Tuesday, January 29, 2008
Recently, I’ve hit the topic of Internet safety and even “reputation defense” pretty hard on a couple of my other blogs. As I reflect upon the then endless days of my mainframe career in the 80s and 90s, and recall that all of today’s problems about prudent behavior online had their foreshadowings in the mainframe world.
Corporate email started coming into use in the 1980s, usually in a CICS region. SYSM was the best known product, and often there was a separate region called SYSMCICS. In those days we had a variety of session management products, like VMan and TPX.
People learned quickly that emails could be forwarded and saved forever, and wrongly conceived emails could have consequences. Back in 1992, I wrote a SYSM about the sharing of dial-up software on diskettes to be taken home for Nightcall, and that generated some resentment. This was in the days that software copyright was just becoming a serious topic.
Right around 2001 or so, companies had big legal debates on the retention of emails. Some companies thought it was safer to discard it. However, in 2004 the federal government started requiring employers to retain email of employees for discovery purposes. A typical resource. Microsoft has a blog on email retention here. "Big I, Little T" has a 2007 article on email discovery practices. I see that I had written about discovery requirements and the TorrentSpy case in June 2007 on another blog, here.
Voicemail came into being in most shops around the late 80s or early 90s with sophisticated systems. Typically, one could play back a message verbally before sending it but then there was no way to save the voice mail messages you had sent.
The mainframe world actually had a form letter-writing application, called Napersoft, and it usually had its own CICS region, too. Batch cycles would generate form letters, filling them in from data pulled off of VSAM files.
In those days, people were told to guard their passwords and to sign off mainframe terminals when they left their desks; they could be responsible if anyone misused their sessions.
CRT 3270 terminals, however, owned their intelligence with “play keys”. People could code their entire signon scripts, including passwords, into the terminals.
Another cultural problem occurred when people started working on terminals all the time to enter documentation. I had saved some notes on a scratch pad in 1981 on a small IBM mainframe, and a librarian picked them up and published them as if they were a legitimate program spec, causing great embarrassment.
Monday, January 21, 2008
Michael A. Fletcher has a story on p A01 of The Washington Post, Jan. 20, "Highly Skilled And Out Of Work: Long-Term Joblessness Spreads in Middle Class," with link here.
He talks about a "CEO recovery" with jobs at the top and the (until recently) the bottom, but not so much in the middle. Of course, I.T. people all know what happened right after 9/11 (it had started at the end of 2000), with steady mainframe jobs (that had grown during the Y2K push) disappearing, and client server and Internet jobs only for the highly specialized. There was this "what's hot, what's not" mentality right around the start of 2002. COBOL was on the not category. But mainframe skills have come back, but always in specialized niches (like MMIS) for professionals who stuck it out in those niches. In 2002 there was a debate as to whether companies would pull back work from overseas and then not be able to find the people in their 50s and even 60s who knew the old mainframe skills. The IRS, for example, still looks for very specialized IBM mainframe assembler programmers. There is little demand overall, so there is little supply. It's like the l'Hopital problem in calculus -- indeterminate.
It's not true that there are no middle level jobs. Everybody talks about the demand for teachers and nurses. This demand is more appropriately filled by young adults or by people making mid-life career changes, with appropriate help; it's more problematic for retirees.
There is a tremendous bias in our economic culture to "get rich" (aka Trump) by selling things, the "we give you the words" mentality. That expectation has spilled over into career advice for retirees. Yet, that may be starting to work less well as in an individualistic world more customers resent being manipulated by sales people when they can take care of themselves.
It looks like we are heading for another 1990 style recession, with the crash in real estate prices and the subprime crisis. It happened in the late 80s in Texas (with the savings and loan debacle) but this time it is much more widespread. There certainly is a tendency for people -- even banks -- not to learn from the mistakes of the relatively recent past.
Still, people in college following technical career paths (IT, medicine or health, security) should be in good shape. In college, the student has an opportunity to look at what employers really need and plan his study accordingly. If he or she is of an activist mentality, there is at least the opportunity to approach the activism "professionally" rather than on Myspace. One can consider social contributions in many areas: engineering (going green, working in developing countries as one distant relative of mine does as a first job from college), law enforcement, intelligence, computer security, knowledge management (maybe the next thing), even the military (where "don't ask don't tell" is still a very serious problem).
For retirees, though, with some accumulated resources to draw on, it's understandable that the approach may be very different. A self-defined program of activism could be very appropriate.
Friday, January 11, 2008
One of the points often made by career coaches in information technology is to “think big.”
It used to be that career stability depended on have very specific hands-on “can do” and “job ready” skills. People who could code, test, actually implement and maintain (and be on call for) production systems were more valuable than middle managers. Particularly central to the computing culture of those days was the nightly and month-end batch cycle, with skills in basic languages (COBOL, ALC), dump analysis, databases, and especially JCL.
