Of course it's not new - the Lisp machines were doing it decades ago, and I never implied that it was. I don't like to use the term tagged pointer in this regard, because that's a massively overloaded term and originated with the kind of tagged pointers that exist on the architecture that I work on.
However, with 64 bits, you have a lot more space to use for values and you have a lot more bits to use for tags. The top 16 bits are unused for address translation, and so can be used to store other information in any valid pointer. 64-bit JavaScript implementations often use this, because those bits are also used for NaN flags in a double, so you can use an invalid NaN pattern to indicate that the low 48 bits are a pointer. Apple's Objective-C implementation uses these bits in the isa pointer of objects to hold the reference count. This shrinks objects and makes reference counting easier. They also adopted a trick that I added to the GNUstep implementation a few years earlier: storing short strings in tagged pointers. It turns out that in desktop applications, strings that are shorter than 7 7-bits of ASCII characters account for 5-20% of all object allocations (path components, keys in dictionaries and so on) and you can conveniently fit these in a 64-bit pointer. When these are used as keys in dictionaries, you also get a cheaper search, as you can use pointer equality rather than invoking an equality method to check for equality.
They'll keep wasting billions on personal computers... and get no where. That ship has sailed, and the Apple ][ and Commodore 64 own the personal computer market. But M$ will toss endless amounts of money after it in a vain hope to gain traction, much like they keep pouring money into MS BASIC to no end.
Microsoft wasn't pushing the DOS-based PC, they were selling DOS to IBM, who used their dominance in the business computing market to push it and Microsoft rode their coat tails.
They'll keep wasting billions on GUI desktops... and get no where. That ship has sailed, and the Macintosh and IBM's OS/2 own the GUI desktop market. But M$ will toss endless amounts of money after it in a vain hope to gain traction, much like they keep pouring money into DOS to no end.
Windows 3 had a dominant market position before OS/2 shipped. The Mac had an advantage, but Microsoft could sell Windows as a DOS upgrade to a huge installed base, whereas switching to a Mac required buying a new machine. As long as DOS (specifically, DOS-running PC clones) were controlling the business computing market, they were able to leverage one monopoly to gain another.
They'll keep wasting billions on word processing... and get no where. That ship has sailed, and Wordperfect owns the word processing market. But M$ will toss endless amounts of money after it in a vain hope to gain traction, much like they keep pouring money into Windows to no end.
Word for DOS didn't get much traction and Microsoft changed APIs at the last minute to break the Wordperfect port to Windows, giving them a first-mover advantage. Once again, leveraging one monopoly to gain another.
They'll keep wasting billions on video game consoles... and get no where. That ship has sailed, and Nintendo and Sony own the video game console market. But M$ will toss endless amounts of money after it in a vain hope to gain traction, much like they keep pouring money into the Zune to no end.
With this one you have a point, except that the network effects for consoles are a lot less than with smartphones. People expect (or, at least, expected) that a new console would come with new games, so not being able to run legacy apps wasn't a problem. Even then, Microsoft was only able to do this because they used almost the same APIs as Windows and so made it easy to port Windows games (of which there were a lot) to the XBox. They were also able to throw a few billion at getting the market share. In contrast, people now expect to be able to run their existing Smartphone apps on their new phone, or at least to get the equivalent version. Not only are there not many Windows Phone apps, far fewer new ones are being written.
It looks like MS is giving up on ARM altogether and doubling down on x86. They probably see Intel weakening and figure they can get cheap chips out of them.
I take it you missed the announcement of the ARMv8 version of Windows 10 with an emulator that allows running x86 programs natively?
Office 365 is basically that: they bundle some cloud nonsense with it, but for most customers it's just a simple way of getting a license to the current version of Office at a flat per-user (not per-device) price.
The 12.9" iPad Pro is $899 with 128GB of flash and 4GB of RAM and the keyboard is another $169, so they're in the same ballpark if they intend this as a high-end iPad competitor, rather than a laptop competitor.
64-bit on iOS makes a big difference with Objective-C, because you can now embed the reference count in the high bits of the class pointer and can also embed a lot more small objects directly in object pointers. AArch64 is designed from scratch as a compiler target, whereas AArch32 has a lot of legacy that was intended for assembly programmers, some of which is hard to use from a compiler and some of which is really hard to implement efficiently on modern pipelines (say 'store multiple' or 'muldiple-wdith floating point registers' to an ARM microarchitect and watch their face sometime - it's entertaining).
