Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
precedence: bulk
Subject: Risks Digest 32.53
RISKS-LIST: Risks-Forum Digest Friday 12 March 2021 Volume 32 : Issue 53
ACM FORUM ON RISKS TO THE PUBLIC IN COMPUTERS AND RELATED SYSTEMS (comp.risks)
Peter G. Neumann, founder and still moderator
***** See last item for further information, disclaimers, caveats, etc. *****
This issue is archived at <http://www.risks.org> as
<http://catless.ncl.ac.uk/Risks/32.53>
The current issue can also be found at
<http://www.csl.sri.com/users/risko/risks.txt>
Contents:
Confusing computer-interface complexity causes train crash (Mark Brader)
Expectations of GPS accuracy contribute to train derailment (Mark Brader)
Boeing calls for global grounding of 777s with Pratt&Whitney engines
(NYTimes)
NOAA begins transition exclusively to electronic navigation charts
(Gabe Goldberg)
Weather Service set to discontinue `advisories' for hazardous weather in
2024 (WashPost)
Never seen anything like this': Chaos strikes global shipping (NYTimes)
New Browser Attack Allows Tracking Users Online With JavaScript Disabled
(The Hacker News)
Calling All Ham Radio Operators (Rebecca Mercuri)
Kentucky mom alleges hospital workers missed her cancer, then
covered up their mistake (NBC News)
Microsoft's dream of decentralized IDs enters the real world (WiReD)
What the worldwide shortage of semiconductor chips is *really*
Why a YouTube chat about chess got flagged for hate speech (WiReD)
Farms are going to need different kinds of robots (bbc.com)
The robots are coming for Phil in accounting (NYTimes)
Spy agencies have big hopes for AI (The Economist via Ross Anderson)
A new type of supply-chain attack with serious consequences is flourishing
(Ars Technica)
Google will remove *facts* if they think they're harmful (geoff goodfellow)
Thousands of Android and iOS Apps Leak Data From the Cloud (WiReD)
Hackers are finding ways to hide inside Apple's walled garden
(Techology Review)
ICE investigators used a private utility database covering millions to
pursue immigration violations (WashPost)
L.A. sheriff's office gets warrant for 'black box' in Tiger Woods' crashed
SUV. (NBC News)
Amazon has become a prime revolving-door destination in Washington
(Mother Jones)
Too much choice is hurting America (Paul Krugman via Richard Stein)
CDC Links Restaurant Dining with Spread of Covid-19 in U.S. (Jonathan Spira)
Those fever scanners that everyone is using to fight covid can be wildly
inaccurate, researchers find (WashPost)
The problems with anti-vaccers' precautionary principle arguments
(The Logic of Science)
You got a vaccine. Walgreens got your data. (Vox)
Research highlights impact of Digital Divide (University of Houston)
ES&S hashcode testing is wrong in 3 ways (Andrew Appel)
At least 30,000 U.S. organizations newly hacked via holes in Microsoft's
email software (geoff goodfellow)
Texas PUC to electricity users who received outragrous bills from grid
mismanagment during winter storm: SCREW YOU! (NPR via Lauren Weinstein)
Rookie coding mistake prior to Gab hack came from site's CTO (Ars Technica)
What lies beneath... on disaster respone (NYU Tandon)
Re: Post Office scandal (Peter Bernard Ladkin)
Re: Fed outage shuts down U.S. payment system (John Levine)
Re: his lights stayed on during Texas's storm. Now he owes $16,752
(John Levine)
Abridged info on RISKS (comp.risks)
----------------------------------------------------------------------
Date: Thu, 4 Mar 2021 19:19:11 -0500 (EST)
From: Mark Brader <msb@Vex.Net>
Subject: Confusing computer-interface complexity causes train crash
In November 2019, the driver of an "Intercity Express" class 800 train was
heading into the maintenance depot at Neville Hill, near Leeds, England.
Another empty train was in front of him on the same track into the depot, so
he waited for it to start moving, then put on power at the lowest setting.
About 30 seconds later the two trains collided. (There were no injuries,
but his train was derailed and both trains needed repairs.)
Although an experienced driver, the man was relatively unfamiliar with the
class 800, and didn't realize that even the lowest power setting would
accelerate this train so fast. So he'd assumed it was safe to look down at
the computerized controls for a little while.
The reason he needed to do this was that, partly for political reasons, the
class 800 is equipped both with on-board diesel engines and with a
pantograph for overhead electric power. So the pantograph should be raised,
and the diesels turned off, if and only if there *is* an overhead wire.
