back to article The dev-astating truth: What's left to develop? Send in the machines

Historian Francis Fukuyama in 1992 reckoned with the fall of the Iron Curtain and the replacement of Communist systems behind it with liberal democracies, we had reached the end of history. Can we say the same about software development? Has the arrival of the age of Agile meant that we can now talk about a similar full-stop? …

  1. Steve Button Silver badge

    The problem with predictions...

    Is that they are really hard to make. Especially about the future.

    Yes, of course AI will write all software at some time in the future, but not before most of us have retired.

    1. Anonymous Coward
      Anonymous Coward

      Re: The problem with predictions...

      Yes, of course AI will write all software at some time in the future, but not before most of us have retired.

      "...but not before most of us have been annihilated by SkyNet"

      FTFY.

  2. Tim 11

    Is this article really saying anything?

    If you define agile as "everything that happens from now on in software engineering" (which is what it seems the author is trying to do) then yes, by definition, nothing will come after it, just the same as MS is trying to do with Windows 10.

    However, that doesn't mean things won't continue to progress. There will continue to be both evolution and revolution, as there will in all aspects of life, whatever terminology you use to define them.

  3. DrStrangeLug

    Trying to get rid of those nasty expensive devs again.

    You see this kind of article every ten years and it always boils down to "in the future you technology X will save you from having to spend that money on expensive developers." It NEVER happens.

    Remember in the mid 90's when visual programming languages would mean you could write software without the need to code ?

    Remember being told that all those cheap offshore developers were going to take your jobs and there would be almost no domestic developer jobs in the 2010-2020 timeline ?

    1. FelixReg

      Re: Trying to get rid of those nasty expensive devs again.

      Yep, DrStrangeLug. Scan through Datamation magazine in the late '50's, early '60's era. You'll find an article telling managers COBOL is going to let managers program the computer directly without needing those odd-ball, troublesome, different-from-you programmers.

      The article was right! Just try to distinguish a post-50's corporate data processing programmer from a '50's manager in thoughts, understanding and attitude.

  4. Anonymous Coward
    Anonymous Coward

    Strange as it may seem it would be a great advantage if those developing software actually got most of the bugs and glitches out of it before inflicting it on users and expecting them to be the beta testers.

    Win10 anyone?

    1. EarthDog

      Most of the defects I have seen has been from the intense pressure to load up software w/ features, which is sales and marketing driving. Quality is always sacrificed for perceived capabilities , which is often sold as "slideware".

  5. yogiatdev

    DevOps the holy grail ?

    the question about 'devops' is, whether this concept i applicable to a situation at all. what about a COTS support-project that is supposed to go agile? YES, they can do that - just that they'd better use agile-kanban and not agile-scrum (as the differentiator is, that they're event driven in the work and don't have (too much of) a planned-feature list to be added to a backlog).

    but does it make sense that they really go 'devops'? i dare to doubt that ...

    so please do a reality check! to do 'devops' just should have a 'dev' component - right ? and if you don't have little to any 'dev' in your project?

    so, like usual: there's quite a hype factor around 'devops' which must not prevent you from questioning whether it's the right approach for the given problem

    1. Anonymous Coward
      Anonymous Coward

      Re: DevOps the holy grail ?

      Under Betteridge's law of headlines, applied here to a title of a comment, I declare...

      ....no.

      As soon as DevOps snake oil peddlers get all the money they can there will be a new! improved! way of doing things. Just keep reading.

  6. Unep Eurobats
    Terminator

    Unep Eurobats has already outsourced his El Reg comments to a bot

    I'm much wittier than he ever was.

    1. horse of a different color
      Terminator

      Re: Unep Eurobats has already outsourced his El Reg comments to a bot

      The Register is just a very elaborate Turing test.

  7. Kubla Cant

    Strange article

    I've been a software developer for over 30 years, and I've worked in agile* development for at least the past 5 years. So this article should be about stuff I'm very familiar with, but that's not how it reads. It's like when you're questioned about development skills and practices by recruitment agents - they talk the talk, but there's something funny about their gait.

