Trusting Dell hardware is like trusting Facebook anti-malware. Once customers can't trust you not to deliberately and repeatedly screw them, your brand name has been permanently poisoned and the kiss of death *should* follow.
Regardless of the wishes of the many, for me, Dell is forever dead.
It's a shame Alienware too was lost to the Dark Side of Dell.
There's an even bigger demand for robots where labor is expensive.
If a developed country can build its widgets locally, it can avoid many costs that come with manufacturing remotely where labor is cheap: lower shipping costs, shorter time delays, less dependency on other governments and their dysfunctions, etc. When you control your own manufacturing (via local robots), you can more readily redesign and tailor your widgets to better serve the needs & interests of local customers. Eventually, all economies will be local, or at worst, regional. Why ship finished products long distances if you can get them made nearby? That's something that will become possible once labor costs fall to zero.
The other constraints to local manufacturing are the availability and cost of raw materials, the cost to dispose of factory chaff (pollution), and the cost of power. With the likely rise of cheap solar, power may no longer be in short supply, but the first two desiderata will shape the choice of manufacturing locales for a long time yet. In transition, those countries willing to shit where they eat will keep their factories. But once dirty manufacturing processes can be cleaned up, they'll be free to relocate anywhere.
As the OP suggests, the workers in economies that depend on the export of cheap labor (e.g. China and India) will be devastated if they lose their work to robots. Not only with upward mobility cease, but the jobs themself will go poof by the hundreds of millions, with no place for unskilled human workers to go but down.
Yes, and a GAN can easily combine features in unnatural ways that never existed in nature.
Unless the natural probability distribution of each clinically important feature can be accurately modeled by the deep net, combinatoric perversity is all too likely. (Features combos will arise synthetically that do not occur together in nature, but these images won't be excluded due to the lack of counterexamples in the training set).
I work in medical image analysis professionally, and I'm certain no physician or biologist would approve of faking images to learn about abnormalities / disease, no matter how 'plausible' they look. (However, GANs might be warranted if the objective is only to model the range of natural anatomic variation, rather than unnatural variations.)
There are better ways to augment pathology image data that avoid fakery.
An isopropyl alcohol bath is resoundingly insufficient to sterilize surgical instruments. This has been known for decades. Likewise, nobody in their right mind assumes a quick wipe with an alcohol pad will make your skin sterile either. Thus this news story adds nothing really new, except that some MRSA bugs may have become somewhat more resistant to a halfhearted swat of alcohol. Stop the presses...
They're similar. CAR-T is one form of ACT (Adoptive Cell Transfer), which the article names as their method. CAR-T employs CAR (chimeric antibody receptors) toward reengineering T cells only. The variant of ACT used here was TIL (tumor infiltrating lymphocytes) rather than T cells, apparently because this target involves a smaller number of target mutations than is suitable for CAR-T.
(I hope I've interpreted this correctly.)
And yes, this form of ACT should be just as expensive as CAR-T -- about a half million dollars per patient.
T1 and T2 diabetics need much more information than a simple binary +/-.
To be useful, a glucometer must provide a continuous readout of your glucose level. Based on that number, 1) a T1 knows how much insulin to inject, or 2) a T2 knows how much excess glucose they just ate, so they can adjust their diet or medication dose.
Raising a binary flag is worthless. This company's gadget is a gimmick, intended to announce victory before a competitor can steal their thunder with a *real* product.
Instead of planting landmines, plant sensors? LOTS of 'em? Maybe the true advantage of embedding computer simulation in real world battle wouldn't be improving military control and tactics as much as maximizing sensor data and finally *lifting* the fog of war.
Let's call the policy of planting billions of sensors everywhere 'BoSE'. In embattled settings, police and peacemaker orgs would willingly adopt BoSE as the ideal way to minimize the element of surprise, both during battle, as well as when preparing for it, like when observing those who plant IEDs.
Better yet, the presence of the *jillions* of BoSE sensors in urban settings would also enable persistent, independent, and global oversight of (esp. urban) violence, allowing independent monitors and courts to reconstruct the details of past events from the gobs of redundant sensor info. This unbiased evidence would be invaluable toward imparting real accountability on all actors in war. All actors would find it much harder to misbehave if crystal clear evidence of who did what when were available to all, and thus inescapable.
'If you look at all the engineers at Facebook, more than one in four are users of our AI platform,' says Mr. Candela [head of applied AI]. 'But more than 70% [of those] aren't experts.'
