Yes, they do. If you run a business, the repair policies are insane. Dell will have parts or visit onsite within 24 hours. Apple make you go do a store (in a different city, we didn't have one), or send it off through an authorised repair company, and it will take two weeks to turn it around. I worked in a university in a group which was primarily Mac-based. A couple of years ago, the funding bodies who fund academic research banned the purchase of Mac hardware with their funds, the justification being that it wasn't cost effective and it was wasting valuable money which could be spent better on alternatives. And to be honest, they aren't wrong, are they.
If I have a laptop like a Dell, HP or Thinkpad, I can replace the keyboard using a vendor-supplied replacement in just a few minutes. Apple now rivet the keyboard to the case, requiring both to be replaced. It's not even screwed on. That's unreasonable. If I need to replace the screen, I can get replacement screen parts. But Apple require the whole lid assembly to be replaced in its entirely because it's all glued together. That's equally unreasonable and wasteful of materials. That's just two points. But the entire systems made by Apple are like that. It all adds up to expensive hardware which is not repairable or manageable at the scale of a medium or large organisation. We used to have one full-time staff member who dealt solely with Mac imaging and facilitating hardware repairs.
If you're wanting to use a computer as a business system, rather than a personal toy, you need better than what Apple can offer. They have never really made much effort to cater for this type of use in their entire history. This stuff ends up costing companies a lot of money they could spend more productively on other things.
That's a typical MBA attitude, and it's devastating to the long-term prospects of a company. It doesn't matter that they are the worst selling. They are needed, because they fill a niche. If you continually axed the "worst selling" product, you'd drop everything except the iPhone... Oh, wait...
The Mac Pro should not take much R&D time. It's a box with a PC mainboard in it. A slightly nicer box than other midi towers, but it's still just a box. They could rebadge a Dell Precision and I'd buy it. The Mac Mini could be a standard mini-ITX or equivalent.
The problem here is that Apple wants to over-engineer these systems to use highly custom board designs and cases to make these as small as possible. But for the niche they occupy, the end user is unlikely to care about that. That's the strategic mistake. The Pro should be powerful and expandable, but it's neither. It's dated and restricted. The mini is smaller, but there's no need to make it so small it can't be upgraded by an end user. A little bigger, and it could have M2 interfaces and maybe a couple of 3.5" bays internally. But you have to dismantle half the internals just to access the RAM slot, and the SSD is soldered. What a pain.
I want a new Mac mini (or Pro) for my consulting development work. But the capabilities and price of current hardware makes it pointless. Even if I invest in one, who would want to run my code on such anaemic and overpriced hardware?
They need to remember that while the phone and iMac are to a large extent fashion products, the high-end PC depends primarily upon functionality and price, and they've missed the mark for years on that front.
What's also bad is that if you alone took the time to prepare for it, everyone else there is still clueless and unable to meaningfully participate! Some people even treat you badly because you're the only one with a clue! Sometimes I've emailed a summary of all my research/investigatory findings to the participants a day or so beforehand so they could easily get up to speed and been told that I wrote "a wall of text" and they couldn't read it (2 pages)! You just can't help some people. Meetings really are lowest common denominator affairs.
With the new apple keyboards, they are failing all the time. It really loses its value proposition when you might be without it for days while they replace half of the case.
The answer two posts up was implying that Linux ACLs were comparable with Windows ACLs. I explained that they are rather different than Windows ACLs, being both more limited in their functionality and being non-standard. I also explained that there was a third ACL type, NFSv4 ACLs which were a superset of both and available on several other Unix systems and which interoperate seamlessly with Windows systems as a result. The point being, that Linux here is the odd one out. It lacks a rich ACL model in its VFS layer, and consequently Linux can't support NFSv4 ACLS and is poorly interoperable with other systems using rich ACLs, as well as local filesystems using rich ACLs. I understand this is because the VFS maintainers don't like them, but this is a fairly significant functionality gap at this point.
