Like VB.net. I've never seen a tidy simple language morph into anything so monstrous. C++ comes close, though.
Frankenly, I'm forever amazed at how low the standard is for programming language enrapture. I'm sure it's due partly to 1) a lack of experience with other languages, and 2) a lack of imagination.
Sadly, I suspect the biggest source of this pathology is that most programming tasks are mindblowingly dull: load/unload a form, query a RDBMS, check for bad data, etc). Developers inevitably want a way to spice up deadend tasks like these (which admittedly is a better response than suicide). So they willing adopt / invent wildly overcomplex tools in the hopes that their humdrum existence will gain meaning. Or distraction at least.
Many less mainstream languages are *far* smaller, more legible, and sufficiently expressive as those at the Top of the Pops (e.g. C++, Java, Python), but they don't receive the needed mindshare (or libraries) needed to compete. Perhaps they just need their syntax to be a *bit* more Kafkaesque. (So what *is* Larry Wall doing these days?)
I strongly agree with most of your post, but direct democratic governance is an invitation to manipulate the uninformed voter. Left to a direct democratic vote, we'd have dozens of added fatuous amendments, like outlawing flag burning, embracing christianity over other religions, and requiring onerous voter ID enforcement.
A preferable alternative might be to ask registered voters to take a knowledge test that apportions a greater/lesser weight to their vote in proportion to their score. That way the informed electorate would have greater impact on policy and the clueless something less.
A thorny problem. But almost any change would be an improvement over today's status quo.
I don't tweet or text, but I skim through a lot of online news and articles. In the past couple of years, I've found it increasingly difficult to work my way through serious technical writing (e.g. research papers), math, or worse yet, my old code.
Yes, science & engineering papers are notoriously tersely (badly) written, as are most math and eng books. But these days I find myself almost unable to slow down and step through difficult passages. I gloss over the sticky stuff much more than I did maybe 20 years ago.
Consider Colorado State. They offer numerous on-line-only degree programs. Look at their Master's program in Computer Science.
(There's no point in earning another bachelor's when a MS is just as fast and requires only 10 semester courses. It's done all the time. I did it with a BS in zoology. You usually take a couple prerequisite courses at a local comm college then enroll as a grad student.)
I strongly recommend Georgia Tech's new MS in CS too. For the price, I'm confident you can't do better, although it will take several years before GT can offer courses on the full range of CS topics.
If you could pony up $50k somehow, you might also consider Stanford or Columbia, both of whom offer excellent MS in CS programs entirely on-line.
Personally I would stay VERY FAR away from schools that are on-line ONLY. AFAIK, all major tech employers have no respect for them. If you compare the workload (difficulty of textbooks, homework, and exams) with those at excellent state schools (like Georgia Tech), they do not compare well.
If you do consider such a school, I strongly recommend you contact several managers at companies you respect (via LinkedIn?) and ask if they hire graduates from those schools. Don't just assume that they do. And avoid HR staff. They know little about assessing candidate abilities.
OK, you've identified a bug that happened to fly past your tiny laser beam. Ignoring the zillion other bugs in your yard that did NOT fly past your beam, now you need to track this bug to confirm its location before you:
1) Turn on/off a gigantic bug zapper that will zap ONLY the bug you've targeted. And you'll do this by instantly powering up a large UV lamp and power grid that draws your moth straight to your flame?
2) Shoot the bugger down? With what, a missile? A laser? How long is your gun targeting system going to continue to work (safely) when left outdoors in-or-near the weather for months or years?
3) How many of these contraptions will you need to control your bad bug population? A laser bug zapper is unlikely to de-bug more than 1/8 acre. Is a farmer really going to buy 320 of these to patrol his back 40?
Lit has to convince drivers (not bikers) that the C1 is worth the premium over a bike. But you could pour million$ into ads to promote that difference, or get a dozen road test reviews in the major car/bike mags, and *still* not get the word out.
