The Risks Digest

The RISKS Digest

Forum on Risks to the Public in Computers and Related Systems

ACM Committee on Computers and Public Policy, Peter G. Neumann, moderator

Volume 31 Issue 54

Tuesday 28 January 2020

Contents

Boeing 737s can't land facing west
FAA via Clive D.W. Feather
GPS jamming expected in southeast during military exercise
AOPA
Election Security At The Chip Level
SemiEngineering
Russians Hacked Ukrainian Gas Company at Center of Impeachment
Nicole Perlroth and Matthew Rosenberg
Scientists Deliver, Once Again, a Horrifying Report About How Hot Earth Is Getting
VICE
Ransomware attack forces cancer patients to re-schedule
CBC Web
An Avenue by Which It Might Be Technically Possible to Give an iPhone The Software Equivalent of Cancer
Pixel Envy
Please Stop Sending Terrifying Alerts to Our Cell Phones
WIRED
Update Firefox now, says Homeland Security, to block attacks
9to5mac
A field guide to Iran's hacking groups
Web Informant
Iran hackers have been password-spraying the U.S. electric grid
WiReD
Re: The shooting down of flight PS752 in Iran
Martyn Thomas
In a desperate bid to stay relevant in 2020's geopolitical upheaval, N. Korea upgrades its Apple Jeus macOS malware
The Register
Inside Documents Show How Amazon Chose Speed Over Safety in Building Its Delivery Network
ProPublica
Feds Are Content to Let Cars Drive, and Regulate, Themselves
WIRED
Should Automakers Be Responsible for Accidents?
Gabe Goldberg
Paul Krugman's no-good, very bad Internet day
Ars Techica
Hackers Cripple Airport Currency Exchanges, Seeking $6 Million Ransom
NYTimes
Hacker offers for sale 49M user records from US data broker LimeLeads
Security Affairs
Over two dozen encryption experts call on India to rethink changes to its intermediary liability rules
Tech Crunch
Chosen-Prefix attack against SHA-1 Reported
Ars Technica
Patch Tuesday, January 2020
Rapid7
Facebook Says Encrypting Messenger by Default Will Take Years
WiReD
China's new Cryptolaw
Cointelegraph
Some consumers have noticed that computerization isn't always the answer
Star Tribune
At Mayo Clinic AI engineers face an acid test: Will their algorithms help real patients?
StatNews
AI Comes to the Operating Room
The New York Times
A Very Real Potential for Abuse: Using AI to Score Video Interviews
CNN
5G, AI, blockchain, quantum, …
Marketoonist
Inside the Billion-Dollar Battle Over .Org
Steve Lohr
A lazy fix 20 years ago means the Y2K bug is taking down computers now
New Scientist
When 2 < 7 => failure
Ars Technica via Jeremy Epstein
Make It Your New Year's Resolution Not to Share Misinformation
Mother Jones
Inside the Feds' Battle Against Huawei
WiReD
Apple Is Bullying a Security Company with a Dangerous DMCA Lawsuit
iFixit
How to Protect Yourself From Real Estate Scams
NYTimes
Dutch Artists Celebrate George Orwell's Birthday By Putting Party Hats On Surveillance Cameras
BuzzFeed News
Re: reliability of computers
Chris Drewe
Info on RISKS (comp.risks)

Boeing 737s can't land facing west (FAA)

“Clive D.W. Feather” <clive@davros.org>
Fri, 10 Jan 2020 20:24:07 +0000

“The FAA received reports earlier this year of three incidents of display electronic unit (DEU) software errors on Model 737 NG airplanes flying into runway PABR in Barrow, Alaska. All six display units (DUs) blanked with a selected instrument approach to a runway with a 270-degree true heading, and all six DUs stayed blank until a different runway was selected. […] The investigation revealed that the problem occurs when this combination of software is installed and a susceptible runway with a 270-degree true heading is selected for instrument approach. Not all runways with a 270-degree true heading are susceptible; only seven runways worldwide, as identified in this AD, have latitude and longitude values that cause the blanking behavior.”

(Note that this is all 6 displays on each plane, not 2 displays on each of three planes.)

The runways in question are:

Runway 26, Pine Bluffs, Wyoming, USA (82V) Runway 28, Wayne County, Ohio, USA (KBJJ) Runway 28, Chippewa County, Michigan, USA (KCIU) Runway 26, Cavern City, New Mexico, USA (KCNM) Runway 25, Barrow, Alaska, USA (PABR) Runway 28, La Mina, La Guajira, Colombia (SKLM) Runway 29, Cheddi Jagan, Georgetown, Guyana (SYCJ)

(The numbers are magnetic bearings, whereas the problem is apparently related to true bearing.)