Absolutely agreed, it must be available and convenient to work. I'll bet that we'll see the existing service stations replaced by places you can relax for a bit while you wait for a charge half way. Coffee shops, cafes, restaurants, tourist attractions etc. with chargers in all the parking spaces. I'd be happy to take a half hour break four hours in. But even before we get to this point, the vast majority of trips are much shorter than this. The people I work with who have them (mainly Nissan Leafs) are all really happy with them, and either have a second family car or rent one for long distance trips. They use the electric for all their trips around town, school, work, shops, sports etc. and it already serves that need very well. They are now coming onto the second hand market in larger numbers, and I have no doubt they will affect the market economics of fuel powered vehicles over the next few years.
In the short term, yes. But only the short term. Once there's less demand, that's going to result in knock-on effects in the medium term. Individual service stations aren't going to be economical, so there will be closures and consolidation (this already happened a decade or so back in the UK, due to the high prices). That means less deliveries, less road tankers needed, less storage capacity needed, and less demand for refineries, which will have to lower production. As demand reduces, the whole complex refining and distribution network will become less economical to operate. Once refineries themselves start to become uneconomical to operate, consolidating and closing, that will start to turn liquid fuels from a high-volume low-price commodity into a low-volume high-price specialty product because you'll have to get custom delivery from one of the few remaining refineries. The whole distribution network took decades to put in place, and it's costly to maintain. It will fall apart like a deck of cards when the foundation is knocked out. Without that distribution network, hauling your delivery is going to be expensive. A hundred years ago, hay production and distribution was of key importance to keeping cities and farms running, and was planned and monitored (such was the entrenched bureaucracy, original car registrations in London had a "hay allowance" IIRC). Today, hay is a niche market. The same will eventually come true for fossil fuels, and electric adoption will only hasten its demise. The transition might become sooner and more swiftly than we might think, once the tipping point is reached.
In the US anyway. It's already very expensive in much of the rest of the world. And that's possibly not going to last. As electric takes over, it's going to reduce demand for liquid fuel, and that will have knock-on effects on refineries and the vast distribution infrastructure which delivers it to all the local stations. Once the negative feedback kicks in, and local stations close, the distribution network contracts and refineries are cut back, it's going to get increasingly expensive. And that will further drive electric adoption and be self-reinforcing. Probably not for a good while yet, but if electric takes off seriously it's just a matter of time. Plus, if the rest of the world switches over to electric before the US, there will be pressure due to supply chain economics as well.
We don't need a new version of the IDE every 2 years. But the compiler and associated tools bring much needed bugfixes and stuff like C++17 and soon C++20. Shame it's all bundled so tightly together.
100% is likely an unattainable goal. But Wine has been painstakingly reverse engineered. They had to identify all the compatibility workarounds and play a never ending game of catch-up. If Microsoft were to make their own equivalent, they would be able to have much better coverage, with the original sources.
Linux has "POSIX.1e DRAFT" ACLs. They are functional, but limited, and based upon an unratified and abandoned draft standard.
Other Unix sytems, like Solaris, IllumOS and FreeBSD, implemented NFSv4 ACLs which are both a ratified published standard and are compatible with both POSIX.1e DRAFT and Windows ACLs. (They are a superset of both.)
If you're using NFSv4 ACLs they are modifiable and queryable from the command-line with get/setfacl and they are also modifiable and queryable from the Windows security/permissions property pages in the explorer (if exported via Samba or NFSv4).
The whole ACL situation is limited by the fact that Linux hasn't implemented NFSv4 ACLs in the VFS. Yet filesystems like ZFS and NFSv4 use them, but they are hidden and inaccessible on Linux. If Linux implemented them, we would have pretty comprehensive and interoperable ACL support between all the major platforms.
Protein structure is intrinsically a computational chemistry / physics modelling problem. It requires modelling of solvent and solute interactions with the protein structure, van der Waals, dipole, ionic and other electrostatic interactions, hydrophobic/hydrophilic interactions with the solvent and itself (and lipid bilayer for membrane proteins), minimising free energy of the whole structure, of which there may be multiple stable variants under different conditions, and interactions between different subunits as well as multimeric associations. It's hard, and it's computationally expensive. I used to work with a team of such modellers, all hardcore physics and maths people, in the computational biology department I used to work in.