OR... You could feature the C1 in a movie, maybe sci fi or better yet, a movie about Makers. Those visuals would go a long way, especially footage of the C1 swooping through some esses -- "Where no bike has gone before".
Maybe the C1 could be one of Tony Stark's ubertech toys in the next Iron Man flick?
The company *is* located in LA, not 10 miles from Tinseltown...
Sure, GCC, Linux, sendmail, SCCS, any many more are essential to the open software stack I would die for. But the importance of a modern, full-featured, open browser simply can't be overstated. Without Firefox, the web would be a much less trustworthy world, and I'd be much less willing to take part in it.
Breaking up the NSA will force the issues of what data on Americans may be gathered by whom. Prior to the Patriot Act(s), no data on American citizens could be gathered unless authorization was requested through a search warrant. The FISA court has been a clumsy blunt force attempt to circumvent this, and it has relied on _temporary_ wartime provisions from the Patriot Acts. If the Constitution is to survive, the Pat Acts must be rewritten (or better yet, repealed whole). Only then can these lacerations in law heal, and official abuses of intel will end.
Until the Supremes act and Pat Acts "meet their makers", the only way to minimize this abuse of data is rigorous formal legal oversight either at the agency level, where the FBI has to defend WHY it has gathered this data.
Failing that, we have tio rely on oversight where it matters most but is hardest to apply consistenly -- the evidentiary discovery process in each court case where this data or products of its analysis is used. Unflinching feedback like this is necessary since the interior of agencies like the NSA are black.
In these times of Rising Police State USA, "1984" and "Brave New World" could not be more relevant or compelling as reflections on where unmonitored unchecked government power will lead. Likewise, "The Federalist Papers" serve as perhaps the best reminder that freedom requires cool careful consideration and eternal vigilence -- by each of us, of our own government.
On a lighter tech note, I also choose "Elements of Style" by Strunk and White. Not only is it the best book ever written on how to express yourself clearly, but I know no other book that better prepares you to think clearly too. Likewise, Kernighan and Ritchie's "The C Programming Language" is a paragon of clarity and simplicity in describing the principles and syntax of a programming language. The brevity of the original edition (at 228 pgs) is also a reminder of the need for economy and focus in design, *especially* among today's enormous, complex, one-tool-for-all-problems languages and the 1000+ page tomes needed to describe them.
GEB *is* a great book. Like it or not, it's a read that you will never forget. Hofstader's writing is a fugue itself, composing and playing variations on many themes as the narrative unwinds. Think of it as a fictional tale on metaphysics where the plot ricochets and glances off real ideas grounded in mathematical principles, some of which are apparent, while others need a little coaxing to show themselves.
I think the trick is for the reader to have patience. Don't push yourself to finish the book in a timely manner. Feel free to skip ahead or put it down periodically. As you read, consider adding a subplot or two of your own invention, so that the words of Achilles and the Tortoise might reveal more of your own Matrix.
After much thought, I've concluded that robotics is a Faustian bargain. The best policy to their onset is to delay and obstruct them by any means necessary.
Yes, automation will make products and services more available. But in every case the cost will be the loss of a human skill and a job. This trend will (and must) continue until all human skills and jobs finally perish. Ultimately all human endeavors, not just life's difficulties like work but it's joys like art will be better done by a robot. This progression will be unstoppable.
In a vain attempt to keep up, man will have been upgrading ourselves cybernetically. In the end we will have no biology left -- we'll be 100% robot.
No thanks. It's time to get off this merry go round.
You know C. C is simple, as fast as any alternative, it's straightforward to optimize (aside from pointer abuse), and you always know what the compiler/runtime is doing. And threading libraries like pthreads or CUDA are best served via C/C++. Why use anything else?
Another thought: scientific libraries. If you need external services/algorithms then your chosen language should support the libraries you need. C/C++ are well served by many fast machine learning libs such as FANN, LIBSVM, OpenCV, not to mention CBLAS, LinPACK, etc.