How so many Facebook engineers can use its AI algorithms without necessarily knowing how to build them, Mr. Candela says, is that the system is 'a very modular layered cake where you can plug in at any level you want.' He adds, 'The power of this is just hard to describe.' Pieces of that platform are performing all kinds of 'domain-specific' tasks across Facebook's properties, from translation to speech recognition.
This implies of the 25% of FB's engineers who use company AI services, 70% invoke it via a simple API without delving into the infrastructure or tuning it themselves.
Therefore only 7.5% of FB's AI users (30% of 25%) pass the Turing Test.
To be precise, actions by media owners that constrain or oppose *paid* speech (including false ads) are *not* violations of the First Amendment. Such constraints are well within the rights of all media outlets to deny unless they are 'common carriers' like telephone (or internet) providers.
Unless they are broadcast over public airwaves (which happens rarely anymore), any private- or corporate- owned medium has the right to dis/approve any paid promotional content from any customer for any reason. They cannot be forced to publish any 'free' speech (except perhaps Public Service Announcements coming from the government). However because denying political ads is likely to annoy many consumers, the airing of mainstream political ads largely unavoidable, even on subscription-only media.
The First Amendment ensures only that no non-commercial agent can *suppress* speech in a public forum -- unpaid speech in an unpaid medium. The law does not apply to any commercial (for profit) medium. Historically, because television and radio were broadcast using public airwaves, the FCC required they 'serve the public good' by guaranteeing *some* public access (like a community channel or late night shows where the public could speak out individually). But with the fading of over-the-air broadcast media, those channels of public free speech have largely been closed. At the same time, corporate media owners have steadily diminished public access to their outlets, such that I'm aware of *no* open forums it TV or radio that remain.
While the net has subsumed virtually all free speech that remains, any website may deny almost anyone access using means that circumvent lawful speech guarantee, such as violations of community guidelines, for example.
Now that all major media in the US is corporate- owned and public access to major media has been eradicated, in practice, free speech no longer exists in America. All that remains is net noise.
The article talks about Amazon's desire to compete with pharmacies, but the Bloomberg title (echoed verbatim by Slashdot) states Amazon wants to 'enter the prescription drug market'.
That wording is confusing. It implies Amazon wants to make AND sell prescription drugs -- the prescription drug market -- which is only half true.
Americans have been sitting motionless in front of TVs for hours each night for the past 50 years. Why would sitting in front of a computer be any more sedentary / unhealthy? And how could anyone measure the marginal difference between the two, since many of us do both at the same time?
IBM's "innovation" is to insert synchronizations in a render farm that enable the gathering of intermediate GPU results across a distributed batch run.
Of course, Google made the same "revolutionary discovery several years ago when Dean and crew first developed DistBelief. Later they abstracted it into TensorFlow's compute graphs. When was this, 2010?
Seems like you could put a vast array of solar panels in several places throughout Europe where cloud cover is minimal, like Greece or Spain. Surely these countries would be amenable to adding those new jobs and and are much more politically stable and secure than anywhere in north Africa, making this big investment far less risky.
Once Europe's solar farms are profitable, their success will encourage that economic model to spread and attract investment elsewhere, even where security and infrastructure is less stable, like the Sahara.
The house subcommittee has done nothing yet. They only allowed a proposal to be debated in committee en route to a vote by the full committee. Even if approved by the committee, the legislation has to be approved by a house vote, then a senate vote, then signed by Trump.
This initiative is still a long way from becoming law. No point in getting excited until it gets a lot closer to reality.
Yeah, this is still the best TV I have *ever* seen, documentary or not. Burns' style of combining still photo montage with low key multiple voice narration has spoiled me forever for actor-based recreations. They all seem disingenuous to me now.
Seconded. Burke was *the* reason to watch TLC. He followed "Connections" with "The Day the Universe Changed" 7 years later. I rank them both up there with "Cosmos" (Sagan).
And the Osborne 1 *shipped* in July 1981. Yeah it was a luggable portable not a laptop, but imagining that it would shrink and add a battery isn't exactly a big leap.
Now: "Driverless: Intelligent Cars and the Road Ahead" - Hod Lipson & Melba Kurman. Terrific so far. Rich with tech details.
Next: "American Spies: Modern Surveillance, Why You Should Care, and What to Do About It" - Jennifer Stisa Granick. 1984 has arrived. Time to face the enema.
"Wonderful Life: The Burgess Shale and the Nature of History" - Stephen Jay Gould. I once visited the Shale in the rain. It made many thousands of 100 million year old fossils clearly visible. An amazing experience.