"AI" might be able to recognise certain patterns. And it might be able to predict certain structural motifs with a reasonable degree of accuracy. But it will always be limited by the training dataset as the parent tried to explain. It's not modelling, it's guessing by interpolation. There's no intelligence; it can't extrapolate to make predictions which it hasn't been trained specifically for. And there's no guarantee that the structures it predicts will be stable or valid in any way. "AI" of this sort isn't magic, and it's certainly not intelligent. It's questionable that it's even a valid technique for good science. Because science and data analysis should be understandable, not a black box.
If the rest of Amazon is loss making, maybe it's the best thing to do in the medium term. You can't subsidise loss-making subsidiaries forever. There are plenty of other companies out there to pick up the slack. I for one have abandoned Amazon retail almost entirely; the marketplace damaged the trust I have in being able to purchase legitimate merchandise, and not some co-mingled knockoff. It's also not at all price competitive with other companies; products are often cheaper elsewhere, and I have zero reason for loyalty to them.
Yeah, I know there's more. But I can't believe there's that much more; not an order of magnitude more. I've worked as a graduate student and postdoc. The entire research institute I was in got by with a single secretary and a lab manager. Centralised university functions included payroll, copyright and patents, and wider departmental functions included purchasing and shared facilities. None of these individually employed that many people. I've personally had discussions regarding copyright and patents and commercialisation of research with the tech transfer people, and these functions were still pretty small. Just a couple of people for the entire university (UK). Conferences require a few staff to organise and take care of the logistics, for sure. But when we organised meetings, the organising researchers and departmental secretary handled it all for the most part. I saw them get seriously stressed about it, because it was all done on top of their regular commitments! Combined, I just don't see four large buildings full of people as justifiable. Large corporations can manage with a handful of admin staff, and I can't see why universities can't either, for the most part.
A few times recently I've had the IPv4 networking randomly break, but all the IPv6 services and websites worked without interruption. The autoconfiguration is worth something. Better than NetworkMangler which is the cause of the IPv4 outages, no doubt.
Most of that can be ignored though. Tunelling is dead; ignore it. It's 0.00% for the last couple of years, 0.01 or less for the last 6 years. 26% native today. So not important to learn-just go native. Link-local can be ignored for the most part; avahi/zeroconf and the like make it transparent. Subnets are also ignorable since it's part of the first 64 bits after the routing prefix; there is nothing to configure. For regular setup and use, all of this should be transparent and ignorable for the common case scenarios. On my network, it's all automatic with SLAAC, connect and it works.
It's happening. Look at this graph. Growth was exponential from 2010 to 2017, taking it from 0.1% to 16%. The last two years have been mostly linear, from 16% to 26% (~5%/year). The last 8 weeks alone have seen a 1.5% increase; that's equivalent to all the growth from 2010 to July 2013! 3.5 years compared to 8 weeks for the same improvement. We're well into the implementation phase now, with over 1 in 4 users using Google services over IPv6; the actual number is even higher, because Google underestimates it by requiring IPv6 to be explicitly whitelisted by them. It is taking time, no doubt about that, but it is happening at a decent clip now, and the pressure to provide it will increase. Already most of my internet traffic is over v6, and it's also more reliable. We're not far off the tipping point where it will start to be required.
It's a direct copy of the fundamental Unix system call and library APIs, with some extra stuff on top. It's a reimplementation of existing system design. The points about SCO are irrelevant; that was about the values of some constants in a header. Which by the way, were copied directly for compatibility. The SCO case was about whether that merited copyright protection, not about whether it was copied (it was, and that point wasn't in question).
You're confusing copying an implementation with copying a design. The Linux kernel, GNU utilities, compiler and the rest are all reimplementations of existing ones. With a number of improvements and extensions on top, of course. The SCO ranting has rather missed the point I was making.
The explosion of administration staff is both surprising and absurd. For many years, I wondered what they all did! Lots of make-work, by the look of it. Lots of meetings, training, reading and sending emails to each other. But how much actual productive work? Precious little, IMO. How much "administration" do students and staff actually need to run a university?