Original FAA notice: <http://rgl.faa.gov/Regulatory_and_Guidance_Library/rgad.nsf/0/3948342a978cc27b862584dd005c1a60/$FILE/2019-25-17.pdf>

[Clive, Can you think of the significance of 270? Perhaps an instance of Buridan's Ass algorithm, in this case being halfway between 180 and 360, and not being able to decide? PGN]

[I have no idea. Also, why don't all runways facing 270 have the problem? I suspect we'll never find out. Clive]


GPS jamming expected in southeast during military exercise (AOPA)

Paul Saffo <paul@saffo.com>
Fri, 17 Jan 2020 07:30:56 -0800

Dan Namowitz, AOPA, 14 Jan 2020

GPS reception may be unavailable or unreliable over a large portion of the southeastern states and the Caribbean during offshore military exercises scheduled between January 16 and 24. aopa.org/news-and-media/all-news/2020/january/14/gps-jamming-expected-in-southeast-during-military-exercise

Graphic depicting area of GPS interference testing. Courtesy of the FAA. The FAA has posted a flight advisory for the exercises that will require jamming of GPS signals for periods of several hours each day of the event. Navigation guidance, ADS-B, and other services associated with GPS could be affected for up to 400 nautical miles at Flight Level 400, down to a radius of 180 nm at 50 feet above the ground.

The flight advisory encourages pilots to report any GPS anomalies they encounter. Reports may be submitted using this online form.

AOPA reported on a similar event in the southeastern United States in 2019.

AOPA is aware of hundreds of reports of interference to aircraft during events around the country for which notices to airmen were issued, and we consider the risks to GA aircraft highly concerning.

In one example, an aircraft lost navigation capability and did not regain it until after landing. Other reports have highlighted aircraft veering off course and heading toward active military airspace—and the wide range of reports makes it clear that interference affects aircraft differently. In some cases, recovery from signal interference may not occur until well after the aircraft exits the jammed area.

In a January 2019 AOPA survey, more than 64 percent of 1,239 pilots who responded noted concern about the impact of interference on their use of GPS and ADS-B.

AOPA continues to advocate for officials to place more focus on efforts to address the well-documented safety concerns raised by such events.


Election Security At The Chip Level (SemiEngineering)

Gabe Goldberg <gabe@gabegold.com>
Wed, 15 Jan 2020 00:40:24 -0500

https://semiengineering.com/how-secure-are-electronic-voting-machines/


Russians Hacked Ukrainian Gas Company at Center of Impeachment (Nicole Perlroth and Matthew Rosenberg)

“Peter G. Neumann” <neumann@csl.sri.com>
Wed, 15 Jan 2020 15:11:02 PST

Nicole Perlroth and Matthew Rosenberg, *The New York Times* 13 Jan 2020, updated in the online version 15 Jan 2020 https://www.nytimes.com/2020/01/13/us/politics/russian-hackers-burisma-ukraine.html

Offices in Kyiv of a subsidiary of the Ukrainian energy company Burisma. Security experts suggest the hackers may have been looking for damaging information on Joe Biden.

With President Trump facing an impeachment trial over his efforts to pressure Ukraine to investigate former Vice President Joseph R. Biden Jr. and his son Hunter Biden, Russian military hackers have been boring into the Ukrainian gas company at the center of the affair, according to security experts.

The hacking attempts against Burisma, the Ukrainian gas company on whose board Hunter Biden served, began in early November, as talk of the Bidens, Ukraine and impeachment was dominating the news in the United States.

It is not yet clear what the hackers found, or precisely what they were searching for. But the experts say the timing and scale of the attacks suggest that the Russians could be searching for potentially embarrassing material on the Bidens - the same kind of information that Mr. Trump wanted from Ukraine when he pressed for an investigation of the Bidens and Burisma, setting off a chain of events that led to his impeachment.

The Russian tactics are strikingly similar to what American intelligence agencies say was Russia's hacking of emails from Hillary Clinton's campaign chairman and the Democratic National Committee during the 2016 presidential campaign. In that case, once they had the emails, the Russians used trolls to spread and spin the material, and built an echo chamber to widen its effect.


Scientists Deliver, Once Again, a Horrifying Report About How Hot Earth Is Getting (VICE)

the keyboard of geoff goodfellow <geoff@iconia.com>
Thu, 16 Jan 2020 14:20:00 -1000

“These are big numbers for our planet,” one NASA scientist told VICE News

EXCERPT:

In 2019, parts of the planet were hotter than they've ever been before, according to NASA and NOAA's annual temperature report. And scientists are warning the world won't be able to reverse the damage.