"Reclaiming Conversation: The Power of Talk in a Digital Age" - Sherry Turkle. I love language and ideas too much to merely broadcast my life online.
This kind of retaliation is no different from a cellphone service provider jamming your RF signal. The FCC (if we still had one) should step in and either fine the manufacturer for retaliatory misbehavior, or punitively shut down their internet access for a nominal period (at least a week) for abusing the privilege of being online.
Doing this periodically would send a really constructive message to many others who routinely abuse others on the net, be they bad businesses or just trolls. Access to the net is a privilege, not a right.
"Based on the current version of the Handbook, the fact that a person may be employed as a computer programmer and may use information technology skills and knowledge to help an enterprise achieve its goals in the course of his or her job is not sufficient to establish the position as a specialty occupation. Thus, a petitioner may not rely solely on the Handbook to meet its burden when seeking to sponsor a beneficiary for a computer programmer position. Instead, a petitioner must provide other evidence to establish that the particular position is one in a specialty occupation as defined by 8 CFR 214.2(h)(4)(ii) that also meets one of the criteria
at 8 CFR 214.2(h)(4)(iii). Section 214(i)(1) of the INA; see also Royal Siam Corp. v. Chertoff, 484 F.3d 139, 147 (1st Cir. 2007).8."
Now you must offer more than a 2 year degree and no experience. You must somehow substantiate that you possess expertise. You should be "prominent", or a "recognized authority", or expert (as demonstrated by referreed publications or a thesis).
Your occupation must "require theoretical and practical application of a body of highly specialized knowledge in fields of human endeavor including, but not limited to, architecture, engineering, mathematics, physical sciences, social sciences, medicine and health, education, business specialties, accounting, law, theology, and the arts, and which requires the attainment of a bachelor's degree or higher in a specific specialty, or its equivalent, as a minimum for entry into the occupation in the United States."
Almost all successful software dev isn't about algorithms; it's business process engineering. If you don't understand how the business works now and how it needs to work in the future, your software will fail no matter how fast or cheap it's made.
Obviously business process design requires domain expertise but it really lives or dies on the presence of *outstanding* communication skills. If the actors don't exchange *all* the essential requirements and business dependencies, the deliverable will hit the wrong target. And the cost of failure is a lot higher than the savings from outsourcing.
This essential exchange of info is hampered badly by domain inexperience, distance, and latency between provider and consumer. Not only do comm skills enable focus on the desired goals, but also on the (many) unspoken needs and dependencies that often decide whether a deliverable actually plugs the big holes in the business' ship or the water pours through at will. I think US businesses have only discovered this recently (in the past decade) after many 'projects thrown over the ocean' have come back to haunt them as zombie projects (half dead creatures that suck the life out of you but somehow can't be killed). Until this need can be addressed internationally, the value of oursourcing forever will be limited.
As a ~60 year old developer, I've followed two major shifts since I began as a pro (in 1986).
First, Dev tools have become much bigger and more interdependent and mastery of the dev idioms takes longer now and can no longer be learned from books, as was once possible for C/Unix. Frameworks are used a lot more now (for better or worse), and S/W deliverables now depend on mixtures of languages and their libraries / components, third party APIs, and O/S service components / stacks. This requires broader awareness of the S/W ecosystem but allows less time for mastery of each component, since the ecosystem is so vast and the components change often.
Second, there are more forms of computing now that require a "deep dive" technically. These usually demand deeper skills in math, statistics, engineering, and maybe hardware (e.g. DSP, image processing, computer vision, real time O/S and Arduino, music, crypto, machine learning, or graphics). The classic academic CS preparation I got before 1990 did not prepare me to work in this space foundationally, and I've had to learn the math through remedial courses or self instruction.
So I guess I'm suggesting that any resource that can provide a good clear intro overview of A) software stacks and ecosystems, or B) the math and engineering beneath advanced tech subjects, I'd find both of these helpful.
A good example of this is Michael Nielsen's web site "Neural Nets and Deep Learning", which provides an excellent overview of those two topics while minimizing the jargon, arcane math notation, and flummery that so often pervades tech talk. Another example is Lyons' excellent book for non-engineers, 'Understanding Digital Signal Processing'. I'd love to see more material like those.
Slashdot Mirror
Trusting Dell hardware is like trusting Facebook anti-malware. Once customers can't trust you not to deliberately and repeatedly screw them, your brand name has been permanently poisoned and the kiss of death *should* follow.
Regardless of the wishes of the many, for me, Dell is forever dead.