In the '90s, my department was run by three people, two undergraduate admins, one graduate. They handled admissions, collecting coursework, and all the other stuff. So three people for about 500 students. And some departmental secretaries/lab managers. It didn't seem too unreasonable. But a decade later, they increased their headcount by one or two as the department grew, while the central university administration had expanded from one building into four separate office blocks! W T F is all that for?! It's a bit of a mystery. I suspect that when they take ~£10k per student per year, they are so awash with cash (hundreds of millions, plus gouging overseas postgrad students) that there is simply no restraint upon their spending or expansion. They are charging such obscene amounts, which doesn't get spent on education for the most part; the amount going to lecturers, lab space and other teaching resources is a fraction of that, so where does it go? I can only guess onto admin, with liberal budgets and salaries to match. But where is the oversight of this, with someone to question the necessity of it all?
I find the whole thing rather unsavoury and obscene, with vice chancellors on £500k and up, while students are fleeced and saddled with a lifetime of debt. Value for money, it isn't. While I have been fortunate to do an undergrad, postgrad and doctorate, it's most likely that I would not go today; it's not affordable given the cost, and hard to justify for the benefits it provides.
The scheme is simple and takes just a few minutes to familiarise yourself with. That's all it is, familiarity. By the way, you only need two colons "::", which means "pad blocks with zeros". The rest are redundant. For example my link-local address is currently fe80::e2d5:5eff:fea8:50c9; my global address is something like 2001:8b0:860:ccbe:243b:81de:43b2:fb37. So it's 8 blocks of 4 hex digits, separated by colons, with optional eliding of ":0000:" with "::". That's it. Your nine year old should be able to understand it just as well as IPv4. He won't even need to learn about all the different IPv4 network classes.
I know you meant it as a joke, but language can change exactly like this. When I was at school, a popular insult was to call people a "spaz" (spastic). Because of its use as a generic insult, some do-gooders required everyone to refer to disabled people as "special". Result: "special" becomes a generic insult. Like: "You're a bit special, aren't you" is equivalent to "you're a spaz" or "you're a retard". More recently, the charity the Spastic Society renamed themselves to "Scope". Result: spastic children are now insultingly referred to as "scopers". So the word "scope" is now also an insult in its own right.
The point being, no matter how much do-gooders and SJWs clamp down on acceptable language use, the remaining "permitted" words will be adapted to replace them, turning the "new inclusive terminology" on its own head. This isn't a battle they can win, because human nature, at its nastiest best, will always be creative enough to come up with new "bad" language! And while I don't condone bullying or insulting, it does make me happy to see people sticking their fingers up at the SJW-enforced norms.
When you look at what made Linux successful, little of it was originally to do with being revolutionary. The vast majority of it was re-implementing existing software under a free licence. The Linux kernel is a copy of Unix kernels. The GNU utilities are copies of Unix utilities. Likewise the compiler, desktop environments, and most of everything else. Being open and free made it more useful and compelling than the proprietary equivalents. Being a direct copy and following the existing standards made it easy to migrate to and use with little disruption. Having a good number of enhancements and improvements on top was the icing on the cake, but the core stuff was what made it indispensable.
It's the "revolutionary" parts which have caused the most disruption, inconvenience and upset. They are also the parts which are the most poorly designed and implemented, and it's not a coincidence. A good number of these people are now arguing that POSIX and other standards are no longer relevant, but they are completely ignoring the main historical reasons why we have popular open source systems. Projects like Gnome, systemd have made some terrible design choices, and have also repeatedly broken compatibility with themselves over the years. Were they designed and implemented by competent professionals who could design and engineer systems to the standard of what came before them, that friction would not exist.
Agreed to a point. But they could have gone straight to chip+pin rather than the chip+signature setup which is almost pointless. When the rest of the whole world nearly is using chip+pin for nearly two decades now, it seems a bit odd to not use it. And regarding the magstripe fallback, has a date been set to drop it yet? If it was withdrawn from use and on new cards starting 2020, that would significantly curtain fraud.
