The software industry will dissolve into a soup of micro-detailed web services delivered over the cloud by 2022, with IT departments reduced to “guiding” users to prevent them from leaking their companies’ crown jewels onto the net. That was the extremist version of the vision sketched out by a panel considering “The Software …
Plus Ca Change
Well I don't see that this is going to kill off either development or product consultancy due to the rules of reusability.
When something is sufficiently capable as to be widely useful it requires complex configuration to work exactly how you want it to (hence product consultancy).
When a group of things are sufficiently simple to be understandable then a number of them must be integrated to do anything of value (hence bespoke development).
There are advantages and disadvantages to either approach so neither is likely to be universally applicable and so the train rumbles on.
Just like all other forms of engineering
We won't software engineers in the same way we don't need structural engineers, mechanical engineers, etc, etc, because all the big problems have been solved... right?
How on earth do these pointless committees even get off the ground, let alone find anyone to publish its ramblings?
What if "the cloud" is just a fad?
Basically a "cloud" is a very similar environment to a mainframe batch operation of years gone by. You submitted your "job", something, somewhere did something with it and produced your results. The person who initiated all this had little or no control (JCL notwithstanding) over the process.
While this sort of set-up provided a solution, like the cloud, it wasn't very flexible and like the cloud, the person who wanted the work done would often want a little more control - or assurance - over the nuts'n'bolts of the process.
As a consequence, it's easy to see that the huge datacentres that house "cloud" service providers these days are analogous to the manframe operations of yore. It also follows that in the IT world, nothing lasts forever - so what we see as a cloud-based solution today will be seen as a cloud-based problem, tomorrow.
So if we're looking forwards 10 years, sure; there WILL be cloud operations, but there will also be other ways to do thing. Ways that haven't yet been invented (just like cloud computing didn't happen in 2002). What they will be is difficult to say, but if the cycle keeps spinning round, I'd guess that the users would be emerging from the remains of cloud-based architectures and wanting their own systems to run their own applications in their own way.
What was old is new again...
[...] just like cloud computing didn't happen in 2002 [...]
But we did have "cloud computing" in 2002...and earlier, too. We just didn't call it that...we used much more utilitarian terms like "thin-client computing" or "distributed processing". We just hadn't evolve such a shiny, shiny name fore it yet.
Re: What was old is new again...
And before that "client/server". And before that "service bureau". Yes, those things are not all the same, but they're all similar to "cloud computing" in various ways. (In fact, the "cloud" model is closest to the service bureau - it's just the utility version of it.)
People who predict the death of anything in the IT industry are nearly always wrong, or at least greatly underestimate when old technology will cease to be prominent. FORTRAN and COBOL applications are still going strong, and much of the source code in use is from older versions of those languages. Timesharing did not spell the death of batch processing (indeed it's seeing a resurgence, thanks to Hadoop and the like). PCs did not kill mainframes. Linux did not kill proprietary OSes. Disk and solid-state storage haven't killed tape. Document imaging hasn't killed paper.
Of course, knowing a bit of history hasn't wiped out these idiot pundits who predict the Death of X every few weeks.
There's No Silver Bullet
[John Harris] said:
“The accessibility of platforms means that software development is no longer such a specialist activity.”
I dunno what he means by "accessibility", but Fred Brooks pretty much shot the "silver bullet" theory of software development several decades ago, but I guess this idiot subscribes to the theory that anything learned more than a few years ago doesn't apply to him and his brand of snake oil.
Software development is "hard" in the same way that differential equations are hard. That is, if you have a particular mindset and enough training then actually they aren't, but very few people have that combination so in practice it remains a specialist activity.
Thats not all...
Microsoft will release Windows under GNU
VMWare will release Virtual Kitchens, allowing my wife to cook/clean from her office desk
Pigs will be genetically engineered with A.N.a.L Thrust to help them achieve flight
The WWW will be available in print
AI will finally allow HAL to become reality and claim an Oscar for his role in 2001
echo "Lots of tiny focused sas apps eh? I suppose we could pipe them together in some manner for custom solutions, oh my this all sounds a bit familiar." | /bin/mail -s "everything old is new again" "firstname.lastname@example.org"
Back in the real world: A new focused app rolls downhill, gathering users and cruft as it goes until it settles into an expensive unmanageable swamp that no one likes to navigate. There's plenty gravy train for software devs for the foreseeable future.
"Consultant - meet Jack Shitt"
"Jack Shitt - meet consultant"
There we are.. consultants definitely know jack shit.
"Oh My" - do people really use that phrase any more? Even in America? I've seen it several times recently in Reg comments. I'm sure the last time I heard it anywhere else was Dorothy in the Wizard of Oz.
Hazaah! A report of quality.