And you're claiming the work is invalid because you're unimpressed by the lack of pubs of a new research program. At Stanford?
In short, the Wired article is interesting while your criticism adds nothing. My advice, FWIW: if you must criticize, be specific. Don't gainsay with, "Your work is uninteresting because I'm unconvinced."
Treason is a rare and essentially non-legal label, normally used only by political blowhards like Senator Joseph McCarthy. And now Melissa Mayer.
I suspect Mayer is trying to make Yahoo look good after they sniveled, cowered, and cringed their way into the NSA's good graces by handing over every imaginable user confidence. While joyously waving the flag, of course.
It's tough enough reading PDFs on a full iPad *with* a retina display. On a smaller form factor like a Mini combined with its lower resolution -- fagetaboudit.
And don't believe for a minute that non-PDF textbooks are an option. Books with equations, graphics, tables, or color render quite poorly and inconsistently as ebooks.
Clearly this school plans to graduate only readers of plaintext fare like novels and poetry. And in that case, why not use cheaper B&W Kindles or Nooks?
The blog does NOT say the son searched for instructions on how to build a bomb. Here it is:
" Most of it was innocent enough. I had researched pressure cookers. My husband was looking for a backpack. And maybe in another time those two things together would have seemed innocuous, but we are in âoethese timesâ now. And in these times, when things like the Boston bombing happen, you spend a lot of time on the internet reading about it and, if you are my exceedingly curious news junkie of a twenty-year-old son, you click a lot of links when you read the myriad of stories. You might just read a CNN piece about how bomb making instructions are readily available on the internet and you will in all probability, if you are that kid, click the link provided.
Which might not raise any red flags. Because who wasnâ(TM)t reading those stories? Who wasnâ(TM)t clicking those links? But my son's reading habits combined with my search for a pressure cooker and my husbandâ(TM)s search for a backpack set off an alarm of sorts at the joint terrorism task force headquarters.
Thatâ(TM)s how I imagine it played out, anyhow. Lots of bells and whistles and a crowd of task force workers huddled around a computer screen looking at our Google history. "
She assumes her son could have clicked on a link. But she does *not* say he did, contrary to your claim.
If the schools think using cameras to automatically identify students is going to be cheaper or work better... I doubt it. Students in hoodies, hats, or sunglasses are going to play hell with face recognition software, not to mention partial occlusions, bad angles, shadows, and poor illumination. That means many (most?) kids will go unrecognized or misrecognized, and miscounted.
Add to that the inevitable bright idea that they also look for unfamiliar faces and then sound an alert when a stranger is seen lurking on the premesis... I foresee many false alarms.
Are these school administrators dimwits? Don't they first *test* a cool new idea before adopting it? Or maybe in Texas they can afford to waste taxpayer megabucks...
Soulskill may well be right. This is NOT your everyday intel analyst:
" As part of a pioneering IT function, you will be a key participant in the design and delivery of a variety of strategic programmes and business critical initiatives within a dynamic, challenging workplace....
The role
You will provide tactical solutions and operational support to business users of information exploitation systems. You will create new data ingest processes, advise team members about their design and implementation and support data management processes within our organisation. "
I've worked inside several US intel orgs, but I've never seen a spy role where you "support business users" to better exploit THEIR data.
For 8 years I worked as a "Compuational Scientist" at 3 university supercomputing centers, helping students and faculty to use/program parallel computers. I don't have a PhD.
I saw several kinds of computing staff in academia. Some assist in research; most don't. They are: - graduate students (CS, engineering, physical sciences, psychology, medical research, etc) - post docs (w/ same areas of research) - IT support staff (cluster or workstation or network admins, univ administrative app programmers, etc) - research support staff (usu. project-based, paid for by a professor / lab, lasts as long as the contract, usually 1-3 years)
The exception to the above roles is the permanent staff research programmer (at large specialized computing centers or univs with large labs within engineering departments). These are funded for longer cycles (usually 4-5 years), and have numerous executive staff who do nothing but maintain the flow of funds. Examples: NCSA, Texas Advanced Computing Center, San Diego Supercomputing Center, etc.