That became very much the mentality in the 1980s, until the picture got more complicated with corporate mergers and buyouts and data center mergers. In the 1990s, the picture became complicated with Y2K, particularly with older mainframe systems (like the life insurance industry’s CFO from the 1960s), the emergence of mini-computers (particularly because of defense applications, originally), and then user-defined computing, client server, and the Internet. After 2000 (and the economic pressures that followed 9/11 and the various scandals), many jobs were lost to offshoring. The market became unpredictable, with emphasis on in-depth expertise of many arcane skill sets, some of which would become obsolete.
The neo-hype was that visible career advancement was important. That’s even more apparent now with all the recent talk about “online reputation defense.” The changes are more cultural than legal, and have more to do with the way people (especially employers) get information, in new volumes before they digest its significance and relevance to what they already have.
One kind of job that you heard a lot about was “lead architect.” As work was offshored (and sometimes brought back, as the savings were not always what was expected), people were needed to oversee it. It was more like technical project leadership than management, and sometimes certified project leaders were in demand.
As I look around, I can see some more areas for obvious architectural synergy. For example, consider the issue of digital rights management, and how quickly music and movie companies are finding that their older notions of how to enforce copyrights and protect revenue sources just don’t work in practice, whatever the legal action they take and whatever their victories in court. Electronic Frontier Foundation has been promoting the idea of voluntary collective licensing. The white paper (from April 2004) is “A Better Way Forward: Voluntary Collective Licensing of Music File Sharing,” here. . All the work to develop DRM seems to have been a lot of barking up the wrong tree. But, sifting down through the details of the proposal, one thing comes through: the music and movie industries probably need to do a lot of work to make sure that the artists are paid properly, and automatically.
We see the same concept again with the WGA Writer’s Strike. Much of the conflict is about “royalties” for Internet redistribution of films and video, but one critical component would seem to be just to have the mechanisms in place to pay them promptly, and automatically, and reliably, with EFT deposits, according to predetermined mathematical formulas. From what I see written about these problems, it doesn’t seem that properly automated systems are in place to compensate the artists.
There is a model for this, with the payment of bloggers for ad revenue, with systems like AdSense. The same concept ought to be used in designing systems to pay musicians and writers for residuals. I just wonder if anyone has sat down and thought about putting the pieces together. They call this “systems architecture.” It’s hard to say who would write the systems like this: Google, or maybe companies like EDS, Perot, IBM, or CA. They would be a mix of mainframe, client server, with loads of XML feeds. But it’s easy to see how today’s bitter legal controversies about fair compensation can generate tomorrow’s jobs in IT.
Tuesday, January 01, 2008
Year end processing—that’s what a lot of systems people on call have to deal with today. January 1, New Years Day, and the offices are officially closed, but most systems people in financial institutions carry pagers and don’t get compensated if they have to come in for a problem (unless on W-2 contracts).
In 1990, toward the end of my first year on a “new job” as a regular salaried employee, I did come in as a precaution New Years Night and happened to look at some control totals for a commission runs, and noticed that they were about one tenth of normal for a month. I called my boss, then called production control. It seemed that there was a PARMLIB date card wrong in the set up. VSAM files had to be restored (from FAVER backups) and the entire month-end commission cycle rerun. Shortly thereafter, there was an effort to automate the datecard setup for many production jobs with sophisticated DYL280 programs. (And remember good old DYL260? It looked like RPG. It looked ancient.)
Back on New Years Day 1986 I had another memorable year-end, then working for Chilton (now Experian), the credit reporting company. I was responsible for end-of-month billing. It the time it was in ALC, and it had been converted from DOS (under DUO, running under MVS with UPSI parameters in the JCL, and DTF’s converted to DCB’s). The EOM cycle started in a number of jobs in groups of five bureaus and ran all day (about eight hours), usually beginning around 6 AM. I still remember the prosaic DOS program names “BA162” and “BA165” that sound like something out of kindergarten (they even invaded our dreams Stephen King style) but became basic vocabulary in the company systems department, then on Fitzhugh in the heart of Dallas ‘s trendy Oak Lawn neighborhood, two miles from downtown. (The Redskins would meet their matches about twelve miles away, in Irving; maybe two miles to the west, EDS had been started in Exchange Park back in 1962; now the Cathedral of Hope is near that site).
Now, it isn’t funny, but my father had died at 3 AM that New Years morning, after a long (age 83), very productive and active life and a sudden short illness (with absolutely no spectacle, and it was the way he said he had wanted it to be). I went in to babysit the cycle – it was mine – before flying “home”
Picture (unrelated): National Presbyterian Center in NW Washington DC.