It's a shame Alienware too was lost to the Dark Side of Dell.
/editmodeoff
There's an even bigger demand for robots where labor is expensive.
If a developed country can build its widgets locally, it can avoid many costs that come with manufacturing remotely where labor is cheap: lower shipping costs, shorter time delays, less dependency on other governments and their dysfunctions, etc. When you control your own manufacturing (via local robots), you can more readily redesign and tailor your widgets to better serve the needs & interests of local customers. Eventually, all economies will be local, or at worst, regional. Why ship finished products long distances if you can get them made nearby? That's something that will become possible once labor costs fall to zero.
The other constraints to local manufacturing are the availability and cost of raw materials, the cost to dispose of factory chaff (pollution), and the cost of power. With the likely rise of cheap solar, power may no longer be in short supply, but the first two desiderata will shape the choice of manufacturing locales for a long time yet. In transition, those countries willing to shit where they eat will keep their factories. But once dirty manufacturing processes can be cleaned up, they'll be free to relocate anywhere.
As the OP suggests, the workers in economies that depend on the export of cheap labor (e.g. China and India) will be devastated if they lose their work to robots. Not only with upward mobility cease, but the jobs themself will go poof by the hundreds of millions, with no place for unskilled human workers to go but down.
Yes, and a GAN can easily combine features in unnatural ways that never existed in nature.
Unless the natural probability distribution of each clinically important feature can be accurately modeled by the deep net, combinatoric perversity is all too likely. (Features combos will arise synthetically that do not occur together in nature, but these images won't be excluded due to the lack of counterexamples in the training set).
I work in medical image analysis professionally, and I'm certain no physician or biologist would approve of faking images to learn about abnormalities / disease, no matter how 'plausible' they look. (However, GANs might be warranted if the objective is only to model the range of natural anatomic variation, rather than unnatural variations.)
There are better ways to augment pathology image data that avoid fakery.
Wadha has a bachelor's in computing and an MBA. If he knows anything about arts or humanities, it's unlikely he learned it in college.
And despite the OP's attribution, he's not an engineering professor either.
An isopropyl alcohol bath is resoundingly insufficient to sterilize surgical instruments. This has been known for decades. Likewise, nobody in their right mind assumes a quick wipe with an alcohol pad will make your skin sterile either. Thus this news story adds nothing really new, except that some MRSA bugs may have become somewhat more resistant to a halfhearted swat of alcohol. Stop the presses...
They're similar. CAR-T is one form of ACT (Adoptive Cell Transfer), which the article names as their method. CAR-T employs CAR (chimeric antibody receptors) toward reengineering T cells only. The variant of ACT used here was TIL (tumor infiltrating lymphocytes) rather than T cells, apparently because this target involves a smaller number of target mutations than is suitable for CAR-T.
(I hope I've interpreted this correctly.)
And yes, this form of ACT should be just as expensive as CAR-T -- about a half million dollars per patient.
There's a nice summary here:
https://www.cancer.gov/news-ev...
T1 and T2 diabetics need much more information than a simple binary +/-.
To be useful, a glucometer must provide a continuous readout of your glucose level. Based on that number, 1) a T1 knows how much insulin to inject, or 2) a T2 knows how much excess glucose they just ate, so they can adjust their diet or medication dose.
Raising a binary flag is worthless. This company's gadget is a gimmick, intended to announce victory before a competitor can steal their thunder with a *real* product.
https://www.mckinsey.com/globa...
The 5 MB PDF:
https://www.mckinsey.com/~/med...
Instead of planting landmines, plant sensors? LOTS of 'em? Maybe the true advantage of embedding computer simulation in real world battle wouldn't be improving military control and tactics as much as maximizing sensor data and finally *lifting* the fog of war.
Let's call the policy of planting billions of sensors everywhere 'BoSE'. In embattled settings, police and peacemaker orgs would willingly adopt BoSE as the ideal way to minimize the element of surprise, both during battle, as well as when preparing for it, like when observing those who plant IEDs.
Better yet, the presence of the *jillions* of BoSE sensors in urban settings would also enable persistent, independent, and global oversight of (esp. urban) violence, allowing independent monitors and courts to reconstruct the details of past events from the gobs of redundant sensor info. This unbiased evidence would be invaluable toward imparting real accountability on all actors in war. All actors would find it much harder to misbehave if crystal clear evidence of who did what when were available to all, and thus inescapable.
More of the article:
'If you look at all the engineers at Facebook, more than one in four are users of our AI platform,' says Mr. Candela [head of applied AI]. 'But more than 70% [of those] aren't experts.'