Many folks do make a living as an academic developer, but they tend to have a specialized skills (HPC / scientific computing, statistical analysis / data mining, data visualization, engineering, etc). Most such jobs do not require a PhD, but univ employers prefer to hire as advanced a degree as possible, especially if the research/work is likely to lead to academic papers. If you can't do that, your role will be marginalized (subordinate to the university's main mission of granting degrees and writing papers).
The ideal research programmer can perform like a research scientist -- find their own funding, write their own research proposals, collaborate with other people's research projects, etc. Sometimes this work extends to US gov't SBIR projects (small business investigative research), which often bleeds over into US military contract work. If you can help to write your own ticket, you can remain working in academia indefinitely. If not, you will need to look for a new research position every 2-4 years.
Personally, I think the two best reasons to get an academic IT job are: 1) to be on-campus for just long enough to finish a grad degree, or 2) to settle into a low-stress long-term IT support job that offers a relaxed lifestyles and good benefits. From my experience, academic support jobs are not especially challenging or rewarding, and academic research jobs are generally narrow-focused and short-lived.
Slashdot Mirror
Like VB.net. I've never seen a tidy simple language morph into anything so monstrous. C++ comes close, though.
Frankenly, I'm forever amazed at how low the standard is for programming language enrapture. I'm sure it's due partly to 1) a lack of experience with other languages, and 2) a lack of imagination.
Sadly, I suspect the biggest source of this pathology is that most programming tasks are mindblowingly dull: load/unload a form, query a RDBMS, check for bad data, etc). Developers inevitably want a way to spice up deadend tasks like these (which admittedly is a better response than suicide). So they willing adopt / invent wildly overcomplex tools in the hopes that their humdrum existence will gain meaning. Or distraction at least.
Many less mainstream languages are *far* smaller, more legible, and sufficiently expressive as those at the Top of the Pops (e.g. C++, Java, Python), but they don't receive the needed mindshare (or libraries) needed to compete. Perhaps they just need their syntax to be a *bit* more Kafkaesque. (So what *is* Larry Wall doing these days?)
Or I suppose they could just add pointers...
I strongly agree with most of your post, but direct democratic governance is an invitation to manipulate the uninformed voter. Left to a direct democratic vote, we'd have dozens of added fatuous amendments, like outlawing flag burning, embracing christianity over other religions, and requiring onerous voter ID enforcement.
A preferable alternative might be to ask registered voters to take a knowledge test that apportions a greater/lesser weight to their vote in proportion to their score. That way the informed electorate would have greater impact on policy and the clueless something less.
A thorny problem. But almost any change would be an improvement over today's status quo.
I don't tweet or text, but I skim through a lot of online news and articles. In the past couple of years, I've found it increasingly difficult to work my way through serious technical writing (e.g. research papers), math, or worse yet, my old code.
Yes, science & engineering papers are notoriously tersely (badly) written, as are most math and eng books. But these days I find myself almost unable to slow down and step through difficult passages. I gloss over the sticky stuff much more than I did maybe 20 years ago.
Maybe my brain is getting old. Maybe not.
Consider Colorado State. They offer numerous on-line-only degree programs. Look at their Master's program in Computer Science.
(There's no point in earning another bachelor's when a MS is just as fast and requires only 10 semester courses. It's done all the time. I did it with a BS in zoology. You usually take a couple prerequisite courses at a local comm college then enroll as a grad student.)
http://www.online.colostate.ed...
I assume you live in Colorado and would pay a lot less for in-state tuition there. That's why I suggest CSU. Or University of Colorado.
http://cuengineeringonline.col...
