An IT lesson from the subterranean boroughs


The New York subway system’s controls are even more antiquated than your Windows 2003 servers, but the principles of dealing with both are the same.

By Bob Lewis | January 4, 2016
Topics: Industry Commentary, NewPost, Organizational Effectiveness, Policies and Procedures, Strategy and Tactics, Technology


The value of information struck me while I was waiting for an elevator.

In the lobby of most buildings each elevator’s position is displayed above its door. Armed with that information you can intelligently decide whether to wait for it or climb the stairs instead.

But elevators give you no positional information on any other floor, forcing you to guess whether waiting or using the stairwell is the better vertical travel decision.

Fixing this is trivial engineering. As the Rutles once sang, all you need is cash.

Not so for the New York subway system, as explained in “Why New York Subway Lines Are Missing Countdown Clocks,” (James Somers, The Atlantic, November, 2015), a charming and fascinating article that explains the answer to the author’s question, “I honestly just wanted to know why the F train didn’t have clocks. I never expected it to be so complicated.” (And thanks to long-time correspondent Leo Heska for calling the article to my attention.)

Turns out, the answer isn’t what’s complicated. The New York subway system relies on early 20th century technology whose architects had a clear and well-chosen design goal: Subway trains must not collide.

To accomplish this goal they engineered an elegant combination of sensors, switches, and on-track displays, through which drivers know whether the next section of track is occupied by another train. If so they slow down. If they don’t, a mechanical relay tied to the train-detection sensor automatically applies the brakes.

The system, that is, was designed for Operations, and relies on a highly decentralized combination of human and automated decision-making. Nothing about it identifies individual trains and their positions, so there’s nothing in it to repurpose to tell passengers when the next train will arrive, let alone support any management analytics.

Solving this is a non-trivial problem.

If these were trucks, a GPS receiver, IoT chip, and Google Maps hack would make it pretty easy. But we’re talking about subway trains. They don’t have line of sight to any GPS satellites, and so, never mind.

Maybe there’s nothing to solve. Knowing each train’s position and velocity is, after all, a luxury, not a necessity. Well, okay, except for this small detail: The entire system is worn out, there’s no source of spare parts, and even the wiring’s insulation is about shot.

Oh, and the estimates for replacing this 1930’s vintage technology with something modern start at $20 billion.

Does any of this sound familiar — a legacy system that would be good enough except its architecture is obsolete, the platforms it runs on aren’t around anymore, and:

Of course it does.

We’re dealing with a linked pair of seldom-discussed IT disciplines: Lifecycle management and migration management. Lifecycle management is about detecting incipient obsolescence and preventing it. Migration management is about becoming excellent at replacing obsolete or near-obsolete systems.

Together they make obsolescence avoidance an operational matter, from both a budgeting and an execution perspective.

Competence at migration management is what makes IT very good at moving from obsolete technology to something modern enough to last a while. Lifecycle management is what says it’s time to repeat the cycle.

Here’s how they might have helped New York’s Metropolitan Transit Authority:

By 1985 (to pick a year out of the air), the subway system relied on 50-year-old technology. Computerization was by then mainstream. The subway control system was clearly obsolete.

So imagine if the MTA had started migrating to a modern system in 1985 through a phased, route-by-route plan.

By now it would probably be time to start the next migration … but it would be from a far better base state, with no looming crises from a lack of spare parts and failing insulation driving a high-risk replacement project along.

Depreciation is the mechanism through which the general ledger depicts how a capital asset … the New York subway system’s control system being an example … loses value over time.

What’s strange is how many business executives consider it an accounting fiction. If they just believed their financial statements they’d bank the funds needed for capital asset replacement as standard operating procedure instead of lifecycle management starting with hat-in-hand supplication.

That’s right: The problem isn’t executives managing by the numbers.

It’s executives choosing to ignore them.

9 Responses to “An IT lesson from the subterranean boroughs”

  1. Bob Harris` says:

    An excellent presentation on investment center challenges. But offering features your customers don’t know they need or want can be very risky, unless you happen to be at a company like Apple, where that kind of investment is a fundamental and rewarded part of the corporate mission and the corporate culture.

    It seems to me that sometimes it would be good for IT to collaborate, or at least consult with the organization’s sales department to find the most effective ways of selling end of life cycle changes to management and the organization’s customers. I suspect most in IT didn’t go into computers to become salesmen, but asking the advice of the best of the sales force can be useful for what you learn, as well as an opportunity to gain an important ally through honest good will.

    Besides, what can it hurt?

    • Bob Lewis says:

      I’d agree if I hadn’t heard so many gripe sessions among sales folk about their inability to convince “management” to do various things they consider critical to their ability to sell stuff.

      • Bob Harris` says:

        Sigh…I hear you. Sometimes it can seem that there is no limit to the amount of true foolishness within areas of an organization. But, as a black kid in Chicago, we all thought there would never be an end to discrimination, and yet by the time I entered high school, Dr. King had shown us a way forward. While I’m not quite ready to give you the Nobel Peace Prize yet, I think it highly commendable that you decided to respond in your article.