No, I don't trust, and what company can anyone actually trust to do right by buyers nowadays? IMO, it's all a crap shoot, Android's just always seemed to me to be the "best bang for my buck"
There are different degrees of trust. Having worked with both the iOS and Android security teams and with colleagues tracking vulnerabilities in mobile platforms, I treat my Android phone as something that comes pre-compromised: I don't trust it with any sensitive data and it doesn't get any of my passwords and definitely won't be used for Internet banking. In contrast, I'm perfectly happy to use my iPad for all of these things.
In terms of lock-in, both Apple and Alphabet have an interest in keeping me on the platform, and if I buy anything from either of their app stores then I'll treat it as a disposable purchase (i.e. when I calculate the value to me, I'll assume it might stop working at any point). My phone has a bunch of open source things installed via F-Droid to avoid this.
In terms of data harvesting, Apple tries to avoid knowing too much about me because they don't want to deal with legal issues, whereas Alphabet makes 95+% of its revenue from advertising and so wants to harvest as much from me as possible. As a result, I'm happy to use the built-in web browser and mail client on my iPad, but installed Firefox and K9 on my phone (before I stopped using the phone for email).
Manufacturer profit doesn't necessarily translate to bad value for you. Apple has a much more limited lineup, which gives them better economies of scale than most other manufacturers. Their SoC design is in-house, which also helps with the margins (for almost anyone else except Samsung, the profit on the SoC is counted to a different company, and even for Samsung is to a different business unit and so doesn't count as profit on the handset). They have also been using that cash surplus to buy factories to make flash for their suppliers, in exchange for deep discounts for the first few years, so they are paying less for flash chips than their competitors. Most companies would find it difficult to make something equivalent to the iPhone for the retail price of the iPhone.
If you get used to a high-salt diet, i.e. from childhood, you'll have difficulty cuttiing down because things just taste bland without the amount of salt you're used to, and it can be difficult to switch to other flavours instead
It takes about a month to do this. Tastebuds have a very high turnover and are mostly replaced after a month. You become gradually more sensitive to salt over this time until the food that you used to eat becomes too salty (the same applies to sugar).
Are you sure it's $1/kW, not $1/W? The former would mean that it would pay for itself in about two days, the latter in a year or two. If the RoI is under a week, then I'd expect a lot more construction than exists currently.
The advantage of 1AU is that the available sunlight is enough that you'd be able to walk outside without shielding. The down side of making it closer to the sun is that you reduce the amount of available living space (although that's not likely to be a real concern for a long time after building one). A swarm wouldn't necessarily have this as a design goal.
Rooftop solar is about the least cost-effective way of running solar power. The panels are pretty cheap, but the cost of rooftop installation is high, the cost of the inverter is relatively high (and not amortised across a large number of panels) and the cost of the grid-feeding equipment is also relatively high. A field full of solar panels has a far lower per-panel cost, but the same per-panel power output. These typically have a 3-5 year RoI without subsidies.
The real problem with domestic solar power is the rate at which panels are improving (which is something the subsidies were intended to encourage). If I'm going to install something with a 20 year lifetime, I don't want to be able to get one that's 50% more efficient a year later for the same price. I started looking at roof-top solar about 10 years ago. Back then, the cheap panels were 8% efficient. Now they're 16-20% efficient for about the same price (and have longer warranties). Given how small a fraction of the total price of a small installation the panels are, it seems like a bad idea to buy them before they improvements slow down a bit (above 30% is likely to be really hard, so 25-30% efficient is probably the time to buy, as the difference between 25% and 30% is far less pronounced than the difference between 16% and 25%).
The problem with this argument is that it's circular. People don't get value from Facebook, they get value from the fact that other people are on Facebook and, conversely, by being on Facebook they give value to Facebook for others. If you close your account, then Facebook becomes a little bit less useful to all of your friends. If you and a quarter of your friends close their accounts, it becomes a lot less useful to the others who still have accounts, and as each of those closes their account it becomes less useful to the remainder.
The problem with enforcement is that it requires complete visibility into the company's accounts (including the off-shore shells) and then time spent tracking them. It's trivial to spin up a couple of thousand companies and move money between them until it's almost impossible to figure our where the money comes from. One of the recent sets of leaks showed a scheme in Europe that rearranges corporate structures within the EU every couple of weeks. It takes a regulator about a month to figure out exactly what the relationship between companies is and by that time it's changed. At any point, it looks to the regulators in one country as if tax is only owed in another country. You need a lot of international cooperation to address this.