I strongly recommend Georgia Tech's new MS in CS too. For the price, I'm confident you can't do better, although it will take several years before GT can offer courses on the full range of CS topics.
If you could pony up $50k somehow, you might also consider Stanford or Columbia, both of whom offer excellent MS in CS programs entirely on-line.
Personally I would stay VERY FAR away from schools that are on-line ONLY. AFAIK, all major tech employers have no respect for them. If you compare the workload (difficulty of textbooks, homework, and exams) with those at excellent state schools (like Georgia Tech), they do not compare well.
If you do consider such a school, I strongly recommend you contact several managers at companies you respect (via LinkedIn?) and ask if they hire graduates from those schools. Don't just assume that they do. And avoid HR staff. They know little about assessing candidate abilities.
I was moments away from buying Oculus' 2nd gen SDK just to play with the thing. It could have been a blast.
But now that they've been assimilated by the Borg, Oculus VR has been mortally poisoned. What a shame.
OK, you've identified a bug that happened to fly past your tiny laser beam. Ignoring the zillion other bugs in your yard that did NOT fly past your beam, now you need to track this bug to confirm its location before you:
1) Turn on/off a gigantic bug zapper that will zap ONLY the bug you've targeted. And you'll do this by instantly powering up a large UV lamp and power grid that draws your moth straight to your flame?
2) Shoot the bugger down? With what, a missile? A laser? How long is your gun targeting system going to continue to work (safely) when left outdoors in-or-near the weather for months or years?
3) How many of these contraptions will you need to control your bad bug population? A laser bug zapper is unlikely to de-bug more than 1/8 acre. Is a farmer really going to buy 320 of these to patrol his back 40?
Not buggy likely.
Lit has to convince drivers (not bikers) that the C1 is worth the premium over a bike. But you could pour million$ into ads to promote that difference, or get a dozen road test reviews in the major car/bike mags, and *still* not get the word out.
OR... You could feature the C1 in a movie, maybe sci fi or better yet, a movie about Makers. Those visuals would go a long way, especially footage of the C1 swooping through some esses -- "Where no bike has gone before".
Maybe the C1 could be one of Tony Stark's ubertech toys in the next Iron Man flick?
The company *is* located in LA, not 10 miles from Tinseltown...
Sure, GCC, Linux, sendmail, SCCS, any many more are essential to the open software stack I would die for. But the importance of a modern, full-featured, open browser simply can't be overstated. Without Firefox, the web would be a much less trustworthy world, and I'd be much less willing to take part in it.
Breaking up the NSA will force the issues of what data on Americans may be gathered by whom. Prior to the Patriot Act(s), no data on American citizens could be gathered unless authorization was requested through a search warrant. The FISA court has been a clumsy blunt force attempt to circumvent this, and it has relied on _temporary_ wartime provisions from the Patriot Acts. If the Constitution is to survive, the Pat Acts must be rewritten (or better yet, repealed whole). Only then can these lacerations in law heal, and official abuses of intel will end.
Until the Supremes act and Pat Acts "meet their makers", the only way to minimize this abuse of data is rigorous formal legal oversight either at the agency level, where the FBI has to defend WHY it has gathered this data.
Failing that, we have tio rely on oversight where it matters most but is hardest to apply consistenly -- the evidentiary discovery process in each court case where this data or products of its analysis is used. Unflinching feedback like this is necessary since the interior of agencies like the NSA are black.
Like the souls of lazy pols and bureaucrats...
Google just spent 100 times as much as Dyson ($700M) to hire ~100 top AI talent.
At that market rate Dyson's 5M pounds would yield a staff of five.
In these times of Rising Police State USA, "1984" and "Brave New World" could not be more relevant or compelling as reflections on where unmonitored unchecked government power will lead. Likewise, "The Federalist Papers" serve as perhaps the best reminder that freedom requires cool careful consideration and eternal vigilence -- by each of us, of our own government.
