Original URL: http://www.theregister.co.uk/2013/05/01/indemand_ancient_it_skills/

Is the IT industry short on Cobolers? This could be your lucky day

Sometimes a CV needs a few fossils

By Dominic Connor

Posted in Jobs, 1st May 2013 08:37 GMT

Let's make one thing clear: your previous jobs are not the reason why you were hired. You were hired for having skills that bosses need.

People are employed because they are needed to do things that must be done, not because they can do something that is merely desired.

It’s not all bad news. The current Big Data hype means firms are reopening mature stable systems to suck in data. In this case, a “mature” system means that most, if not all, of the programmers who wrote it are long gone, so bosses often need archaeological skills to work out why the hideous old Oracle 3 and VB6 system behaves in a way that mystifies developers who’ve used versions that belong to this century.

Also, headcount freezes and attrition means that the support of some systems has dropped below a level that even Capita would find unacceptable.

What this all adds up to, is that some of the motley collection of skills that you’ve picked up over the years can be more valuable than you think. If you spin them right.

Your CV is wrong

First things first, drop the idea there is a perfect CV, even if you avoid the stupid mistakes I list here. You’re shooting in the dark and taking the time to read this article, which implies you’re not hitting as often as you would like. That means you need a Plan D. Plan A might be Python on top of MongoDB, B might be to go back to the bad old days of using Oracle, and C is to write SQL for whatever people will pay you to drive. Plan D is a shameless pitch to elbow your way in doing things other can’t or won’t do.

I suppose you can try to punt the idea that you’re smart enough to pick any API in a couple of days and since all programming languages either look like Lisp or C none really scare you. But realistically that won’t get your mortgage paid because you already know that the whole recruitment process fixates on having the latest version of the fashionable tech buzzwords and of course every single programmer on the planet is an Extreme Agile Top Down Business Oriented developer. Trust me on this because I read CVs for a living. We both know that when you say your last job was fiveyears where you used Hadoop and MapReduce, the reality is that for over four years you were the last remaining REXX guy, tending an old, critical system on life support. But by not mentioning it you are betting that a casual reader might think you only did the sexy skills. I have to admit that this spin can work, but at the price of that is missing jobs where your fossil skills are wanted.

Upgrade your pandering

A big mistake is only to have one CV that panders to this defective process, you need to pander to it in a more cynical way. Recruiters use keyword matching systems, so when a client says they want someone who knows Excel 4 macros (still in use) that’s what goes into the search field. Nor am I going to kid you that my firm doesn’t do the same - we just have a better search engine, one that lets me type in regular expressions and has a “sort of like” operator.

What no database can tell a recruiter is what you haven’t bothered to tell it. Please don’t tell me that you’re shocked when I say that job ads are spun to make the job look good, so there is a toxic symmetry where they don’t say that they need to migrate from Delphi and you don’t tell them that you know it. Whatever the skills needed, if the process is working properly, then you’ll be competing with people with roughly similar experience and sharing a fossil makes you stand out. Or not; you can’t know which of the jobs your CV gets sent to actually need your legacy skill which is why you need more than one. Fortunately there are still a lot of agencies so you can send a usefully different variant to each one.

That means you need to have multiple CVs, some with bright shiny leading edge buzzwords easily identified out from job ads, but others that include some of the things you think that no one wants.

Scripting and sucking

A good way to present fossil skills is with a “Migration and Upgrade” section that lists them in comma separated form so that the buzzword scanners spot you and pass it to a human (or at least a recruiter) who will check them out. A decent recruiter will contact you before firing your CV at his client and you should seize the chance to expand and explain the fossil skill because otherwise you’ll be claiming CICS skills but without it being referenced in a specific job and of course you can delete the non-relevant fossils to make space. This is more work for you and I’ve had pushback from candidates who seemed to think the whole sales pitch is my problem. It’s not. You want the best job you can get. I want the cash. Play that game or lose.

A lot of migration work is reading and reformatting data, extracting business logic from old code and often generating something that’s functionally pretty much identical, but in the new framework. Ideally that should be automated, which means scripting in Python or Perl plus Regex can make you a better candidate and since you produce lots out output there is good visible productivity which makes you look better once you get the job. There’s also MS SQL Server’s Data Transformation Services which although obviously designed for importing to MS SQL is good as a staging post. One very large client was hugely impressed by me using DTS to deal with a sick old Oracle database, it’s also pathetically easy to learn and tack on to your fossil skill set.

As you saw in my piece on expert witnessing, there is a useful stream of billable hours looking at what people have done to important systems and working out where the code came from. To get that you do need both depth and the ability to explain to lawyers (and occasional outraged management) without resorting to explaining the guts of a needlessly complex derivatives order matching system as “like your car”.

COBOL, forgotten but not gone

Recently those nice people at Micro Focus wrote to me with research about a skills shortage in Cobol. Perhaps they hadn’t read my article where I spat scorn on the idea that there could ever be a skill shortage, rather than a shortage of employers willing to pay the market rate which is high considering how the age of the skill, but pretty average if you’re using it to pay your bills.

But there is a lot of Cobol out there and although I don’t believe their ambitious claim that 65 per cent of currently working code is in Cobol, brutal Darwinism means that the large blocks that have survived are deeply resistant to being replaced. Part of that is the surprisingly short tenure of CIOs who realise that the payback time for replacing the bottom of their house of IT cards is longer than they expect to stay in the job and there’s no sexy visible output to impress golf playing execs. They don’t want or expect you to make major changes. In fact the work is mostly to prevent bad things. Talking about re-architecting is a high risk strategy but if you pull it off then you have a job for your life and that of your children.