Slashdot Mirror
Yes, they do. If you run a business, the repair policies are insane. Dell will have parts or visit onsite within 24 hours. Apple make you go do a store (in a different city, we didn't have one), or send it off through an authorised repair company, and it will take two weeks to turn it around. I worked in a university in a group which was primarily Mac-based. A couple of years ago, the funding bodies who fund academic research banned the purchase of Mac hardware with their funds, the justification being that it wasn't cost effective and it was wasting valuable money which could be spent better on alternatives. And to be honest, they aren't wrong, are they. If I have a laptop like a Dell, HP or Thinkpad, I can replace the keyboard using a vendor-supplied replacement in just a few minutes. Apple now rivet the keyboard to the case, requiring both to be replaced. It's not even screwed on. That's unreasonable. If I need to replace the screen, I can get replacement screen parts. But Apple require the whole lid assembly to be replaced in its entirely because it's all glued together. That's equally unreasonable and wasteful of materials. That's just two points. But the entire systems made by Apple are like that. It all adds up to expensive hardware which is not repairable or manageable at the scale of a medium or large organisation. We used to have one full-time staff member who dealt solely with Mac imaging and facilitating hardware repairs. If you're wanting to use a computer as a business system, rather than a personal toy, you need better than what Apple can offer. They have never really made much effort to cater for this type of use in their entire history. This stuff ends up costing companies a lot of money they could spend more productively on other things.
That's a typical MBA attitude, and it's devastating to the long-term prospects of a company. It doesn't matter that they are the worst selling. They are needed, because they fill a niche. If you continually axed the "worst selling" product, you'd drop everything except the iPhone... Oh, wait... The Mac Pro should not take much R&D time. It's a box with a PC mainboard in it. A slightly nicer box than other midi towers, but it's still just a box. They could rebadge a Dell Precision and I'd buy it. The Mac Mini could be a standard mini-ITX or equivalent. The problem here is that Apple wants to over-engineer these systems to use highly custom board designs and cases to make these as small as possible. But for the niche they occupy, the end user is unlikely to care about that. That's the strategic mistake. The Pro should be powerful and expandable, but it's neither. It's dated and restricted. The mini is smaller, but there's no need to make it so small it can't be upgraded by an end user. A little bigger, and it could have M2 interfaces and maybe a couple of 3.5" bays internally. But you have to dismantle half the internals just to access the RAM slot, and the SSD is soldered. What a pain. I want a new Mac mini (or Pro) for my consulting development work. But the capabilities and price of current hardware makes it pointless. Even if I invest in one, who would want to run my code on such anaemic and overpriced hardware? They need to remember that while the phone and iMac are to a large extent fashion products, the high-end PC depends primarily upon functionality and price, and they've missed the mark for years on that front.
So, what they are really saying is that their illicit tracking is so effective that they already know who you are without an account.
What's also bad is that if you alone took the time to prepare for it, everyone else there is still clueless and unable to meaningfully participate! Some people even treat you badly because you're the only one with a clue! Sometimes I've emailed a summary of all my research/investigatory findings to the participants a day or so beforehand so they could easily get up to speed and been told that I wrote "a wall of text" and they couldn't read it (2 pages)! You just can't help some people. Meetings really are lowest common denominator affairs.
With the new apple keyboards, they are failing all the time. It really loses its value proposition when you might be without it for days while they replace half of the case.
The answer two posts up was implying that Linux ACLs were comparable with Windows ACLs. I explained that they are rather different than Windows ACLs, being both more limited in their functionality and being non-standard. I also explained that there was a third ACL type, NFSv4 ACLs which were a superset of both and available on several other Unix systems and which interoperate seamlessly with Windows systems as a result. The point being, that Linux here is the odd one out. It lacks a rich ACL model in its VFS layer, and consequently Linux can't support NFSv4 ACLS and is poorly interoperable with other systems using rich ACLs, as well as local filesystems using rich ACLs. I understand this is because the VFS maintainers don't like them, but this is a fairly significant functionality gap at this point.