How so many Facebook engineers can use its AI algorithms without necessarily knowing how to build them, Mr. Candela says, is that the system is 'a very modular layered cake where you can plug in at any level you want.' He adds, 'The power of this is just hard to describe.' Pieces of that platform are performing all kinds of 'domain-specific' tasks across Facebook's properties, from translation to speech recognition.
This implies of the 25% of FB's engineers who use company AI services, 70% invoke it via a simple API without delving into the infrastructure or tuning it themselves.
Therefore only 7.5% of FB's AI users (30% of 25%) pass the Turing Test.
To be precise, actions by media owners that constrain or oppose *paid* speech (including false ads) are *not* violations of the First Amendment. Such constraints are well within the rights of all media outlets to deny unless they are 'common carriers' like telephone (or internet) providers.
Unless they are broadcast over public airwaves (which happens rarely anymore), any private- or corporate- owned medium has the right to dis/approve any paid promotional content from any customer for any reason. They cannot be forced to publish any 'free' speech (except perhaps Public Service Announcements coming from the government). However because denying political ads is likely to annoy many consumers, the airing of mainstream political ads largely unavoidable, even on subscription-only media.
The First Amendment ensures only that no non-commercial agent can *suppress* speech in a public forum -- unpaid speech in an unpaid medium. The law does not apply to any commercial (for profit) medium. Historically, because television and radio were broadcast using public airwaves, the FCC required they 'serve the public good' by guaranteeing *some* public access (like a community channel or late night shows where the public could speak out individually). But with the fading of over-the-air broadcast media, those channels of public free speech have largely been closed. At the same time, corporate media owners have steadily diminished public access to their outlets, such that I'm aware of *no* open forums it TV or radio that remain.
While the net has subsumed virtually all free speech that remains, any website may deny almost anyone access using means that circumvent lawful speech guarantee, such as violations of community guidelines, for example.
Now that all major media in the US is corporate- owned and public access to major media has been eradicated, in practice, free speech no longer exists in America. All that remains is net noise.
The article talks about Amazon's desire to compete with pharmacies, but the Bloomberg title (echoed verbatim by Slashdot) states Amazon wants to 'enter the prescription drug market'.
That wording is confusing. It implies Amazon wants to make AND sell prescription drugs -- the prescription drug market -- which is only half true.
Why not just say: "Amazon: Your Next Pharmacy"?
Americans have been sitting motionless in front of TVs for hours each night for the past 50 years. Why would sitting in front of a computer be any more sedentary / unhealthy? And how could anyone measure the marginal difference between the two, since many of us do both at the same time?
IBM's "innovation" is to insert synchronizations in a render farm that enable the gathering of intermediate GPU results across a distributed batch run.
Of course, Google made the same "revolutionary discovery several years ago when Dean and crew first developed DistBelief. Later they abstracted it into TensorFlow's compute graphs. When was this, 2010?
Yeah. Another Big Blue Breakthrough.
Seems like you could put a vast array of solar panels in several places throughout Europe where cloud cover is minimal, like Greece or Spain. Surely these countries would be amenable to adding those new jobs and and are much more politically stable and secure than anywhere in north Africa, making this big investment far less risky.
Once Europe's solar farms are profitable, their success will encourage that economic model to spread and attract investment elsewhere, even where security and infrastructure is less stable, like the Sahara.
The house subcommittee has done nothing yet. They only allowed a proposal to be debated in committee en route to a vote by the full committee. Even if approved by the committee, the legislation has to be approved by a house vote, then a senate vote, then signed by Trump.
This initiative is still a long way from becoming law. No point in getting excited until it gets a lot closer to reality.
Yeah, this is still the best TV I have *ever* seen, documentary or not. Burns' style of combining still photo montage with low key multiple voice narration has spoiled me forever for actor-based recreations. They all seem disingenuous to me now.
Seconded. Burke was *the* reason to watch TLC. He followed "Connections" with "The Day the Universe Changed" 7 years later. I rank them both up there with "Cosmos" (Sagan).
And the Osborne 1 *shipped* in July 1981. Yeah it was a luggable portable not a laptop, but imagining that it would shrink and add a battery isn't exactly a big leap.
Now:
"Driverless: Intelligent Cars and the Road Ahead" - Hod Lipson & Melba Kurman. Terrific so far. Rich with tech details.
Next:
"American Spies: Modern Surveillance, Why You Should Care, and What to Do About It" - Jennifer Stisa Granick. 1984 has arrived. Time to face the enema.