For the first time ever, the average temperature in Alaska was above freezing. And Australia, at more than 1.5 degrees Celsius above normal, was as hot as the UN hopes the world will ever get.

As a whole, 2019 was the second hottest year on record, according to the report, published by government scientists on Wednesday. That caps off the hottest decade in recorded history. The last half of the decade was also one for the record books: All five years, together, were the hottest on record. The cause, the scientists say, is clearly human-emitted greenhouse gases.

“The last ice age, where we had ice covering North America and most of Europe was only five degrees [Celsius] colder than the pre-industrial planet,” Gavin Schmidt, director of NASA's Goddard Institute for Space Studies, told VICE News.

“We've warmed up a fifth of that,” he added. “These are big numbers for our planet.”

In addition to Alaska and Australia, Poland and other parts of eastern Europe also broke temperature records, as did Madagascar, New Zealand, parts of Southern Africa, and eastern South America. And on top of the high temperatures, glaciers are melting at record rates <https://www.businessinsider.com/greenland-ice-melting-is-2070-worst-case-2019-8> in Greenland. Hurricanes and typhoons are becoming more intense. And wildfires are getting bigger and more frequent.

The planet' has already warmed a full degree Celsius above pre-industrial levels—and scientists say there's likely no turning back. Just because the planet wasn't *quite* as warm in 2019 as it was in 2016 that shouldn't not be misinterpreted as climate change turning around.

“This whole, ‘Oh, we've been cooling since 2016' point—that's just bullshit,” Schmidt said…

[…] https://www.vice.com/en_us/article/884gx3/scientists-deliver-once-again-a-horrifying-report-about-how-hot-earth-is-getting


Ransomware attack forces cancer patients to re-schedule (CBC Web)

“David E. Ross” <david@rossde.com>
Thu, 16 Jan 2020 14:36:55 -0800

eHealth is the provincial health authority in Saskatchewan, Canada. Note that they have a backup plan for such situations. The attack began 6 January. Treatments for affected patients were delayed 24 to 48 hours. By 14 January, the effects of the attack were apparently resolved.

The news article on the Canadian Broadcasting Company Web site had the headline:

Ransomware attack on eHealth forces 31 cancer patients to re-schedule radiation treatment

The article read:

Six patients booked for chemotherapy also affected.

A ransomware attack on the computer system that stores confidential medical data for Saskatchewan residents ended up affecting almost 40 patients getting cancer treatment in Saskatoon and Regina.

The attack on eHealth Saskatchewan began Jan. 6. Antivirus software immediately began sending alerts to staff.

When eHealth officials attempted to open files on affected servers they received a message that the files had been encrypted and would remain inaccessible until a payment was made.

The Saskatchewan Cancer Agency oversees the two cancer clinics in Saskatoon and Regina. It disconnected from the eHealth network after learning of the assault on the system.

While the move served to protect patient data, it also meant that staff could not immediately access provincial lab results, imaging pathology and pharmacy and medical information.

eHealth hit by ransomware attack but personal health data is secure, says CEO.

The clinics have contingency plans for when the electronic records are not accessible but it took time to co-ordinate retrieving the information.

As a result, 31 patients booked for radiation and another six with chemotherapy appointments had their treatment delayed by between 24 and 48 hours.

Each patient was given a personal explanation and apology for the delay and inconvenience, officials with Saskatchewan Cancer Agency said in an emailed statement.

The agency fully reconnected with the eHealth network on Jan. 14.


An Avenue by Which It Might Be Technically Possible to Give an iPhone The Software Equivalent of Cancer (Pixel Envy)

Gabe Goldberg <gabe@gabegold.com>
Thu, 16 Jan 2020 18:23:10 -0500

https://pxlnv.com/blog/software-equivalent-of-cancer/


Please Stop Sending Terrifying Alerts to Our Cell Phones (WIRED)

Gabe Goldberg <gabe@gabegold.com>
Tue, 7 Jan 2020 20:04:15 -0500

https://www.wired.com/story/please-stop-sending-terrifying-alerts-to-my-cell-phone/


Update Firefox now, says Homeland Security, to block attacks (9to5mac)

Monty Solomon <monty@roscom.com>
Fri, 10 Jan 2020 11:30:15 -0500

https://ww.9to5mac.com/2020/01/10/update-firefox-now/


A field guide to Iran's hacking groups (Web Informant)

Gabe Goldberg <gabe@gabegold.com>
Fri, 17 Jan 2020 09:54:15 -0500

https://blog.strom.com/wp/?p=7529


Iran hackers have been password-spraying the U.S. electric grid (WiReD)

Gabe Goldberg <gabe@gabegold.com>
Fri, 10 Jan 2020 20:50:38 -0500

A state-sponsored group called Magnallium has been probing American electric utilities for the past year. https://www.wired.com/story/iran-apt33-us-electric-grid/


Re: The shooting down of flight PS752 in Iran

Martyn Thomas <martyn@thomas-associates.co.uk>
Mon, 13 Jan 2020 10:10:55 PST

It seems to me that commercial aircraft shouldn't fly within range of anti-aircraft systems at a time of high military alert, because human error or computer system error is too likely. If that wasn't obvious before the USS Vincennes shot down Iran Air 655 in 1988, it should have become obvious immediately afterwards. Iran Air 655 has been regarded in the literature as a “Normal Accident”, using Chick Perrow's terminology.