OSX is a terrible interface, riddled with plenty of gotcha's that just frustrate the user eventually. Maybe its better now, but I used it back in 2008 or so.
Care to expand on that? A few counter examples:
The buttons in dialog boxes are all the correct way around. The 'proceed' item is on the right, the 'back' equivalent is on the left[1]. In contrast, Windows has them the wrong way around and Linux DEs have them inconsistently ordered (which is even worse: at least with Windows you eventually get used to the counterintuitive behaviour).
Dialog box buttons are all labelled with verbs. For example 'print' or 'download' not 'okay' for the forward action. If a app has unsaved data that will be lost on exit (rare now), the dialog on attempting to close will say 'cancel' (shortcut: escape), 'exit without saving', 'save' (default). Research has shown that most users don't read the dialog box text, just the button text.
Most apps now support sudden termination, so you can kill -9 an app and have it restart without losing any data (this is used with the OOM behaviour - apps that opt into this will be killed. The window server keeps a copy of their window content and when you switch back to them they're restarted. The user doesn't notice this, they just notice that they're swapping less. Oh, and it also helps defragment the heap). This means that accidentally exiting an app is unlikely to result in data loss.
The default is 'ask forgiveness, not permission'. Only dangerous actions prompt the user to confirm, the rest are all expected to support undo (and NSUndoManager is sufficiently easy to use that most third-party apps do as well). This includes settings changes and so on: they're applied immediately and support undo, rather than teaching the user to always hit 'okay' to every dialog that pops up will changing settings.
The menu bar is at the top of the screen, where it gets a Fitts' Law bonus for hitting it. In contrast, Windows not only puts it in the window but also hides it unless you hit alt, so you need to press a key then move the mouse to a small target.
All menu items (including submenus) are searchable. Type a word into the text field in the help menu and it will find help items[2] that match the term, but also menu items, so you can always find the relevant menu item even from a deeply nested menu in a few keystrokes.
All common actions in all applications have keyboard shortcuts and these are consistent.
Command-space brings up the Spotlight search box, which searches into items and also includes plugins for quickly viewing many kinds of document (e.g. searching for a person's name will show their address book entry in the pop-up, searching for an arithmetic expression will show the result and let you either open it in the calculator or copy the result). Most document-driven applications ship with Spotlight and QuickView plugins to support this.
[1] Actually, I think they're the wrong way around in right-to-left reading order locales, because older research indicated that this preference for rightwards meaning forwards was dependent on reading order. More recent research indicates that it's universal.
The webserver stuff is super-primitive, not even permitting local forwards to other servers
The 'web server stuff' is Apache with a little config GUI. There are many reasons to prefer nginx to Apache, but not being able to configure forwarding is a bit of an odd one.
But these are increasingly the same people. TV production companies are running their own online streaming services. Bands are providing digital downloads from their own web sites or via marketplaces that work for them. Authors are self-publishing e-books.
It doesn't work that way. DRM requires huge economies of scale (it's not worth developing a DRM scheme for a single album, you won't get enough devices to support it). Your choices are to either go through a distributor like Amazon or Apple who will use their own DRM model and gain the strong negotiating position, or license a DRM scheme from a company like Adobe or Microsoft and give them the same control.
Why? It's in Neftlix's interest for Netflix to have control over the channel, which is what Netflix DRM gives them. It's just not in the interests of content producers other than Netflix.
There was a study from the Harvard Business School almost 10 years ago now that looked at the optimum price point for music. They found that it was around 5 cents per track. At that price, people will buy an entire album without thinking and the sales will go up by more than enough to cover the reduction in per-sale profit. Something similar is probably true for most forms of digital entertainment. The distribution cost is small and impulse purchases can dwarf considered purchases if you get the price right. From personal experience, I've bought a number of games from GOG.com on sale that I've never played for about £1. I'll probably get around to them at some point, but £1 is cheap enough that 'that looks fun, I might find time to play it at some point' is enough of a motivation to make the purchase, whereas at £20 it has to be 'I want to spend hours playing that right now'.