    * Or should that be Agile?

    1. Rafael 1

      Re: Or should that be Agile?

      I was being agile before it was cool -- I was quite proficient as hiding and running away from managers and users with suggestions "that could easily be implemented in your free time". They never caught me.

  8. Mike Shepherd
    Meh

    "According to some observers, there's still some way to go"

    If cars were like computers, we'd still have to "get out and get under" almost every day. We'd spend half our evenings applying the latest compulsory updates and fixes, but still wonder what difference they made.

    The notion that the computer industry is approaching maturity is about as comical as it gets. Maybe it's at the stage of the automobile sometime in the 1930s. There simply isn't a comparable approach to quality (achieving a product that works and doesn't need tinkering throughout its life).

    1. Concrete Gannet

      Re: "According to some observers, there's still some way to go"

      Yes, we're nowhere near maturity, but you miss the scale of the problem we're trying to solve.

      You don't want to *change* important subsystems in your car almost every day.

      In contrast, software is soft. We implement software systems precisely so they can be changed. The right analogy is not your car, but trying to reconstruct the plane while it's still flying (https://www.youtube.com/watch?v=L2zqTYgcpfg).

      This is an ambitious goal, and it's not at all surprising that we haven't achieved it. What is interesting is the progress we've made.

      1. Mike Shepherd
        Meh

        Re: "According to some observers, there's still some way to go"

        I'm not missing the scale of the problem we're trying to solve. I'm pointing out that the problem arises from a poor approach to quality control. I don't want to change important subsystems in my computer every day either: it's the suppliers who want to do that.

        You say that "We implement software systems precisely so they can be changed". This is like saying "My house is built of bricks, so it can survive a tough environment", then firing mortars at it to get your money's worth.

        1. Chris 155

          Re: "According to some observers, there's still some way to go"

          Except your analogy is completely wrong on both counts.

          You change things in your computer all the time. You install new software, use it for new tasks, connect it to new devices. You and everyone else does this all the time. There are millions upon millions of different combinations. If you spent half as much time making modifications to your car it would break down all the time too. You may not understand what Microsoft is changing under the hood in the latest update, but you sure want the thing that uses it to work first time every time and fast.

          Your mortar analogy is also bad, because unlike your house, your computer literally is under attack. All the time. In ever changing ways.

          If you provided a list of what your computer had to do, and you never wanted to change any of that, and you disconnected it from the internet you could build a system that doesn't crash. A custom OS that only worked on that hardware, and only worked for that list of tasks and didn't have to deal with being attacked. You don't want that though, you want an OS you can install onto any hardware, and run any software on, and connect to the internet.

    2. Vic

      Re: "According to some observers, there's still some way to go"

      There simply isn't a comparable approach to quality (achieving a product that works and doesn't need tinkering throughout its life).

      On the contrary - embedded development has been like that for years.

      Unfortunately, the up-front cost of that sort of work is much higher - so the bean-counters go for the cheaper option. And that means lots of re-work, because the "cheaper" developers aren't doing the same job.

      The end result is that the product goes to market in a less-robust state. It takes longer to get it to a good state - meaning a huge loss of customer confidence in the meantime - and ultimately costs more to boot. But the management who put it all into action will congratulate themselves on having "saved" money on developers, and so award themselves big bonuses. Trebles all round!

      Vic.

    3. EarthDog

      Re: "According to some observers, there's still some way to go"

      Who are these observer? How did they reach their conclusions? Or is the author just a BS factory?

  9. Chris Evans

    "The Last Program" you will ever need.

    There was a program called something like "The Last Program", it was certainly advertised as the last program you would ever need. This was about 25 years ago....

    Historian Francis Fukuyama is looking a bit foolish now and I think rumours of DevOps demise are exaggerated.

    1. Anonymous Coward
      Anonymous Coward

      Re: "The Last Program" you will ever need.

      "There was a program called something like "The Last Program", it was certainly advertised as the last program you would ever need. This was about 25 years ago...."