"Wonderful Life: The Burgess Shale and the Nature of History" - Stephen Jay Gould. I once visited the Shale in the rain. It made many thousands of 100 million year old fossils clearly visible. An amazing experience.
"Reclaiming Conversation: The Power of Talk in a Digital Age" - Sherry Turkle. I love language and ideas too much to merely broadcast my life online.
This kind of retaliation is no different from a cellphone service provider jamming your RF signal. The FCC (if we still had one) should step in and either fine the manufacturer for retaliatory misbehavior, or punitively shut down their internet access for a nominal period (at least a week) for abusing the privilege of being online.
Doing this periodically would send a really constructive message to many others who routinely abuse others on the net, be they bad businesses or just trolls. Access to the net is a privilege, not a right.
According to the US code:
https://www.nafsa.org/_/file/_...
"Based on the current version of the Handbook, the fact that a person may be employed as a computer programmer and may use information technology skills and knowledge to help an enterprise achieve its goals in the course of his or her job is not sufficient to establish the position as a specialty occupation. Thus, a petitioner may not rely solely on the Handbook to meet its burden when seeking to sponsor a beneficiary for a computer programmer position. Instead, a petitioner must provide other evidence to establish that the particular position is one in a specialty occupation as defined by 8 CFR 214.2(h)(4)(ii) that also meets one of the criteria
at 8 CFR 214.2(h)(4)(iii). Section 214(i)(1) of the INA; see also Royal Siam Corp. v. Chertoff, 484 F.3d 139, 147 (1st Cir. 2007).8."
Now you must offer more than a 2 year degree and no experience. You must somehow substantiate that you possess expertise. You should be "prominent", or a "recognized authority", or expert (as demonstrated by referreed publications or a thesis).
Your occupation must "require theoretical and practical application of a body of highly specialized knowledge in fields of human endeavor including, but not limited to, architecture, engineering, mathematics, physical sciences, social sciences, medicine and health, education, business specialties, accounting, law, theology, and the arts, and which requires the attainment of a bachelor's degree or higher in a specific specialty, or its equivalent, as a minimum for entry into the occupation in the United States."
Almost all successful software dev isn't about algorithms; it's business process engineering. If you don't understand how the business works now and how it needs to work in the future, your software will fail no matter how fast or cheap it's made.
Obviously business process design requires domain expertise but it really lives or dies on the presence of *outstanding* communication skills. If the actors don't exchange *all* the essential requirements and business dependencies, the deliverable will hit the wrong target. And the cost of failure is a lot higher than the savings from outsourcing.
This essential exchange of info is hampered badly by domain inexperience, distance, and latency between provider and consumer. Not only do comm skills enable focus on the desired goals, but also on the (many) unspoken needs and dependencies that often decide whether a deliverable actually plugs the big holes in the business' ship or the water pours through at will. I think US businesses have only discovered this recently (in the past decade) after many 'projects thrown over the ocean' have come back to haunt them as zombie projects (half dead creatures that suck the life out of you but somehow can't be killed). Until this need can be addressed internationally, the value of oursourcing forever will be limited.
As a ~60 year old developer, I've followed two major shifts since I began as a pro (in 1986).
First, Dev tools have become much bigger and more interdependent and mastery of the dev idioms takes longer now and can no longer be learned from books, as was once possible for C/Unix. Frameworks are used a lot more now (for better or worse), and S/W deliverables now depend on mixtures of languages and their libraries / components, third party APIs, and O/S service components / stacks. This requires broader awareness of the S/W ecosystem but allows less time for mastery of each component, since the ecosystem is so vast and the components change often.
Second, there are more forms of computing now that require a "deep dive" technically. These usually demand deeper skills in math, statistics, engineering, and maybe hardware (e.g. DSP, image processing, computer vision, real time O/S and Arduino, music, crypto, machine learning, or graphics). The classic academic CS preparation I got before 1990 did not prepare me to work in this space foundationally, and I've had to learn the math through remedial courses or self instruction.
So I guess I'm suggesting that any resource that can provide a good clear intro overview of A) software stacks and ecosystems, or B) the math and engineering beneath advanced tech subjects, I'd find both of these helpful.
A good example of this is Michael Nielsen's web site "Neural Nets and Deep Learning", which provides an excellent overview of those two topics while minimizing the jargon, arcane math notation, and flummery that so often pervades tech talk. Another example is Lyons' excellent book for non-engineers, 'Understanding Digital Signal Processing'. I'd love to see more material like those.