Rather than leave the control of this to fallible humans, it is implemented
by computer ("Automatic Power Changeover" or APCO).
Trackside beacons mark the key positions where things have to be done, and
the specific train service must be identified to the computer by the driver
typing its 4-character code into the control console touchscreen.
The train's previous code of 5D29 had expired when it got to Leeds, and the
driver should have changed it to 1D29 to go into the depot, but the system
would not accept it. This was because, after typing the 4 characters, he
was supposed to touch a "Check Stops" button, which would cause the computer
to display a list of the stops for this train service, and then a "Home"
button to return the console to the home screen. Well, the driver knew
there weren't any stops before getting to the depot, so he skipped Check
Stops and just touched "Home". But what his training on the controls had
*not* mentioned was that this canceled the change of code.
Seeing that this hadn't worked, and not wanting the diesels to come on
unnecessarily, he used the controls to disable the APCO, and then when he
passed the last beacon before the depot, re-enabled it -- as he had been
trained to, so that the next driver wouldn't be confused. And it was the
re-enabling step what distracted him from avoiding the crash.
More detail in "Modern Railways" magazine, January 2021, pages 30-32. Full
report, raising a number of points I haven't mentioned here, at:
https://assets.publishing.service.gov.uk/media/5fb3c146d3bf7f63e1b6f55a/R132020_201118_Neville_Hill.pdf
------------------------------
Date: Thu, 4 Mar 2021 20:26:16 -0500 (EST)
From: Mark Brader <msb@Vex.Net>
Subject: Expectations of GPS accuracy contribute to train derailment
In January 2020 a freight train derailed on an overpass over a road in north
London, England. There were no injuries, but the train was damaged as well
as a considerable length of track.
The track on the bridge was supported on wooden beams running lengthwise
under the rails (like on the old Great Western Railway) and connected by
steel cross-members. This design is lighter than conventional track and
therefore allowed the bridge to be more lightly built, but the long-term
plan is to eliminate it from use.
In this case the configuration did not allow the wood to be easily
inspected, and it had deteriorated, but this was not known. What could
readily be done, and was done regularly, was to check the position of the
rails by running a track-recording train over the line, and repair any
problems seen.
On several occasions in the preceding months, this train had reported a
problem in the area of this bridge: the rails were wide to gauge, i.e. too
far apart. Crews were sent out to the location it reported, repaired some
faults they saw in the track, and then checked that the rails were within
the allowed tolerance of the correct gauge.
But the train had reported its position using GPS, and while this was
claimed to be accurate to 1 meter, it wasn't. Crews were expected to use
handheld GPS devices to locate the same position within 3 meters, but the
work they did was a few meters west of the bridge, not where they might have
seen the actual problem.
This wasn't the only cause of the accident, but it is the one I'm mentioning
here. Full report at:
https://assets.publishing.service.gov.uk/media/5faea3a68fa8f55de55af7c9/R122020_201116_Wanstead_Park.pdf
------------------------------
Date: Sat, 6 Mar 2021 23:32:14 +0800
From: Richard Stein <rmstein@ieee.org>
Subject: Boeing calls for global grounding of 777s with Pratt&Whitney engines
(NYTimes)
https://www.nytimes.com/2021/02/21/business/faa-boeing-777-engines.html
"Boeing said on Sunday that all 128 of its 777 jetliners powered by a
particular Pratt & Whitney engine model should be grounded worldwide until
the Federal Aviation Administration determines the best way to inspect the
engines."
Several hundred passengers experienced "the flight of their lives" in two
incidents, apparently due to engine blade fracture from metal fatigue.
Curiously, discloses the FAA met a few days before the UAL 328 incident on
20FEB2021 to discuss airline engine inspection frequency.
https://edition.cnn.com/2021/02/24/politics/faa-meeting-engine-inspections/index.html
The investigatory dust will eventually settle. A revised Pratt & Whitney jet
engine inspection protocol will become regulation standard operating
procedure.
The flying public relies on trained and qualified professionals to certify
an aircraft as safe-to-fly. Their ethics, competence, and professional
judgment are key to sustain aircraft maintenance life cycle, and the air
transportation industry reliability. Public safety depends on inspection to
determine essential aircraft maintenance actions.
Aircraft maintenance is an example of the "expert service problem." It is a
well-known subject in economics. See "When Trust in an Expert Is Unwise,"
via https://www.nytimes.com/2007/11/07/business/07leonhardt.html
"...the same expert who is diagnosing the flaw is the one who will be paid