I don't often feel the need to defend our industry's regulator, the Federal Railroad Administration.
Part of that is me. Like many who have stumbled into a career in this industry (and I literally stumbled my way into railroad employment, half-blinded and three-quarters frozen by a blizzard in Chicago), I've always had a problem with authority. Not that I begrudge anyone his or her authority, title, rate of pay-- any of that stuff. I just don't like other people telling me what to do, and I positively hate it when others think they need to tell me what to do.
Ask some VP-Os. Or my poor wife, that loving woman from Honolulu, Tucson (Tucson?), San Mateo, Santa Cruz, San Francisco now stuck in partial lockdown with me, here in Manhattan No-Beach. New York's a lonely town when you're the only...surfer left in Soho.
There was a general manager who once said to me between firings: "You know what? I'd rather have people working for me who aren't all that smart but do as they're told than have someone like you who insists on thinking for himself."
I said, "I know. That explains why you have so many people working for you who need to be told what to do."
(dialogue guaranteed verbatim).
That general manager went on to bigger and better things. I went on to a different railroad.
Interesting career choice, given the (once upon a time) rather clear lines of authority on the railroad, and the fact that there's always someone telling someone else what to do, no? Interesting career choice, yes. Remember I said I had a problem. I didn't say I had a solution.
Anyway, I'm not usually rushing to the defense of FRA, partly because of my above described personality disorder/order, and partly because FRA's record over the long-haul, over the last 40 years, pretty much defends itself.
Perfect? No, but pretty good. Any jack-ass maneuvers? Of course. Who in this industry has not pulled a jack-ass maneuver? Has not made a move so bizarre that the only explanation is that your body and brain were taken over by an evil extra-terrestial?
When I do feel the need to defend FRA, it's usually because NTSB has issued, or reissued some finding, or recommendation that either
a)ignores the adjudication, reconciliation, compromise functions that are fundamental, to enacting good regulations or
b) requires the regulator and the railroads to expend a maximum of time and effort without a compensating benefit to safe train operations.
Regarding (a) above:
doesn't always work, the adjudication, reconciliation, compromise process, but the effort is always necessary. If compromise is possible so that a reconciled regulation is accepted by all parties, so much the better. Remember, FRA is in this for the long haul. Holding out for the "perfect" regulation gets us nowhere. Why that would be like holding out for PTC 50 years ago after an accident in timetable/train order territory, rather than advocating the installation of CTC and automatic speed control.
Sometimes the reconciliation process doesn't work, as it hasn't regarding medical fitness for duty standards and testing for obstructive sleep apnea. Then the regulator should say, as it should have already said regarding medical fitness for duty standards: "This is the regulation. You have three months to submit your compliance plans. Whimpering not allowed."
Regarding (b) above...would you like an example? Sure you would.
NTSB held a virtual meeting of September 15, 2020 to review the circumstances, findings, and recommendations resulting from its investigation into the collision ot two CSX Transportation freight trains. The colllision occurred at 0509 hours, August 12, 2019 resulting in derailment to both trains, no fatalities, and damages estimated at $4.9 milllion.
According to NTSB's synopsis of the accident, the locomotive engineer of the westbound train had disengaged the PTC in order to facilitate switching moves between a siding ("a track adjacent to a main track used for the meeting or passing of trains") and a yard or industrial track. This is not unusual, given the current configuration of PTC which doesn't always accommodate itself to reverse movements.
What was unusual was that this locomotive engineer was impaired by alcohol while performing this service, and also tested positive for marijuana. While moving his train west on the siding, this impaired locomotive engineer failed to stop at the stop signal, and ran into the eastbound train.
What was unusually unusual is that this locomotive engineer operating the westbound train had escaped the "net" of random drug testing for ten (10) years prior to the accident. The NTSB does not, in this summary, report the date of the engineer's last test for alcohol.
Unusual indeed, but certainly not impossible. The "ordinary" requirements for the rate of random alcohol testing stipulate that the railroad conduct tests equivalent to 25 percent of its covered service employees annually. For drugs the minimum acceptable rate is 10 percent.
However, since these are random tests, since these tests have to be truly random to be effective, to provide any reliable indication of the breadth of a problem, there is no guarantee that 100 percent of the covered employees will be tested in 4 years, or 10 years, or 30 years. That's the nature of random. Everybody goes back in the pool at the beginning of the new testing period when the clock goes back to zero and we work our way up to 25 percent again.
I don't know if that's what took place on CSX or if this is another part of the legacy of Hunter Harrison and good old precision scheduled book-cooking, but it's possible that this engineer was not tested for the simplest and most innocent of reasons, which gives additional meaning to the words of the great Lefty Gomez, "Better to be lucky than good."
