it by appointing someone to do it. But still with the bitching.
The biggest slip-up in DNS was how the .us domain was managed for many years. And it was a policy mistake, not a technical one.
Pretty much every other DNS server besides MS supports DNSSEC
Can you see someone with deeeeeeeep pockets doing what they can to keep DNSSEC from becoming popular?
But why would they? Funnily enough, leaving their customers wide open in this area isn't actually in MS's best interest, and they've for sure got the developers to be able to fix this. I'm a cynic, but I expect that we'll see updates in this area.
This inspired me to negate the regex to find the longest word I could type with my right hand.
Yeah, but they're thoroughly beaten by phyllophyllin, which at 13 letters is the longest right-side word in this machine's dictionary...
It's a noun that means a "monocarboxylic acid derived from chlorophyll".
It's easy to make an interface that works at both 800×600 and 1920×1080.
And yet so many developers continue to fail miserably at that. And fail doubly hard when they come up against users who require large fonts.
I must be getting a bit too cynical...
Revenue is not taxed. Profit is taxed. A company tries to maximize profit no matter what. The amount of tax on their profit has no impact on prices, except for stock prices.
Actually no. Profit is what is left after tax. Before it is taxed it is called "earnings" (or "ebitda", which is an acronym for "earnings before interest, tax, depreciation and amortization"). Companies try to maximize profit, as that gives the best ability to both reinvest in growing the business and returning money to shareholders as dividends. From the perspective of shareholders (i.e. the people who appoint the board) there's no point in increasing earnings if that increase is more than swallowed up by the tax man!
Imagine a beowulf cluster of those.
Meh. You put beowulf clusters inside grids, but you don't build beowulfs out of grids because the latency will totally slay you.
It is not "free" , stop believing the lies. Every time you sit through a commercial you have "paid".
No. You're just not the customer. You're the user (well, assuming you like country music) but it is the advertiser who is the customer.
Why on earth do you get a scholarship for your sporting ability?
Because colleges make a lot of money from their TV deals. Don't tell me you were thinking they were there to teach or carry out research, were you?
Here's an example... I can't go to the expressions window and edit an existing expression. I also can't copy/paste that expression somewhere else to edit it. I can give up and use gdb: print x and that's ok I guess, but it's not a persistent, updated value.
Relatedly, did you know that if you're running a terminal-based program inside the debugger inside Xcode, you can't cut-n-paste strings into it? That sucks, and yet it should be pretty easy for Apple to fix.