Air defence systems are major intelligence targets, so several states with significant cyber capability will have been trying to compromise the Iranian system over an extended period. It would surprise me if they had all completely failed. This heightens the probability that an aircraft may be misidentified.

If an air defence system identifies (or appears to identify) a radar contact as something that will strike fatally within a small number of seconds, the missile defences will be fired, whether there is a human in the loop or not.

I find it impossible to allocate blame.

[As we have said so often in RISKS, blame can often be remarkably widely distributed. Here are subsequent reports of the Iranian revolutionary guards air-defense comms being jammed, and other issues relating to this shootdown. See the NYTimes article “Anatomy of a Lie”, on how the events around the shootdown unfolded: https://www.nytimes.com/2020/01/26/world/middleeast/iran-plane-crash-coverup.html

This item came in recently, although RISKS-31.54 was ready to be sent weeks ago. We are still resolving internal mailer problems that massively rejected delivery of RISKS-31.53 to many readers. It appears to be Office 365 problem or a side-effect of SRI's installation of proofpoint to block executable attachments. Let's see if this issue gets through.

PLEASE submit RISKS items for consideration as ASCII text to RISKS without attachments to facilitate my efforts. Office 365 is now introducing several hundred lines of headers, which makes things even worse. PGN]

WARNING: I've had a slew of mailman messages dropping readers's subscriptions. If you did not get this message via the normal mailing, you need to resubscribe. SORRY. I have no control over this. PGN


In a desperate bid to stay relevant in 2020's geopolitical upheaval, N. Korea upgrades its Apple Jeus macOS malware (The Register)

Monty Solomon <monty@roscom.com>
Thu, 9 Jan 2020 11:56:01 -0500

https://www.theregister.co.uk/2020/01/08/applejeus_malware_returns/


Inside Documents Show How Amazon Chose Speed Over Safety in Building Its Delivery Network (ProPublica)

Gabe Goldberg <gabe@gabegold.com>
Wed, 8 Jan 2020 23:45:24 -0500

https://www.propublica.org/article/inside-documents-show-how-amazon-chose-speed-over-safety-in-building-its-delivery-network

…but we all want our stuff right now…


Feds Are Content to Let Cars Drive, and Regulate, Themselves (WIRED)

Gabe Goldberg <gabe@gabegold.com>
Sat, 11 Jan 2020 17:29:06 -0500

A new Transportation Department policy on self-driving cars is long on boosting the industry and short on ensuring its safety.

Not all road safety advocates are pleased with that approach. “The DOT is supposed to ensure that the US has the safest transportation system in the world, but it continues to put this mission second, behind helping industry rush automated vehicles,” Ethan Douglas, a senior policy analyst for cars and product safety at Consumer Reports, said in a statement.

https://www.wired.com/story/feds-content-cars-drive-regulate-themselves/


Should Automakers Be Responsible for Accidents?

Gabe Goldberg <gabe@gabegold.com>
Fri, 17 Jan 2020 10:29:53 -0500

What a strange scheme:

Automaker enterprise liability would have useful incentives that driver liability law misses.

My basic argument is that while current negligence-based auto liability rules could in theory work to provide optimal accident-avoidance incentives, in practice they do not. The current system requires courts and drivers to evaluate benefit“cost tradeoffs they are not equipped to make. Also under the current system, much of auto-accident costs are offloaded onto medical and disability insurers or taxpayers. By contrast, under an automaker enterprise liability system, responsibility for those costs would be placed on the parties in the best position to reduce and insure them: vehicle manufacturers. In addition, automakers would be induced to charge enough for cars to fully internalize the costs of automobile accidents. Further, if auto-insurance contracts”and auto-insurance premium adjustments”could be deployed to improve driving habits, auto manufacturers would be induced to coordinate with auto insurers to achieve these deterrence gains. Moreover, to the extent that Level 5s reduce the cost of accidents, they would be cheaper to purchase than conventional vehicles, which would provide a natural subsidy to encourage (and potentially accelerate) their deployment.