So does NTSB recommend to CSX that it supplement its random drug and alcohol testing program with face-to-face interactions between supervisors and covered employees? Oh, child, have you learned nothing in all these years? The simple, direct approach is never the right one.
Speaking of one, that's one. And this is two. NTSB takes this opportunity to re-open and re-recommend this to FRA:
2. Require that railroads regularly review and use in-cab audio and image recordings (with appropriate image limitations on public release, in conjunction with other performance data, to verify that train crew actions are in accordance with rules and procedures that are essential to safety. (R-10-2)
As for FRA's previous response which was an NPRM to require the installation and use of these devices on passenger locomotives but not freight locomotives used in freight service, NTSB reclassifies that from "Open-Acceptable Response," to "Open-Unacceptable Response."
Well, among the problems I don't have, I don't have a problem with my own authority, so here's how I'd respond to NTSB.
First, when FRA requires that railroads do something, that's the beginning of the process for FRA. FRA does not, and cannot require a railroad to do something without providing the metrics, a means to measure the railroad's performance, its compliance with the regulation.
Secondly, how does a regulatory agency provide that metric and then measure the compliance with what is now established as a minimum acceptable standard? Well the regulator will require that the regulatee develop a program for random sampling from the data field made up of the complete record of the events.
Thirdly, what does that mean? It means this. I estimate that locomotive engineers operating over the general railroad system of transportation in the US perform about 60 million hours of service annually.
What size sample would be adequate to our friends at NTSB? Ten percent? OK, lets start there
New NPRM get published in the Federal Register:
FRA intends to issue a regulation requiring passenger, commuter, and all freight railroads with locomotive engineers or any employee acting in the position of a locomotive engineer covered by the hours of service, to review audio and visual records of the actual service performance of said employees. The quantity of review of in-service operation must equal ten percent of the total service time accruing to locomotive engineers performing service in the 12 months prior to adoption of the final regulation.
OK, sound good to you? Scares the hell out of me. But no matter, we've now got 6 million, more or less, service hours to review over 12 months, or 365 days. Using my New York Yankee/Sharp ELS25 calculator (snared on a giveaway day so long ago I can't remember), me and Mr. Sharp figure that works out to reviewing 16,438 hours of audio/visual records every single day for a year. Who does the reviewing? Why, duly qualified transportation officers of the railroad, of course. In this case, that's 2055, give or take, qualified transportation officers doing nothing but this for eight hours a day, every day. Since you can work operating officers seven days a week, cancel their vacations, forego their holidays, we will ignore the incremental adjustment.
What's that? Since you can divide the 8 hours into three operating shifts, each of our 2055 only has to devote 2 hours and 40 minutes of each shift.
Cut it anyway you like, but we're looking at a loss of 1/3 of the officers' available service time. Now does anyone think such an expenditure of time in order to review a 10 percent sample is effective use of an officer's availability?
Maximum effort, miniscule reward.
A bit burdensome, no?
Okay, shall we reduce that sample, say to 1 percent of the total pool, thus reducing our operating officer requirement? But then we're only getting a random sample of 1 percent, and we just saw what happened over on CSX with a random drug sampling requirement of ten percent. They missed one, and the one that makes all the difference in the world, according to the NTSB.
That's the nature of the "beast," managing, supervising, ensuring the safety of this massive operation called the general railroad system of transportation.
We sample, of course. In everything we do, we are sampling although we know that sampling can't be perfect. We might miss one.
We sample when we select and train our employees. We might miss one.
We sample when we make sure our trainees are neither suicidal nor homocidal. We might miss one.
We sample when we observe and test employees. We might miss one.
In fact, we're sampling when we construct algorithms that theoretically do so much of the grunt work for us.
And when we do miss one, as we have to miss one, we go back and review our methodology; we review the assumptions and components of our sampling-- the frequency, the pool, the time of day, etc.-- any and all those elements known as variables that go into making an effective sampling discipline, into a "robust" algorithm.
What we don't do is overwhelm the entire system by adding sampling requirements that cannot, and will not improve the safety and efficiency of the operation. Creating those additional layers, that additional burden can only result in the waste of time, the loss of efficiency, and increasing non-compliance. Those three elements are fatal to system safety. Sooner or later, everyone becomes too busy reviewing yesterday's tapes, or avoiding that task, to pay attention to today's operation, much less tomorrow's.
And I have that on good authority.
David Schanoes
September 19, 2020
I said doctor (doctor),
Mr. MD (doctor)
Can you tell me, what's ailing me (doctor)
He said yeah, yeah, yeah, yeah, yeah (yeah,yeah,yeah, yeah, yeah)
Yes indeed, all you really need
(is good lovin)
-- The Young Rascals
Copyright 2012 Ten90 Solutions LLC. All rights reserved.