Obviously you will get very little good content if DRM goes away and artists begin to basically give away their creations
Why is that obvious? There are two stages involved in consumers getting good content. Step one, someone has to create it. Step two, someone has to distribute it. The first step is difficult and (often) expensive. The second step is basically free with the Internet. If your economic model is to do the first step for free and then charge people for the second, then you're going to have problems.
This is not how content creators actually work, typically. They provide a sample (chapter of a book, pilot for a TV show, whatever) for free and then a content distributor (TV channel, publisher, and so on) pays them up front enough to create the full work, in exchange for the rights to try to make money from distribution. It's easy to imagine cutting out the middle man. Put the pilot for a TV show online for free (pilots are fairly cheap to produce, because they typically don't have the special effects done by the time that they're made available to networks) and then ask people to fund the whole thing. When it's finished, make it available for free and ask for funding for the next season or sequel - the fact that it's freely redistributable makes it easy for fans to share copies with other people who might want to pay for the next project (whether it's a direct sequel or something else from the same creator).
There's only one thing that will kill DRM: when content producers realise how much power it gives to content distributors. DRM on music is completely gone now. Why? Because the big four record labels realised that requiring DRM was giving Apple a much stronger negotiating position than them (want your music to work on iPods? You had to agree to Apple's terms or provide your music DRM free). With TV movies, we're increasingly seeing Netflix and Amazon get a similarly strong position. Netflix maintains streams for around 80 different types of device, including a load of set-top boxes that don't have upgradable firmware. Want to reach those customers? License your content to Netflix or allow it to be distributed without DRM (pretty much anything can play back plain H.264).
I quite enjoy the fact that the organisations insisting on DRM are the ones most harmed by it.
I did, and I found it quite difficult to believe that the authors had read any of the originals, let alone any notes. They completely missed all of the subtlety from the originals and made all of the characters painfully two dimensional. Reading the bit in the foreword when Brian Herbert opines that Kevin J Anderson (who has yet to write a single book with an ending that didn't feel like he got bored and had 5 pages to tie up all of the loose ends and is best known for some embarrassingly bad Star Wars novels) was the only person who could write something on a scale of the Dune sequels tells you that it's not going to go particularly well.
Slashdot Mirror
However, with 64 bits, you have a lot more space to use for values and you have a lot more bits to use for tags. The top 16 bits are unused for address translation, and so can be used to store other information in any valid pointer. 64-bit JavaScript implementations often use this, because those bits are also used for NaN flags in a double, so you can use an invalid NaN pattern to indicate that the low 48 bits are a pointer. Apple's Objective-C implementation uses these bits in the isa pointer of objects to hold the reference count. This shrinks objects and makes reference counting easier. They also adopted a trick that I added to the GNUstep implementation a few years earlier: storing short strings in tagged pointers. It turns out that in desktop applications, strings that are shorter than 7 7-bits of ASCII characters account for 5-20% of all object allocations (path components, keys in dictionaries and so on) and you can conveniently fit these in a 64-bit pointer. When these are used as keys in dictionaries, you also get a cheaper search, as you can use pointer equality rather than invoking an equality method to check for equality.
They'll keep wasting billions on personal computers... and get no where. That ship has sailed, and the Apple ][ and Commodore 64 own the personal computer market. But M$ will toss endless amounts of money after it in a vain hope to gain traction, much like they keep pouring money into MS BASIC to no end.
Microsoft wasn't pushing the DOS-based PC, they were selling DOS to IBM, who used their dominance in the business computing market to push it and Microsoft rode their coat tails.
They'll keep wasting billions on GUI desktops... and get no where. That ship has sailed, and the Macintosh and IBM's OS/2 own the GUI desktop market. But M$ will toss endless amounts of money after it in a vain hope to gain traction, much like they keep pouring money into DOS to no end.
Windows 3 had a dominant market position before OS/2 shipped. The Mac had an advantage, but Microsoft could sell Windows as a DOS upgrade to a huge installed base, whereas switching to a Mac required buying a new machine. As long as DOS (specifically, DOS-running PC clones) were controlling the business computing market, they were able to leverage one monopoly to gain another.
They'll keep wasting billions on word processing... and get no where. That ship has sailed, and Wordperfect owns the word processing market. But M$ will toss endless amounts of money after it in a vain hope to gain traction, much like they keep pouring money into Windows to no end.