https://www.cato.org/sites/cato.org/files/serials/files/regulation/2019/3/regulation-v42n1-1.pdf


Paul Krugman's no-good, very bad Internet day (Ars Techica)

Monty Solomon <monty@roscom.com>
Fri, 10 Jan 2020 12:29:04 -0500

https://arstechnica.com/information-technology/2020/01/paul-krugmans-no-good-very-bad-internet-day/


Hackers Cripple Airport Currency Exchanges, Seeking $6 Million Ransom (NYTimes)

Gabe Goldberg <gabe@gabegold.com>
Thu, 9 Jan 2020 23:07:32 -0500

https://www.nytimes.com/2020/01/09/business/travelex-hack-ransomware.html


Hacker offers for sale 49M user records from US data broker LimeLeads (Security Affairs)

Monty Solomon <monty@roscom.com>
Thu, 16 Jan 2020 14:34:46 -0500

https://securityaffairs.co/wordpress/96432/data-breach/limeleads-data-leak.html


Over two dozen encryption experts call on India to rethink changes to its intermediary liability rules (Tech Crunch)

Monty Solomon <monty@roscom.com>
Fri, 10 Jan 2020 12:17:45 -0500

https://techcrunch.com/2020/01/09/over-two-dozen-encryption-experts-call-on-india-to-rethink-changes-to-its-intermediary-liability-rules/


Chosen-Prefix attack against SHA-1 Reported (Ars Technica)

“Bob Gezelter” <gezelter@rlgsc.com>
Tue, 07 Jan 2020 13:12:37 -0700

As reported in Ars Technica, a team of researchers recently presented a paper reporting a successful chosen-prefix attack against SHA-1. This has implications for OpenSSL, PGP, Git, and other components and processes that rely on the use of SHA-1 message digests for proving authenticity.

The full article can be found at: https://arstechnica.com/information-technology/2020/01/pgp-keys-software-security-and-much-more-threatened-by-new-sha1-exploit/

The underlying paper is at: https://eprint.iacr.org/2020/014.pdf


2020 first Patch Tuesday: Windows' ECC certificates (Rapid7)

Gabe Goldberg <gabe@gabegold.com>
Wed, 15 Jan 2020 23:48:50 -0500

The first Patch Tuesday of 2020 has been hotly anticipated due to a rumour that Microsoft would be fixing a severe vulnerability in a fundamental cryptographic library. It turns out that the issue in question is indeed serious, and was reported to Microsoft by the NSA: CVE-2020-0601 is a flaw in the way Windows validates Elliptic Curve Cryptography (ECC) certificates. It allows attackers to spoof a code-signing certificate that could be used to sign a malicious executable, which would look totally legitimate to the end user. It also enables attackers to conduct man-in-the-middle attacks and decrypt confidential information on user connections to affected systems. This vulnerability exists in Windows 10, Server 2016, and Server 2019. These systems need to be patched immediately, as correct certificate validation is vital for determining trust.

https://blog.rapid7.com/2020/01/14/patch-tuesday-january-2020/

[Steven Cheung noted this (WSJ)

“The flaw at issue involves a mistake in how Microsoft uses digital signatures to verify software as authentic, which helps block malware from being deployed on a computer. The error would potentially enable hackers to install powerful malware on systems undetected.”] https://www.wsj.com/articles/microsoft-releases-patch-to-severe-windows-flaw-detected-by-nsa-11579030780


Facebook Says Encrypting Messenger by Default Will Take Years (WiReD)

Gabe Goldberg <gabe@gabegold.com>
Sun, 12 Jan 2020 16:19:24 -0500

Mark Zuckerberg promised default end-to-end encryption throughout Facebook's platforms. Nearly a year later, Messenger's not even close.

https://www.wired.com/story/facebook-messenger-end-to-end-encryption-default/

No rush…


China's new Cryptolaw (Cointelegraph)

“Peter G. Neumann” <neumann@csl.sri.com>
Mon, 13 Jan 2020 10:26:01 PST

cointelegraph.com/news/china-prepares-for-cbdc-with-cryptography-law-on-encryption-standards

On 1 Jan 2020, China's law governing cryptographic password management came into power. Essentially, the act aims to set standards for the application of cryptography and the management of passwords, and, therefore, ultimately reduces China's cyber vulnerabilities on a nationwide scale. Some local media outlets rumor that the law is paving the way for the long-awaited release of China's central bank digital currency, although it does not make any explicit references in that regard. Meanwhile, the private sector is worried about the anonymity of its data. […]


Some consumers have noticed that computerization isn't always the answer (Star Tribune)