      ... it was called "The Last One" ... and I seem to recall much merriment when they subsequently released "The Last One V2"

    2. Captain DaFt

      Re: "The Last Program" you will ever need.

      -There was a program called something like "The Last Program", it was certainly advertised as the last program you would ever need.-

      Interesting. Was there a version that ran on The computer that will never become obsolete?

  10. allthecoolshortnamesweretaken

    "Historian Francis Fukuyama in 1992 reckoned with the fall of the Iron Curtain and the replacement of Communist systems behind it with liberal democracies, we had reached the end of history."

    Yes. And he was obviously wrong.

    Great start for what is essentially a sales pitch.

  11. hellwig

    Status Quo?

    "It really matters how good your developers are and the best developers are so, so much better than average."

    So, 1% of the population will do the work and 99% will sit around with their thumbs up their asses? This is a useless summation. We have to make use of everyone somehow, and there are only so many toilets that need scrubbing. If the solution does not work with your "average" person, it's a sh*t solution. We'd all like to be Google, hiring only the best and the brightest, running Ph.D students through embarrassing internships just to see who's smart enough yet still easily manipulated enough, but for most other companies, you hire what you can. If agile only applies to the geniocracy, then by definition it can NOT be applied everywhere, and it has failed.

  12. MD Rackham

    Fads Come, Fads Go

    Will the latest fad be the last one? (said breathlessly)

    No.

    Some of us remember when "goto-less programming" was going to fix everything. Sigh....

  13. Concrete Gannet

    Four bullet points is not a framework

    "There's even a question as to whether Agile is a framework in its own right..."

    It is not, no question.

    The Agile Manifesto (http://agilemanifesto.org/) that started it all has four bullet points and talks about principles, not practices.

    Agile is a very broad church. There are many more prescriptive processes like XP and Scrum that add more detail and might be regarded as a framework.

  14. Aging Hippy

    Future in computing

    "There's no future in computing now they can write their own programs" - Quote from my school careers adviser referring to compilers - 1966.

  15. PassiveSmoking

    Yawn

    New day, same old crap.

    I've been hearing predictions about how programming is going to cease to be an intellectual pursuit for years. Either it's going to be replaced with a load of pre-build modules that any idiot can assemble into some kind of super-uber-application (*cough* BPEL! *cough*) or it's all getting outsourced to India and now we're all going to be replaced by robots.

    None of this has happened (except for some outsourcing to India for stuff that doesn't really matter that much, because most code produced at outsourced software houses is crap). BPEL and other such attempts at lego brick programming have all been abysmal failures (Seriously? A programming language implemented in XML? Couldn't anybody working on that see that it was an idea that should have been drowned at birth?) and I doubt AI is going to be any threat to my job so long as Siri can't tell "Play the song Ruby" from "Play with your boobies" (true story, Siri really did think I said the latter when I asked it to do the former once).

  16. John Smith 19 Gold badge
    Unhappy

    I also recall "The Programmers Apprentice" project out of stanford IIRC.

    The developers said it was buggy and clumsy but it showed (sort of) what could be done.

    What I never understood was if that was the case why didn't they run a copy of it through itself and have the system make itself better ?

  17. MatsSvensson

    What the hell does the article have to do with the headline????

  18. T.a.f.T.

    0 bugz

    I remember reading something on the JPL's Laboratory for Reliable Software (LaRS) which suggested we were in the hunter gatherer or early metalworking (bronze age vs. composite material design) stages of programming. Agile is like the idea of crop rotation and organised teams of farmers. At some point in the future we will work out something like artificial fertilizers, pesticides and automated systems that would give us hundreds of times the output per-worker with much lower error rates. It took us a few hundred years to get finance as far as it has come today (which is still a bit error prone at times) so it seems reasonable to expect other professions like programming to take a century or two to find their stride.

    TLDR; things are getting more productive but they will probably get hugely more productive and not be that recognizable to us as programming any more.

    1. Vic

      Re: 0 bugz

      TLDR; things are getting more productive

      I'm really not sure that's true...

      Vic.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like