Absolutely agreed, it must be available and convenient to work. I'll bet that we'll see the existing service stations replaced by places you can relax for a bit while you wait for a charge half way. Coffee shops, cafes, restaurants, tourist attractions etc. with chargers in all the parking spaces. I'd be happy to take a half hour break four hours in. But even before we get to this point, the vast majority of trips are much shorter than this. The people I work with who have them (mainly Nissan Leafs) are all really happy with them, and either have a second family car or rent one for long distance trips. They use the electric for all their trips around town, school, work, shops, sports etc. and it already serves that need very well. They are now coming onto the second hand market in larger numbers, and I have no doubt they will affect the market economics of fuel powered vehicles over the next few years.
In the short term, yes. But only the short term. Once there's less demand, that's going to result in knock-on effects in the medium term. Individual service stations aren't going to be economical, so there will be closures and consolidation (this already happened a decade or so back in the UK, due to the high prices). That means less deliveries, less road tankers needed, less storage capacity needed, and less demand for refineries, which will have to lower production. As demand reduces, the whole complex refining and distribution network will become less economical to operate. Once refineries themselves start to become uneconomical to operate, consolidating and closing, that will start to turn liquid fuels from a high-volume low-price commodity into a low-volume high-price specialty product because you'll have to get custom delivery from one of the few remaining refineries. The whole distribution network took decades to put in place, and it's costly to maintain. It will fall apart like a deck of cards when the foundation is knocked out. Without that distribution network, hauling your delivery is going to be expensive. A hundred years ago, hay production and distribution was of key importance to keeping cities and farms running, and was planned and monitored (such was the entrenched bureaucracy, original car registrations in London had a "hay allowance" IIRC). Today, hay is a niche market. The same will eventually come true for fossil fuels, and electric adoption will only hasten its demise. The transition might become sooner and more swiftly than we might think, once the tipping point is reached.
In the US anyway. It's already very expensive in much of the rest of the world. And that's possibly not going to last. As electric takes over, it's going to reduce demand for liquid fuel, and that will have knock-on effects on refineries and the vast distribution infrastructure which delivers it to all the local stations. Once the negative feedback kicks in, and local stations close, the distribution network contracts and refineries are cut back, it's going to get increasingly expensive. And that will further drive electric adoption and be self-reinforcing. Probably not for a good while yet, but if electric takes off seriously it's just a matter of time. Plus, if the rest of the world switches over to electric before the US, there will be pressure due to supply chain economics as well.
We don't need a new version of the IDE every 2 years. But the compiler and associated tools bring much needed bugfixes and stuff like C++17 and soon C++20. Shame it's all bundled so tightly together.
100% is likely an unattainable goal. But Wine has been painstakingly reverse engineered. They had to identify all the compatibility workarounds and play a never ending game of catch-up. If Microsoft were to make their own equivalent, they would be able to have much better coverage, with the original sources.
Linux has "POSIX.1e DRAFT" ACLs. They are functional, but limited, and based upon an unratified and abandoned draft standard. Other Unix sytems, like Solaris, IllumOS and FreeBSD, implemented NFSv4 ACLs which are both a ratified published standard and are compatible with both POSIX.1e DRAFT and Windows ACLs. (They are a superset of both.) If you're using NFSv4 ACLs they are modifiable and queryable from the command-line with get/setfacl and they are also modifiable and queryable from the Windows security/permissions property pages in the explorer (if exported via Samba or NFSv4). The whole ACL situation is limited by the fact that Linux hasn't implemented NFSv4 ACLs in the VFS. Yet filesystems like ZFS and NFSv4 use them, but they are hidden and inaccessible on Linux. If Linux implemented them, we would have pretty comprehensive and interoperable ACL support between all the major platforms.