I hear that a couple of UK universities teach Cobol, which I suspect is the right number for the graduates, though maybe not for the employers since it keeps the supply/demand favouring developers. Still, I do worry about what they’ll do as the shadow of Cobol recedes from our machines, but the impression I get from Micro Focus is that ¼ of US universities teach it, which makes the UK schools use of Pascal look almost rational.

COBOL still runs a scary percentage of the world’s financial infrastructure and - along with CICS, JCL and other mainframe generation tech - is to be found buried under several generations of 80s then 90s then 2000’s tech. There is a small subset of you reading this for whom getting better Cobol skills is rational. You’re the ones in your 50s. MicroFocus reckon Cobol has upwards of 10 years of undeath left and that fits what I see, combined with the fact that so many Cobolers wlil have retired over that period, pay may even go up, for a while.

A lot of people laud Admiral Grace Hopper for being the first important woman in computing, but they are strangely quiet about her leading role in the creation of Cobol, which IBM at first tried to kill for good reason. Strictly speaking Cobol’s longtime buddy DB2 is not a legacy skill since there are still new installations up so it is probably the oldest non-fossil skill and will remain. Moving from CICS or DB2 is like an organ transplant, such is the risk and pain involved. Indeed you can expect most core databases to outlive the company that build them.

Don’t learn Delphi

There will never be any shortage of Delphi/Pascal programmers because of all the legacy technologies Pascal is the equivalent of inheriting your grandad’s black and white 60s girlie mags. This is because AQA, the largest exam board not only encourages this silliness, but won’t accept A level coursework in languages like C and actually dropped C#. They are cheered on by the British Computer Society with pseudo academic “papers” that attempt to justify pushing a language that stopped being a rational career choice before some IT teachers were even born. So, like everything else that’s worthless there will never be a shortage of Pascal programmers.

Fractional skills

So far I’ve covered languages which inevitably are mass market, but if you’ve been in the game long enough to have legacy skills then you’ve got APIs, OSes and frameworks like OS/2, CICS, Win32, ODBC and MFC. This is another reason to develop multiple CVs since you need to present a consistent story for each silo, don’t blur a clear CICS/Cobol picture with Pascal or Adabas/Natural or vice versa. Ironically this may mean you need to refresh or even build skills that turn you into a complete person to fit a requirement and those nice people at MicroFocus even have a freebie version to do that with.

Industry-standard lies

Be clear about the game I’m coaching you to play here. These are short to mid term tactics to keep you solvent, so unless you’re just looking for a few last paydays before you retire you need to lever yourself into more modern and portable skills.

Management may say that they’ll retrain you in the new system but there is a reason this section is called “Industry Standard Lies” and since you’re betting your career and they’re more interested in budgets and delivery times you have to manage your risks. Many managers genuinely believe you don’t notice that as one Cobol guy ex-Merill Lynch put it, “after 18 years they dumped me like a bag of sh*t”. I know 18 years sounds like forever to some of you, but starting mid 20s means this is mid-40s, not a good time to be dumped like a bag of anything. It’s pretty much impossible to spot this from the outside, BAML (as they like to call themselves) have pious HR statements about developing people which bear as much relation to reality as some of their pricing models for MBS did, conversely I was quite shocked when ITPros at the generally evil EDS told me that career development was actually rather good.

There is a persistent myth that ITPros aren’t interested in “the business”, yet 30 years of doing IT and bossing ITPros around tells me the complete opposite and the odds are that you’ve picked up a good understanding of how money is made in your business and the processes that support it. The reason for the myth is that most ITPros don’t articulate their understanding enough to decision makers and so when their specialist skill is no longer needed, they get dumped. That means the time you are buying using specialist legacy skills can be used to reposition yourself as a business analyst but one who can give a better impression of a real ITPro than most.

How to tell your career is doomed

If you meet me and I ask “how’s your VBA ?”, know now that I’m politely saying that your career is holed below the waterline. As a dialect of VB6 my learned colleague Verity Stob points out that it’s the worst language in the world, but our good friends at Microsoft threw us a lifeline a few years by releasing a version of Excel that’s wildly different from the one we’ve come to know and loathe.

As Vista was a dog and Win8 is not fit for any business purpose I can imagine, corporates are now grasping the nettle of migrating from the XP generations of Office and this will not only break many surprisingly critical worksheets but it needs testing. This project work is mildly well paid and you may be able to elbow your way into replacing some of it with C# or VB.Net solutions.

You’re next

I was the first person to write a VB app in the UK (I can tell you’re impressed), helping found the VB User group and writing about it. For me VB is the big tool I’ve seen through a life cycle from hot skill to legacy, often pushed out by Java. It served as a good escape route for PL/1, Fortran, and Cobol programmers in the 1990s before some got sucked back for the well paid Millennium Bug gig.

Your buzzwords will be different, but the pattern is likely to be the same especially if you’re in Java which is heading towards huge over-supply even if there’s still a lot of code being written, remember it is supply/demand that drives your pay, not the absolute number of jobs.

The Millennium bug was unique only in scale. Lots of code uses base dates that can wrap or crash through short sighted use of short data types and for a good number of the systems I’ve worked on if you told me that it would be running in 2013, we’d have laughed out loud and as you read this people are writing code that someone is going to be rewriting in 2050.

And finally, it's always worth remembering Citizen Robespierre's observation that "graveyards are full of men once called ‘indispensable’". ®

Dominic Connor is not just a City headhunter: he wrote some of the legacy VB, C, XLL, C++ and SQL code that’s paying your mortgage, so be grateful.