Word for DOS didn't get much traction and Microsoft changed APIs at the last minute to break the Wordperfect port to Windows, giving them a first-mover advantage. Once again, leveraging one monopoly to gain another.
They'll keep wasting billions on video game consoles... and get no where. That ship has sailed, and Nintendo and Sony own the video game console market. But M$ will toss endless amounts of money after it in a vain hope to gain traction, much like they keep pouring money into the Zune to no end.
With this one you have a point, except that the network effects for consoles are a lot less than with smartphones. People expect (or, at least, expected) that a new console would come with new games, so not being able to run legacy apps wasn't a problem. Even then, Microsoft was only able to do this because they used almost the same APIs as Windows and so made it easy to port Windows games (of which there were a lot) to the XBox. They were also able to throw a few billion at getting the market share. In contrast, people now expect to be able to run their existing Smartphone apps on their new phone, or at least to get the equivalent version. Not only are there not many Windows Phone apps, far fewer new ones are being written.
It looks like MS is giving up on ARM altogether and doubling down on x86. They probably see Intel weakening and figure they can get cheap chips out of them.
I take it you missed the announcement of the ARMv8 version of Windows 10 with an emulator that allows running x86 programs natively?
You don't want more than 4GB of RAM for each, but you absolutely do want more than 4GB of virtual address space if you want ASLR to do any good.
Office 365 is basically that: they bundle some cloud nonsense with it, but for most customers it's just a simple way of getting a license to the current version of Office at a flat per-user (not per-device) price.
Have you ever seen one in public?
Yes, several, and not all of them owned by Microsoft employees.
The 12.9" iPad Pro is $899 with 128GB of flash and 4GB of RAM and the keyboard is another $169, so they're in the same ballpark if they intend this as a high-end iPad competitor, rather than a laptop competitor.
64-bit on iOS makes a big difference with Objective-C, because you can now embed the reference count in the high bits of the class pointer and can also embed a lot more small objects directly in object pointers. AArch64 is designed from scratch as a compiler target, whereas AArch32 has a lot of legacy that was intended for assembly programmers, some of which is hard to use from a compiler and some of which is really hard to implement efficiently on modern pipelines (say 'store multiple' or 'muldiple-wdith floating point registers' to an ARM microarchitect and watch their face sometime - it's entertaining).
No, I don't trust, and what company can anyone actually trust to do right by buyers nowadays? IMO, it's all a crap shoot, Android's just always seemed to me to be the "best bang for my buck"
There are different degrees of trust. Having worked with both the iOS and Android security teams and with colleagues tracking vulnerabilities in mobile platforms, I treat my Android phone as something that comes pre-compromised: I don't trust it with any sensitive data and it doesn't get any of my passwords and definitely won't be used for Internet banking. In contrast, I'm perfectly happy to use my iPad for all of these things.
In terms of lock-in, both Apple and Alphabet have an interest in keeping me on the platform, and if I buy anything from either of their app stores then I'll treat it as a disposable purchase (i.e. when I calculate the value to me, I'll assume it might stop working at any point). My phone has a bunch of open source things installed via F-Droid to avoid this.
In terms of data harvesting, Apple tries to avoid knowing too much about me because they don't want to deal with legal issues, whereas Alphabet makes 95+% of its revenue from advertising and so wants to harvest as much from me as possible. As a result, I'm happy to use the built-in web browser and mail client on my iPad, but installed Firefox and K9 on my phone (before I stopped using the phone for email).
Manufacturer profit doesn't necessarily translate to bad value for you. Apple has a much more limited lineup, which gives them better economies of scale than most other manufacturers. Their SoC design is in-house, which also helps with the margins (for almost anyone else except Samsung, the profit on the SoC is counted to a different company, and even for Samsung is to a different business unit and so doesn't count as profit on the handset). They have also been using that cash surplus to buy factories to make flash for their suppliers, in exchange for deep discounts for the first few years, so they are paying less for flash chips than their competitors. Most companies would find it difficult to make something equivalent to the iPhone for the retail price of the iPhone.
If you get used to a high-salt diet, i.e. from childhood, you'll have difficulty cuttiing down because things just taste bland without the amount of salt you're used to, and it can be difficult to switch to other flavours instead
It takes about a month to do this. Tastebuds have a very high turnover and are mostly replaced after a month. You become gradually more sensitive to salt over this time until the food that you used to eat becomes too salty (the same applies to sugar).