On a lighter tech note, I also choose "Elements of Style" by Strunk and White. Not only is it the best book ever written on how to express yourself clearly, but I know no other book that better prepares you to think clearly too. Likewise, Kernighan and Ritchie's "The C Programming Language" is a paragon of clarity and simplicity in describing the principles and syntax of a programming language. The brevity of the original edition (at 228 pgs) is also a reminder of the need for economy and focus in design, *especially* among today's enormous, complex, one-tool-for-all-problems languages and the 1000+ page tomes needed to describe them.
GEB *is* a great book. Like it or not, it's a read that you will never forget. Hofstader's writing is a fugue itself, composing and playing variations on many themes as the narrative unwinds. Think of it as a fictional tale on metaphysics where the plot ricochets and glances off real ideas grounded in mathematical principles, some of which are apparent, while others need a little coaxing to show themselves.
I think the trick is for the reader to have patience. Don't push yourself to finish the book in a timely manner. Feel free to skip ahead or put it down periodically. As you read, consider adding a subplot or two of your own invention, so that the words of Achilles and the Tortoise might reveal more of your own Matrix.
After much thought, I've concluded that robotics is a Faustian bargain. The best policy to their onset is to delay and obstruct them by any means necessary.
Yes, automation will make products and services more available. But in every case the cost will be the loss of a human skill and a job. This trend will (and must) continue until all human skills and jobs finally perish. Ultimately all human endeavors, not just life's difficulties like work but it's joys like art will be better done by a robot. This progression will be unstoppable.
In a vain attempt to keep up, man will have been upgrading ourselves cybernetically. In the end we will have no biology left -- we'll be 100% robot.
No thanks. It's time to get off this merry go round.
There is no University of Philadelphia, only a Philadelphia University, which would also be wrong.
The news is from the University of Pennsylvania.
You know C. C is simple, as fast as any alternative, it's straightforward to optimize (aside from pointer abuse), and you always know what the compiler/runtime is doing. And threading libraries like pthreads or CUDA are best served via C/C++. Why use anything else?
Another thought: scientific libraries. If you need external services/algorithms then your chosen language should support the libraries you need. C/C++ are well served by many fast machine learning libs such as FANN, LIBSVM, OpenCV, not to mention CBLAS, LinPACK, etc.
And you're claiming the work is invalid because you're unimpressed by the lack of pubs of a new research program. At Stanford?
In short, the Wired article is interesting while your criticism adds nothing. My advice, FWIW: if you must criticize, be specific. Don't gainsay with, "Your work is uninteresting because I'm unconvinced."
That makes you sound like a Creationist.
CompTop: Applied and Computational Algebraic Topology
http://comptop.stanford.edu/
You need to do more than, "Google: I feel lucky".
Why start small?
Treason is a rare and essentially non-legal label, normally used only by political blowhards like Senator Joseph McCarthy. And now Melissa Mayer.
I suspect Mayer is trying to make Yahoo look good after they sniveled, cowered, and cringed their way into the NSA's good graces by handing over every imaginable user confidence. While joyously waving the flag, of course.
It's OK though. 'Cause she's a PATRIOT.
It's tough enough reading PDFs on a full iPad *with* a retina display. On a smaller form factor like a Mini combined with its lower resolution -- fagetaboudit.
And don't believe for a minute that non-PDF textbooks are an option. Books with equations, graphics, tables, or color render quite poorly and inconsistently as ebooks.
Clearly this school plans to graduate only readers of plaintext fare like novels and poetry. And in that case, why not use cheaper B&W Kindles or Nooks?
The blog does NOT say the son searched for instructions on how to build a bomb. Here it is:
"
Most of it was innocent enough. I had researched pressure cookers. My husband was looking for a backpack. And maybe in another time those two things together would have seemed innocuous, but we are in âoethese timesâ now. And in these times, when things like the Boston bombing happen, you spend a lot of time on the internet reading about it and, if you are my exceedingly curious news junkie of a twenty-year-old son, you click a lot of links when you read the myriad of stories. You might just read a CNN piece about how bomb making instructions are readily available on the internet and you will in all probability, if you are that kid, click the link provided.