        Good on ya’, as the Australians say.

  2. allison says:

    So so true… reminded me of the OPM breach, and the age of some critical government systems…

  3. Chris Miller says:

    In the UK (and I’d bet in the US too) several large retail banks are in this position. The key daily systems are old and fragile, and the people who knew how to fix the inevitable problems are retiring or being outsourced to India because they’re too expensive. There have been a number of outages lasting several days(one is currently in progress), and IMHO it’s only a matter of time before one of them encounters a problem from which they can’t recover.

    Deutsche Bank have announced a huge lift-and-shift to update everything, but UK banks are too focused on next quarter’s results for such a project to ever get a green light.

  4. dave says:

    Not sure the previous comment went out…trying again.

    Eric Hoffer once said that the true measure of an organization (and, it appears to me, a person and a nation) is not in what they build but rather in what they maintain. This seems to remain good guidance even today.

    As for better ways to deal with obsolescence, it appears to me that we have to attack the core of the problem. That is that:

    a. The entire financial sector is both attracted to and pressured to consider only the immediate ROI by their investors (including all of us who have 401Ks or similar). Our unspoken instruction to all of these folks is, “Maximize my profits, or else.”

    b. The financial sector responds very well to that unspoken instruction; in turn they apply the same pressure to the Boards and C-level staff of the companies they invest in.

    c. Those Boards / staffs respond to that pressure by making decisions that tend to maximize ROI at the expense of everything else.

    None of the players believe they have an alternative, even if they really want one, because they believe there will ALWAYS be someone who will play the game that way regardless of the human and other costs, and if they don’t play the same way those other folks will clobber them.

    I don’t know how to change this except to start with the financial folks, and start giving them explicit instructions that they are NOT to chase short-term ROI at the complete expense of maintenance, personnel, and other items, and that that kind of pressure should be applied to the folks they invest in.

    Some folks think that “the government” can fix this. That is flawed because (a) the government is a meataxe when a scalpel is needed, and (b) if “the government” is entrusted with this they will be even MORE of a lobbyist target than they are now. Think “regulatory capture” on steroids.

    This may seem like “pie in the sky”; however, there is no other alternative that doesn’t wind up even more destructive than what we have now.

  5. David Molvik says:

    I was once involved in trying to persuade nonprofit organizations to adopt such practices, but ran up against several obstacles:
    1. Most managers had learned the usual commercial attitude, which was: “depreciation is only an accounting dodge which gives us something to deduct from taxes. Nonprofits don’t pay taxes therefore don’t need it. Or else they knew it only as an accounting “fiction” and never considered budgeting for equipment replacement.
    2. Saving money for future equipment replacement is much less exciting than spending it on new toys.
    3. It’s hard to come up with realistic depreciation schedules. Insurance companies are happy to recommend them, but tend to suggest extremely rapid schedules, for reasons of their own which may be obvious. Most people in management don’t know enough about technical equipment they use to have any idea how long it will last.
    Nevertheless, I still agree that such practice is only sensible, but it is a hard sell!

  6. David Howard says:

    This was excellent – the foresight and courage of IT leadership to work with business to manage both functions is essential for long term success. Sadly, many IT leaders lack these qualities, or if they have them, the business leaders do not value the wisdom and instead kick the can down the road.

  7. Sara says:

    Anyone who manages large physical assets plans for and allocates budget for maintenance and upgrades per the depreciation schedule – IF they are allowed to*. Unfortunately the mgmt of the NYC subway system is in charge of neither its revenue nor its budget. The MTA mgmt and politicians in charge of those have many other priorities to consider. For example, the hub-and-spoke design of the NYC subway emphasizes Manhattan’s importance at the cost of serving the other boroughs. When I lived in Manhattan in the 1980’s this was a significant race-class issue, among other transportation issues that prioritized wealthy white Manhattanites. The MTA may have chosen to shift funds to improve/increase outer-boro bus lines to compensate (I don’t know if they did). If you are thinking you might have to replace your outdated subway lines altogether, why devote extremely limited and highly-fought-over resources to updating it?

    Subway mgmt was likely provided just enough funds to keep operating. Wages and security were very high priorities at the time. No one was worried about keeping the system up-to-date so long as it functioned safely. With so much political clout and pressure and voices devoted to the many many other priorities of NYC and MTA, it would have been impossible to reserve money for non-critical capital upgrades.

    For other examples, see how bridges across the US were allowed to slowly deteriorate until some actually started failing. Railroad tracks as well. At least NYC’s subway system hasn’t had those life-endangering problems (that I’m aware of). Keeping the subway running while preventing deaths was probably the highest mandate the subway was funded for.

    *My local school district recently started a round of replacing the turf at school fields. This in spite of chronic serious classroom funding shortfalls. However, facilities mgmt correctly pointed out that the turf replacement funds had been planned for/allocated the decade(s) or so ago when the turf was first installed. So budgeting for capital upgrades can be done, but it requires very long-term foresight to override all those other pesky demands for immediate operational/improvement funding.