Protein structure is intrinsically a computational chemistry / physics modelling problem. It requires modelling of solvent and solute interactions with the protein structure, van der Waals, dipole, ionic and other electrostatic interactions, hydrophobic/hydrophilic interactions with the solvent and itself (and lipid bilayer for membrane proteins), minimising free energy of the whole structure, of which there may be multiple stable variants under different conditions, and interactions between different subunits as well as multimeric associations. It's hard, and it's computationally expensive. I used to work with a team of such modellers, all hardcore physics and maths people, in the computational biology department I used to work in. "AI" might be able to recognise certain patterns. And it might be able to predict certain structural motifs with a reasonable degree of accuracy. But it will always be limited by the training dataset as the parent tried to explain. It's not modelling, it's guessing by interpolation. There's no intelligence; it can't extrapolate to make predictions which it hasn't been trained specifically for. And there's no guarantee that the structures it predicts will be stable or valid in any way. "AI" of this sort isn't magic, and it's certainly not intelligent. It's questionable that it's even a valid technique for good science. Because science and data analysis should be understandable, not a black box.
If the rest of Amazon is loss making, maybe it's the best thing to do in the medium term. You can't subsidise loss-making subsidiaries forever. There are plenty of other companies out there to pick up the slack. I for one have abandoned Amazon retail almost entirely; the marketplace damaged the trust I have in being able to purchase legitimate merchandise, and not some co-mingled knockoff. It's also not at all price competitive with other companies; products are often cheaper elsewhere, and I have zero reason for loyalty to them.
Yeah, I know there's more. But I can't believe there's that much more; not an order of magnitude more. I've worked as a graduate student and postdoc. The entire research institute I was in got by with a single secretary and a lab manager. Centralised university functions included payroll, copyright and patents, and wider departmental functions included purchasing and shared facilities. None of these individually employed that many people. I've personally had discussions regarding copyright and patents and commercialisation of research with the tech transfer people, and these functions were still pretty small. Just a couple of people for the entire university (UK). Conferences require a few staff to organise and take care of the logistics, for sure. But when we organised meetings, the organising researchers and departmental secretary handled it all for the most part. I saw them get seriously stressed about it, because it was all done on top of their regular commitments! Combined, I just don't see four large buildings full of people as justifiable. Large corporations can manage with a handful of admin staff, and I can't see why universities can't either, for the most part.
A few times recently I've had the IPv4 networking randomly break, but all the IPv6 services and websites worked without interruption. The autoconfiguration is worth something. Better than NetworkMangler which is the cause of the IPv4 outages, no doubt.
Most of that can be ignored though. Tunelling is dead; ignore it. It's 0.00% for the last couple of years, 0.01 or less for the last 6 years. 26% native today. So not important to learn-just go native. Link-local can be ignored for the most part; avahi/zeroconf and the like make it transparent. Subnets are also ignorable since it's part of the first 64 bits after the routing prefix; there is nothing to configure. For regular setup and use, all of this should be transparent and ignorable for the common case scenarios. On my network, it's all automatic with SLAAC, connect and it works.
It's happening. Look at this graph. Growth was exponential from 2010 to 2017, taking it from 0.1% to 16%. The last two years have been mostly linear, from 16% to 26% (~5%/year). The last 8 weeks alone have seen a 1.5% increase; that's equivalent to all the growth from 2010 to July 2013! 3.5 years compared to 8 weeks for the same improvement. We're well into the implementation phase now, with over 1 in 4 users using Google services over IPv6; the actual number is even higher, because Google underestimates it by requiring IPv6 to be explicitly whitelisted by them. It is taking time, no doubt about that, but it is happening at a decent clip now, and the pressure to provide it will increase. Already most of my internet traffic is over v6, and it's also more reliable. We're not far off the tipping point where it will start to be required.
It's a direct copy of the fundamental Unix system call and library APIs, with some extra stuff on top. It's a reimplementation of existing system design. The points about SCO are irrelevant; that was about the values of some constants in a header. Which by the way, were copied directly for compatibility. The SCO case was about whether that merited copyright protection, not about whether it was copied (it was, and that point wasn't in question). You're confusing copying an implementation with copying a design. The Linux kernel, GNU utilities, compiler and the rest are all reimplementations of existing ones. With a number of improvements and extensions on top, of course. The SCO ranting has rather missed the point I was making.