I've not used it, but the guys doing the z/System port of FreeBSD seem happy with it.
Are you sure it's $1/kW, not $1/W? The former would mean that it would pay for itself in about two days, the latter in a year or two. If the RoI is under a week, then I'd expect a lot more construction than exists currently.
The advantage of 1AU is that the available sunlight is enough that you'd be able to walk outside without shielding. The down side of making it closer to the sun is that you reduce the amount of available living space (although that's not likely to be a real concern for a long time after building one). A swarm wouldn't necessarily have this as a design goal.
Rooftop solar is about the least cost-effective way of running solar power. The panels are pretty cheap, but the cost of rooftop installation is high, the cost of the inverter is relatively high (and not amortised across a large number of panels) and the cost of the grid-feeding equipment is also relatively high. A field full of solar panels has a far lower per-panel cost, but the same per-panel power output. These typically have a 3-5 year RoI without subsidies.
The real problem with domestic solar power is the rate at which panels are improving (which is something the subsidies were intended to encourage). If I'm going to install something with a 20 year lifetime, I don't want to be able to get one that's 50% more efficient a year later for the same price. I started looking at roof-top solar about 10 years ago. Back then, the cheap panels were 8% efficient. Now they're 16-20% efficient for about the same price (and have longer warranties). Given how small a fraction of the total price of a small installation the panels are, it seems like a bad idea to buy them before they improvements slow down a bit (above 30% is likely to be really hard, so 25-30% efficient is probably the time to buy, as the difference between 25% and 30% is far less pronounced than the difference between 16% and 25%).
The problem with this argument is that it's circular. People don't get value from Facebook, they get value from the fact that other people are on Facebook and, conversely, by being on Facebook they give value to Facebook for others. If you close your account, then Facebook becomes a little bit less useful to all of your friends. If you and a quarter of your friends close their accounts, it becomes a lot less useful to the others who still have accounts, and as each of those closes their account it becomes less useful to the remainder.
The problem with enforcement is that it requires complete visibility into the company's accounts (including the off-shore shells) and then time spent tracking them. It's trivial to spin up a couple of thousand companies and move money between them until it's almost impossible to figure our where the money comes from. One of the recent sets of leaks showed a scheme in Europe that rearranges corporate structures within the EU every couple of weeks. It takes a regulator about a month to figure out exactly what the relationship between companies is and by that time it's changed. At any point, it looks to the regulators in one country as if tax is only owed in another country. You need a lot of international cooperation to address this.
OSX is a terrible interface, riddled with plenty of gotcha's that just frustrate the user eventually. Maybe its better now, but I used it back in 2008 or so.
Care to expand on that? A few counter examples:
The buttons in dialog boxes are all the correct way around. The 'proceed' item is on the right, the 'back' equivalent is on the left[1]. In contrast, Windows has them the wrong way around and Linux DEs have them inconsistently ordered (which is even worse: at least with Windows you eventually get used to the counterintuitive behaviour).
Dialog box buttons are all labelled with verbs. For example 'print' or 'download' not 'okay' for the forward action. If a app has unsaved data that will be lost on exit (rare now), the dialog on attempting to close will say 'cancel' (shortcut: escape), 'exit without saving', 'save' (default). Research has shown that most users don't read the dialog box text, just the button text.
Most apps now support sudden termination, so you can kill -9 an app and have it restart without losing any data (this is used with the OOM behaviour - apps that opt into this will be killed. The window server keeps a copy of their window content and when you switch back to them they're restarted. The user doesn't notice this, they just notice that they're swapping less. Oh, and it also helps defragment the heap). This means that accidentally exiting an app is unlikely to result in data loss.
The default is 'ask forgiveness, not permission'. Only dangerous actions prompt the user to confirm, the rest are all expected to support undo (and NSUndoManager is sufficiently easy to use that most third-party apps do as well). This includes settings changes and so on: they're applied immediately and support undo, rather than teaching the user to always hit 'okay' to every dialog that pops up will changing settings.
The menu bar is at the top of the screen, where it gets a Fitts' Law bonus for hitting it. In contrast, Windows not only puts it in the window but also hides it unless you hit alt, so you need to press a key then move the mouse to a small target.
All menu items (including submenus) are searchable. Type a word into the text field in the help menu and it will find help items[2] that match the term, but also menu items, so you can always find the relevant menu item even from a deeply nested menu in a few keystrokes.