It must have come from Kensington. Kensington? The Royal Borough? Up top?
I smell overpriced researchers justifying their position.
I remember when assembler died, and my project manager saying. Coding is so fast now, we will need less of you lot.
Badah! Mr Babbage says no. Report is bollocks.
Abstraction and Encapsulation
Abstraction and Encapsulation are what we do, right? All the time? Repeatedly, recursively.
What is SaaS offering:
* "Big" components that have teams supporting them
* Processing power
* Lots of network latency :) (aka. universal accessibility from any device)
In how many scenarios do you actually need those first three things? Not very often. Only the fourth seems universally relevant, but I could just as easily imagine your own personal infrastructure providing universal access in a self-organised P2P style.
The only way I can see it becoming mainstream for development is if the components/functionality being offered as *radically* better than you can get in an off the shelf library, and it's *cheaper* - which in the case of FOSS is going to be difficult.
I mean, I get the "dream", but I'm still missing why anybody thinks it (whatever it wants to market itself as today) is going to extend beyond being a Channel to serve consumer/business apps (which will still be developed the "traditional" way).
Re: "The Last One"
Exactly what I thought when I read the article. Every decade produces a magic solution that's going to put an end to software engineering. TLO, 4GL, RAD. The cloud is mainly different because it has a name instead of a TLA. It's still the same old snake oil.
I'm sure a large part of the problem is that "consultants" like this don't have proper jobs, so they're unaware of the thousands of vital business applications that companies rely on. They think "software" means Facebook.
Most of the devs I know (including me) spend far less time developing and far more trying to navigate a path through miles of red tape, attending numerous (often pointless) meetings, filling in status reports and spreadsheets, multiple copies of timesheets, and poring through (or producing) the vast amounts of project related paperwork (real or virtual) that inhabit most of the available bytes on any server.
"New" things are all well and good, and sometimes do actually save time, but that's only until the business processes catch up with them and envelop them in all this crap.
That's not going to change any time soon - in fact as I get older it seems to get worse - so going by that we'll need more and not less.
Re: Dev time
You have to have lots of meetings. If you took the meetings away, it would be you, doing all the work (which you'd find a solution which is obvious,) and thousands of business analysts, performance testers, project managers, etc.. sitting around looking obviously redundantable.
Consultancies create large numbers of job titles to con their marks into thinking they're getting lots for their money. The government likes that they do this, because otherwise they're are home when they go canvassing.
Where do they think these services people hook together will come from?
There's going to be change, and *maybe* that change will disrupt monolithic software providers. But developers? Someone has to create, maintain and extend the components being discussed that people would access via APIs. Heck, someone has to design the APIs themselves, let alone implement them. Developers and designers will have a place doing just those things.
And I think the claims that "the world" wil be writing tomorrow's software is a bit overblown. The vast majority of people don't want to fiddle with APIs, and use of even highly graphical tools for creating programs has never really taken off.
I don't see the change as all that, personally.
Re: Where do they think these services people hook together will come from?
Is the "world" that will be writing these apps the same "world" that has so thoroughly munged up SuSE12.2 that we don't know when it will ever be released?
This story comes alomg every few years, normally along the lines of drag and drop software, mashup systems etc... will take over the world and BAs/FAs will develop everything, devs will all be out of a job. 15 years ago when I started I was told this, then it was VB and now its SaS.
Is it even remotely possible that these idiots actually believe what they are saying?
Scary thought, isn't it?
SW will just move to a higher level
EEs have been trying to work themselves out of a job for 100 years, by coming up with every faster, more integrated parts. But each improvement increases the utility of the products, so EEs just move to the next higher level f integration. Once it was plugging tubes or transistors together to make a flip-flop. Now it's wiring several million flip-flops (in the form of a CPU chip) together with another several billion flip-flops (in the form of memory chips).
Ah, yes, of course.
>“There are things that can be done. I don’t know what they are,” he said, “but software will be created by the world.”
>As for the IT department, he said: “We won’t be controlling. I think our job will be to guide.”
circa 1985 - "with the massive power of Access and Excel, business users will now be able to define their own business-critical applications, thus obviating the need for specialist IT personnel".
Do we have any history on this groups predictions? What did they predict, 10 years ago, about today?
It would be awful to have to put your name behind such outrageous guesses, and I would guess only drooling crack heads would. Does pcmag not get to contribute to the panel?
Paris; the patron saint of drooling crack heads.
Brilliant comment, and something I do all the time when troubleshooting, check how good they were are predicting where we are now.
I also throw in additional checks, such as "Can you just fix this for me?" and see if they can fix it.
Like the saying goes. "Those who can, do, those who can't, leech."
"Don't go into computing
- there's no future in in" - quote from my school careers advisor in 1966.