The explosion of administration staff is both surprising and absurd. For many years, I wondered what they all did! Lots of make-work, by the look of it. Lots of meetings, training, reading and sending emails to each other. But how much actual productive work? Precious little, IMO. How much "administration" do students and staff actually need to run a university? In the '90s, my department was run by three people, two undergraduate admins, one graduate. They handled admissions, collecting coursework, and all the other stuff. So three people for about 500 students. And some departmental secretaries/lab managers. It didn't seem too unreasonable. But a decade later, they increased their headcount by one or two as the department grew, while the central university administration had expanded from one building into four separate office blocks! W T F is all that for?! It's a bit of a mystery. I suspect that when they take ~£10k per student per year, they are so awash with cash (hundreds of millions, plus gouging overseas postgrad students) that there is simply no restraint upon their spending or expansion. They are charging such obscene amounts, which doesn't get spent on education for the most part; the amount going to lecturers, lab space and other teaching resources is a fraction of that, so where does it go? I can only guess onto admin, with liberal budgets and salaries to match. But where is the oversight of this, with someone to question the necessity of it all? I find the whole thing rather unsavoury and obscene, with vice chancellors on £500k and up, while students are fleeced and saddled with a lifetime of debt. Value for money, it isn't. While I have been fortunate to do an undergrad, postgrad and doctorate, it's most likely that I would not go today; it's not affordable given the cost, and hard to justify for the benefits it provides.
Don't you mean 1998? (I started in 1997, the second-to-last year of the grant system.)
The scheme is simple and takes just a few minutes to familiarise yourself with. That's all it is, familiarity. By the way, you only need two colons "::", which means "pad blocks with zeros". The rest are redundant. For example my link-local address is currently fe80::e2d5:5eff:fea8:50c9; my global address is something like 2001:8b0:860:ccbe:243b:81de:43b2:fb37. So it's 8 blocks of 4 hex digits, separated by colons, with optional eliding of ":0000:" with "::". That's it. Your nine year old should be able to understand it just as well as IPv4. He won't even need to learn about all the different IPv4 network classes.
I know you meant it as a joke, but language can change exactly like this. When I was at school, a popular insult was to call people a "spaz" (spastic). Because of its use as a generic insult, some do-gooders required everyone to refer to disabled people as "special". Result: "special" becomes a generic insult. Like: "You're a bit special, aren't you" is equivalent to "you're a spaz" or "you're a retard". More recently, the charity the Spastic Society renamed themselves to "Scope". Result: spastic children are now insultingly referred to as "scopers". So the word "scope" is now also an insult in its own right. The point being, no matter how much do-gooders and SJWs clamp down on acceptable language use, the remaining "permitted" words will be adapted to replace them, turning the "new inclusive terminology" on its own head. This isn't a battle they can win, because human nature, at its nastiest best, will always be creative enough to come up with new "bad" language! And while I don't condone bullying or insulting, it does make me happy to see people sticking their fingers up at the SJW-enforced norms.
When you look at what made Linux successful, little of it was originally to do with being revolutionary. The vast majority of it was re-implementing existing software under a free licence. The Linux kernel is a copy of Unix kernels. The GNU utilities are copies of Unix utilities. Likewise the compiler, desktop environments, and most of everything else. Being open and free made it more useful and compelling than the proprietary equivalents. Being a direct copy and following the existing standards made it easy to migrate to and use with little disruption. Having a good number of enhancements and improvements on top was the icing on the cake, but the core stuff was what made it indispensable. It's the "revolutionary" parts which have caused the most disruption, inconvenience and upset. They are also the parts which are the most poorly designed and implemented, and it's not a coincidence. A good number of these people are now arguing that POSIX and other standards are no longer relevant, but they are completely ignoring the main historical reasons why we have popular open source systems. Projects like Gnome, systemd have made some terrible design choices, and have also repeatedly broken compatibility with themselves over the years. Were they designed and implemented by competent professionals who could design and engineer systems to the standard of what came before them, that friction would not exist.
Agreed to a point. But they could have gone straight to chip+pin rather than the chip+signature setup which is almost pointless. When the rest of the whole world nearly is using chip+pin for nearly two decades now, it seems a bit odd to not use it. And regarding the magstripe fallback, has a date been set to drop it yet? If it was withdrawn from use and on new cards starting 2020, that would significantly curtain fraud.