In the beginning there was dBase II, designated “II” by Ashton-Tate, its publisher, to convey a level of maturity beyond its actual virtues. It was followed in quick succession by Paradox, Delphi, and Microsoft Access, all of which overcame most of dBase II’s (and III’s, and especially IV’s) numerous limitations.
Compared to traditional programming languages these platforms increased developer productivity by approximately 10,000% compared to traditional COBOL coding – they let me get about a day’s worth of COBOL coding in five minutes or so.
This history was current events more than twenty years ago and yet IT shops still write code and enshrine the practice with various methodologies (Scrum, Kanban, DevOps, add-your-favorite-here) intended to improve IT’s overall app dev effectiveness.
Speaking of deja vu, the pundits who track such things write about no-code/low-code (NC/LC) development environments as if they’re something new and different – vuja de, like nothing they’ve seen before – when in fact they offer little their 1990s-vintage predecessors weren’t capable of way back when.
Should NCLC be in your future? Gartner says yes, predicting that by 2024, “… low-code application development will be responsible for more than 65% of application development activity.”
They make it so easy … to nitpick, is that 65% of all lines of code that will be produced using No Code tools? Probably not, as No Code tools by definition produce no code.
Function points? Maybe, except that nobody uses function points any more.
Probably, Gartner means 65% of all developer hours will be spent using NC/LC tools.
Which is simply wrong, on the grounds that most IT shops license when they can and only build when they have to. In my unscientific experience, looking at total application functionality as the metric, maybe 75% comes from COTS implementations (commercial, off-the-shelf software, which includes but isn’t limited to SaaS packages). Maybe 25% comes from in-house-developed custom applications, and that’s being generous.
As NC/LC platforms don’t touch COTS/SaaS functionality, it’s doubtful that work on 25% of the application portfolio can occupy 65% of all developer hours.
But I digress. The question isn’t whether Gartner has done it again. The question is how much attention IT should pay to this platform category.
Answer: If coding and unit testing are enough of a development bottleneck to care about, then yes. When it comes to optimizing any function, removing bottlenecks is generally a good idea.
Second answer: If in your company DIY application development is a source of a lot of application functionality, then selecting an NC/LC standard, integrating it with your application portfolio’s systems-of-record APIs, and providing training in its use will save everyone involved from a lot of headaches, while removing a source of friction and conflict between IT and the rest of the business.
Third answer: Most COTS/SaaS applications have some sort of no-code or low-code toolkit built into them. These should be IT’s starting point for moving in the NC/LC direction, and for that matter for any new application functionality.
Bob’s last word: It’s easy to fall into the trap of answering the question someone asks. “Are NC/LC tools useful and ready for prime time?” is an example, and shows why Dr. Yeahbut makes frequent appearances in this space.
The answer to the question is, in fact, “Yeah, but.” NC/LC development should, I think, be part of the IT app dev toolkit. But mastering the tools needed to integrate, configure, enhance, and extend the company’s COTS application suites has, for most IT organizations, far more impact on overall IT app dev effectiveness than anything in the way of app dev tools.
Bob’s sales pitch: As a member of the KJR community you might enjoy my most recent contribution to CIO.com, and a podcast I was interviewed for.
The CIO.com article is titled “The hard truth about business-IT alignment.” You’ll find it here.
The interview was for Greg Mader’s Open and Resilient podcast and covered a number of KJR sorts of topics. You’ll find it here.
My experience with many earlier tools purporting to be “low code” is that yes, you can get something built that will (mostly) do the job. BUT, the resulting application is liable to be extremely bloated, AND limited to whatever the providers of the tool thought might be needed.
Can you get something that works? Of course. Do we care that it’s bloated? In these days of unlimited RAM and storage, probably not. Can it cover the bulk of the feature set? Usually.
But if your use case is unusual, or even break-through, will you be able to get everything you expect using one of these tools? I will lay down even money that in the end, someone somewhere is going to have to write code.
My $0.02.
Great Podcast appearance, if I do say so myself!