Which might not raise any red flags. Because who wasnâ(TM)t reading those stories? Who wasnâ(TM)t clicking those links? But my son's reading habits combined with my search for a pressure cooker and my husbandâ(TM)s search for a backpack set off an alarm of sorts at the joint terrorism task force headquarters.
Thatâ(TM)s how I imagine it played out, anyhow. Lots of bells and whistles and a crowd of task force workers huddled around a computer screen looking at our Google history.
"
She assumes her son could have clicked on a link. But she does *not* say he did, contrary to your claim.
The subplot about the son is missing not just from the summary, but from The Atlantic article as well.
Where did you get this quote? Or are you just trolling?
If the schools think using cameras to automatically identify students is going to be cheaper or work better... I doubt it. Students in hoodies, hats, or sunglasses are going to play hell with face recognition software, not to mention partial occlusions, bad angles, shadows, and poor illumination. That means many (most?) kids will go unrecognized or misrecognized, and miscounted.
Add to that the inevitable bright idea that they also look for unfamiliar faces and then sound an alert when a stranger is seen lurking on the premesis... I foresee many false alarms.
Are these school administrators dimwits? Don't they first *test* a cool new idea before adopting it? Or maybe in Texas they can afford to waste taxpayer megabucks...
Soulskill may well be right. This is NOT your everyday intel analyst:
" ...
As part of a pioneering IT function, you will be a key participant in the design and delivery of a variety of strategic programmes and business critical initiatives within a dynamic, challenging workplace.
The role
You will provide tactical solutions and operational support to business users of information exploitation systems. You will create new data ingest processes, advise team members about their design and implementation and support data management processes within our organisation.
"
I've worked inside several US intel orgs, but I've never seen a spy role where you "support business users" to better exploit THEIR data.
This *is* weird.
For 8 years I worked as a "Compuational Scientist" at 3 university supercomputing centers, helping students and faculty to use/program parallel computers. I don't have a PhD.
I saw several kinds of computing staff in academia. Some assist in research; most don't. They are:
- graduate students (CS, engineering, physical sciences, psychology, medical research, etc)
- post docs (w/ same areas of research)
- IT support staff (cluster or workstation or network admins, univ administrative app programmers, etc)
- research support staff (usu. project-based, paid for by a professor / lab, lasts as long as the contract, usually 1-3 years)
The exception to the above roles is the permanent staff research programmer (at large specialized computing centers or univs with large labs within engineering departments). These are funded for longer cycles (usually 4-5 years), and have numerous executive staff who do nothing but maintain the flow of funds. Examples: NCSA, Texas Advanced Computing Center, San Diego Supercomputing Center, etc.
Many folks do make a living as an academic developer, but they tend to have a specialized skills (HPC / scientific computing, statistical analysis / data mining, data visualization, engineering, etc). Most such jobs do not require a PhD, but univ employers prefer to hire as advanced a degree as possible, especially if the research/work is likely to lead to academic papers. If you can't do that, your role will be marginalized (subordinate to the university's main mission of granting degrees and writing papers).
The ideal research programmer can perform like a research scientist -- find their own funding, write their own research proposals, collaborate with other people's research projects, etc. Sometimes this work extends to US gov't SBIR projects (small business investigative research), which often bleeds over into US military contract work. If you can help to write your own ticket, you can remain working in academia indefinitely. If not, you will need to look for a new research position every 2-4 years.
Personally, I think the two best reasons to get an academic IT job are: 1) to be on-campus for just long enough to finish a grad degree, or 2) to settle into a low-stress long-term IT support job that offers a relaxed lifestyles and good benefits. From my experience, academic support jobs are not especially challenging or rewarding, and academic research jobs are generally narrow-focused and short-lived.