Steve Summit
Fri, 10 Jan 2020 10:30:34 -0500

Not the usual sort of risk, but here's a nice article on the premium placed by savvy farmers on tractors built before 1980 or so, in significant part because they're *not* computerized and can therefore be maintained by anyone.

http://www.startribune.com/for-tech-weary-midwest-farmers-40-year-old-tractors-now-a-hot-commodity/566737082/


At Mayo Clinic AI engineers face an acid test: Will their algorithms help real patients? (StatNews)

Richard Stein <rmstein@ieee.org>
Sun, 12 Jan 2020 12:22:00 +0800

https://www.statnews.com/2019/12/18/mayo-clinic-artificial-intelligence-acid-test/

A sobering peak at AI's potential role in medicine at the front line, with patient data-in-the-loop, applied to ferret out atrial fibrillation (a-fib) precursors using a convolution neural network—the same algorithm applied by driverless vehicles to recognize traffic signs and road obstacles, etc.

“The largest share of the data is derived from electrocardiograms (EKGs), a century-old technology that is commonly used to evaluate heart function by recording electrical pulses that cause the heart to beat. About 250,000 EKGs are performed every year at Mayo, which has a digital dataset of 7 million records stretching back to the mid-1990s.”

“EKGs have been able to detect a-fib for decades, but Mayo is seeking to take it a step further ” by trying to predict which patients will experience this arrhythmia in the future.” […]

“In a study published in August, Mayo reported the algorithm was able to accurately identify patients with a-fib at an 80-percent accuracy rate. On a recent afternoon, its power was displayed in the case of a patient who had undergone EKGs over a 30-year period but had never been diagnosed with a-fib. Inside a conference room, a group of engineers and cardiologists scanned the peaks and valleys of the data projected on a screen for any sign of an abnormality.”

“Dr. Samuel Asirvatham, an electrophysiologist who reads EKGs as automatically as most people drive a flat stretch of interstate, jumped up from his chair to take a closer look. He flipped forward in the series of EKGs and then back, but nothing seemed to call out a certainty of atrial fibrillation. However, the AI system, when it was shown the same data, detected a hidden pattern pinpointing two occasions when the patient's risk of atrial fibrillation had increased dramatically.”

“As it turned out, both of those EKGs preceded cryptogenic strokes, or strokes of unknown cause, that, in hindsight, may have been caused by the a-fib.”

Focusing on patient outcome improvement potential is a key performance indicator for effective medical care delivery. That the article does not mention false-negative/positive and area-under-curve/receiver-operating-characteristics (AUCROC) suggests some undisclosed algorithmic sensitivity derived from the MAYO dataset—though it embodies a sizable patient sample history.

As described by the essay, the data used is selective and filtered — presented as evidence of merit for premonitory a-fib detection where none is currently visible in a given cardiogram—normal sinus rhythm presented. That a physician skilled in the art can recognize 'cryptogenic stroke' indicators based on prior cardiogram reading, as can the machine, suggests equivalent detection capability when both are given a sufficiently rich dataset.

Interpreting an isolated electro-cardiogram to predict a-fib occurrence or recurrence risks, independent of patient history, is quack medicine.

Cardiac electrophysiologists often assess a-fib risks using patient factors that antagonize: high blood pressure, obstructive sleep apnea, obesity, high cholesterol, sedentary life style, prior a-fib events, etc. Typically, the CHADS2 score (https://www.mdcalc.com/chads2-score-atrial-fibrillation-stroke-risk) encapsulates these factors to estimate stroke risk.

Perhaps the motive to justify proactive a-fib prediction is to suppress or optimize future medical care expenditures. ~1% of the US population (~3 million people) are diagnosed with a-fib each year.

How many patients will be falsely diagnosed or misdiagnosed by “The Stroke Predictor Model 9000”? What costs (and potential hardships) will be incurred by patients, physicians, and medical system who rely on AI-enhanced incidents? Will these adverse incidents diminish or increase in frequency? Where's the double-blind study to certify and justify adoption of this device into cardiac care protocol?

Risk: AI-based cardiogram signal processing and interpretation.


AI Comes to the Operating Room (The New York Times)

Richard Stein <rmstein@ieee.org>
Wed, 8 Jan 2020 12:14:15 +0800

https://www.nytimes.com/2020/01/06/health/artificial-intelligence-brain-cancer.html

“Images made by lasers and read by computers can help speed up the diagnosis of brain tumors during surgery.”

A ‘frozen section’ analysis of brain tissue only requires ~2 minutes given the candidate technique. In the old days, 30+ minutes elapsed while the patient waited under anesthesia for a carbon-based pathology assessment.