All common actions in all applications have keyboard shortcuts and these are consistent.
Command-space brings up the Spotlight search box, which searches into items and also includes plugins for quickly viewing many kinds of document (e.g. searching for a person's name will show their address book entry in the pop-up, searching for an arithmetic expression will show the result and let you either open it in the calculator or copy the result). Most document-driven applications ship with Spotlight and QuickView plugins to support this.
[1] Actually, I think they're the wrong way around in right-to-left reading order locales, because older research indicated that this preference for rightwards meaning forwards was dependent on reading order. More recent research indicates that it's universal.
[2] And man pages in the Terminal app.
The webserver stuff is super-primitive, not even permitting local forwards to other servers
The 'web server stuff' is Apache with a little config GUI. There are many reasons to prefer nginx to Apache, but not being able to configure forwarding is a bit of an odd one.
But these are increasingly the same people. TV production companies are running their own online streaming services. Bands are providing digital downloads from their own web sites or via marketplaces that work for them. Authors are self-publishing e-books.
It doesn't work that way. DRM requires huge economies of scale (it's not worth developing a DRM scheme for a single album, you won't get enough devices to support it). Your choices are to either go through a distributor like Amazon or Apple who will use their own DRM model and gain the strong negotiating position, or license a DRM scheme from a company like Adobe or Microsoft and give them the same control.
Why? It's in Neftlix's interest for Netflix to have control over the channel, which is what Netflix DRM gives them. It's just not in the interests of content producers other than Netflix.
There was a study from the Harvard Business School almost 10 years ago now that looked at the optimum price point for music. They found that it was around 5 cents per track. At that price, people will buy an entire album without thinking and the sales will go up by more than enough to cover the reduction in per-sale profit. Something similar is probably true for most forms of digital entertainment. The distribution cost is small and impulse purchases can dwarf considered purchases if you get the price right. From personal experience, I've bought a number of games from GOG.com on sale that I've never played for about £1. I'll probably get around to them at some point, but £1 is cheap enough that 'that looks fun, I might find time to play it at some point' is enough of a motivation to make the purchase, whereas at £20 it has to be 'I want to spend hours playing that right now'.
Obviously you will get very little good content if DRM goes away and artists begin to basically give away their creations
Why is that obvious? There are two stages involved in consumers getting good content. Step one, someone has to create it. Step two, someone has to distribute it. The first step is difficult and (often) expensive. The second step is basically free with the Internet. If your economic model is to do the first step for free and then charge people for the second, then you're going to have problems.
This is not how content creators actually work, typically. They provide a sample (chapter of a book, pilot for a TV show, whatever) for free and then a content distributor (TV channel, publisher, and so on) pays them up front enough to create the full work, in exchange for the rights to try to make money from distribution. It's easy to imagine cutting out the middle man. Put the pilot for a TV show online for free (pilots are fairly cheap to produce, because they typically don't have the special effects done by the time that they're made available to networks) and then ask people to fund the whole thing. When it's finished, make it available for free and ask for funding for the next season or sequel - the fact that it's freely redistributable makes it easy for fans to share copies with other people who might want to pay for the next project (whether it's a direct sequel or something else from the same creator).
There's only one thing that will kill DRM: when content producers realise how much power it gives to content distributors. DRM on music is completely gone now. Why? Because the big four record labels realised that requiring DRM was giving Apple a much stronger negotiating position than them (want your music to work on iPods? You had to agree to Apple's terms or provide your music DRM free). With TV movies, we're increasingly seeing Netflix and Amazon get a similarly strong position. Netflix maintains streams for around 80 different types of device, including a load of set-top boxes that don't have upgradable firmware. Want to reach those customers? License your content to Netflix or allow it to be distributed without DRM (pretty much anything can play back plain H.264).
I quite enjoy the fact that the organisations insisting on DRM are the ones most harmed by it.
I did, and I found it quite difficult to believe that the authors had read any of the originals, let alone any notes. They completely missed all of the subtlety from the originals and made all of the characters painfully two dimensional. Reading the bit in the foreword when Brian Herbert opines that Kevin J Anderson (who has yet to write a single book with an ending that didn't feel like he got bored and had 5 pages to tie up all of the loose ends and is best known for some embarrassingly bad Star Wars novels) was the only person who could write something on a scale of the Dune sequels tells you that it's not going to go particularly well.