Speed is important, too: less time on the operating room table, and a “quick second opinion,” albeit by 'deep learning' trained-machine to recognize tumors in the flesh. MRIs apparently don't always yield a conclusive pre-op diagnosis. Hence the need for biopsy supplement.

“The study involved brain tissue from 278 patients, analyzed while the surgery was still going on. Each sample was split, with half going to AI and half to a neuropathologist. The diagnoses were later judged right or wrong based on whether they agreed with the findings of lengthier and more extensive tests performed after the surgery.”

“The result was a draw: humans, 93.9 percent correct; AI, 94.6 percent.” ‘Correct’? No false-positive or false-negative AUC ROC measures?

You should trust your physician—they swear by the Hippocratic Oath. Trust the physician's tool supply chain? Not so fast.


A Very Real Potential for Abuse: Using AI to Score Video Interviews (CNN)

”Bob Gezelter” <gezelter@rlgsc.com>
Thu, 16 Jan 2020 04:01:34 -0700

CNN has published an article on an interesting trend: the use of AI evaluations of candidate video interviews during the selection process for internships and jobs.

As in other cases with AI-based evaluation of imagery, the potential for baked-in bias is clear. Without extensive study, is there a way to validate that such mechanisms are free of explicit or implicit bias concerning race, culture, and other factors. As an example, the subject of “word choice”. In some cultures, directness is valued, in other cultures, precisely the opposite is true. It would be far too simple for a bot to downgrade a candidate for “lack of directness” when their cultural background values it. Would that not be effective discrimination on race, national origin, or other prohibited or suspect factor.

A thought experiment: Consider scoring the statement “The patient has a tumor” with the all-but-required phrasing used by a radiologist “The patient's imagery is consistent with the presence of a tumor”. Is one of these options “evasive”?

One could argue that it is a matter of what questions are asked, but that presupposes a degree of sophistication which is likely not present in practice.

https://www.cnn.com/2020/01/15/tech/ai-job-interview/index.html


5G, AI, blockchain, quantum, … (Marketoonist)

Gabe Goldberg <gabe@gabegold.com>
Mon, 13 Jan 2020 13:19:47 -0500

Smart Devices and 5G cartoon | Marketoonist | Tom Fishburne

With the imminent arrival of 5G, there's a lot of euphoric talk about about the future of connected devices, which is leading to a fair amount of technology-for-technology-sake. And there are many funny and no-so-funny bumps in the road.

On the funny end of the spectrum, GE was mocked <https://www.marketwatch.com/story/this-ridiculous-ge-video-showing-14-steps-to-reset-a-smart-lightbulb-has-suddenly-gone-viral-2019-06-20> a few months ago for releasing a guide to reset their Smart Lightbulb. It requires 14 complicated steps of turning it off and on at exact second counts with a stopwatch (“turn off for two seconds … turn on for eight seconds”). Stephen Fry remarked <https://www.marketwatch.com/story/this-ridiculous-ge-video-showing-14-steps-to-reset-a-smart-lightbulb-has-suddenly-gone-viral-2019-06-20>, “This is insane enough to be joyous.”

On the not-so-funny end of the spectrum, smart-device maker Wyze announced <https://www.marketwatch.com/story/smart-device-maker-wyze-confirms-data-breach-that-could-affect-millions-2019-12-29> two weeks ago that both of the company's production databases were left entirely open to the Internet, exposing the data of 2.4 million users of their smart-home cameras and devices.

These are all reflections of the awkward adolescent stage of technology we're living and working in. We have to continually question just how “smart” all of this “smart” technology really is.

https://marketoonist.com/2020/01/smart.html


Inside the Billion-Dollar Battle Over .Org (Steve Lohr)

Richard Forno <rforno@infowarrior.org>
January 8, 2020 8:14:28 JST

[via Dave Farber]

Steve Lohr, *The New York Times*, 7 Jan 2020

A private equity firm wants to buy the Internet domain used by nonprofits. A group of online pioneers says it is not the place to maximize profits.

Two months ago, Ethos Capital, a private equity firm, announced that it planned to buy the rights to a tract of Internet real estate for more than $1 billion. But it wasn't just any piece of digital property. It was dot-org, the cyber neighborhood that is home to big nonprofits and nongovernmental organizations like the United Nations (un.org) and NPR (npr.org), and to little ones like neighborhood clubs.

The deal was met with a fierce backlash. Critics argued that a less commercial corner of the Internet should not be controlled by a profit-driven private equity firm, as a matter of both principle and practice. Online petitions and letters of concern came from hundreds of organizations, thousands of individuals and four Democrats in Congress, including Senator Elizabeth Warren of Massachusetts.

Rarely has the acronym-strewn realm of Internet addresses—so-called domain names—stirred such passion.

Now, a group of respected Internet pioneers and nonprofit leaders is offering an alternative to Ethos Capital's bid: a nonprofit cooperative corporation. The incorporation papers for the new entity, the Cooperative Corporation of .ORG Registrants, were filed this week in California. […] [PGN-ed, longish item, truncated]

https://www.nytimes.com/2020/01/07/technology/dot-org-private-equity-battle.html?emc=3Drss&partner=3Drss


A lazy fix 20 years ago means the Y2K bug is taking down computers now (New Scientist)

Paul Saffo <paul@saffo.com>
Thu, 9 Jan 2020 21:03:39 -0800

[Re: Martyn Thomas, This might be a genuine Y2K problem—are there more? RISKS-31.50]

Chris Stokel-Walker, *New Scientist*, 7 Jan 2020 https://www.newscientist.com/article/2229238-a-lazy-fix-20-years-ago-means-the-y2k-bug-is-taking-down-computers-now/

[PGN-ed to avoid duplication with RISKS-31.50 and 53.]

[…] Programmers wanting to avoid the Y2K bug had two broad options: entirely rewrite their code, or adopt a quick fix called “windowing”, which would treat all dates from 00 to 20, as from the 2000s, rather than the 1900s. An estimated 80 per cent of computers fixed in 1999 used the quicker, cheaper option.

“Windowing, even during Y2K, was the worst of all possible solutions because it kicked the problem down the road,” says Dylan Mulvin at the London School of Economics.

Coders chose 1920 to 2020 as the standard window because of the significance of the midpoint, 1970. “Many programming languages and systems handle dates and times as seconds from 1970/01/01, also called Unix time,” says Tatsuhiko Miyagawa, an engineer at cloud platform provider Fastly.

Unix is a widely used operating system in a variety of industries, and this v“epoch time” is seen as a standard.

The theory was that these windowed systems would be outmoded by the time 2020 arrived, but many are still hanging on and in some cases the issue had been forgotten.

“Fixing bugs in old legacy systems is a nightmare: it's spaghetti and nobody who wrote it is still around,” says Paul Lomax, who handled the Y2K bug for Vodafone. “Clearly they assumed their systems would be long out of use by 2020. Much as those in the 60s didn't think their code would still be around in the year 2000.”

Those systems that used the quick fix have now reached the end of that window, and have rolled back to 1920. Utility company bills have reportedly been produced with the erroneous date 1920, while tens of thousands of parking meters in New York City have declined credit card transactions because of the date glitch.

Thousands of cash registers manufactured by Polish firm Novitus have been unable to print receipts due to a glitch in the register's clock. The company is attempting to fix the machines.

WWE 2K20, a professional wrestling video game, also stopped working at midnight on 1 January 2020. Within 24 hours, the game's developers, 2K, issued a downloadable fix.

Another piece of software, Splunk, which ironically looks for errors in computer systems, was found to be vulnerable to the Y2020 bug in November. The company rolled out a fix to users the same week—which include 92 of the Fortune 100, the top 100 companies in the US.

Some hardware and software glitches have been incorrectly attributed to the bug. One healthcare professional claimed Y2020 hit a system developed by McKesson, which produces software for hospitals. A spokesperson for McKesson told New Scientist the firm was unaware of any outage tied to Y2020.

Exactly how long these Y2020 fixes will last is unknown, as companies haven't disclosed details about them. If the window has simply been pushed back again, we can expect to see the same error crop up.

Another date storage problem also faces us in the year 2038. The issue again stems from Unix's epoch time: the data is stored as a 32-bit integer, which will run out of capacity at 3.14 am on 19 January 2038.

[In response to a request from Eric Hofnagel, I pulled together a historical list of Y2K-related problems. It is now on my website http://www.csl.sri.com/neumann/neumann.html at http://www.csl.sri.com/neumann/y2k-pgn.txt PGN]


When 2 < 7 => failure (Ars Technica)

Jeremy Epstein <jeremy.j.epstein@gmail.com>
Mon, 13 Jan 2020 13:35:59 -0500

Grocery store system does periodic audits of self-checkout users, but the system doesn't work if you have fewer than 7 items - the audit requires auditing exactly seven items.

Granted, not the biggest risk in the world, but if the venue didn't have in-person employees, what would the customer do?

https://arstechnica.com/staff/2020/01/how-i-broke-my-grocery-stores-app-by-not-buying-enough-stuff/


Make It Your New Year's Resolution Not to Share Misinformation (Mother Jones)

Gabe Goldberg <gabe@gabegold.com>
Tue, 7 Jan 2020 20:18:50 -0500

https://www.motherjones.com/politics/2020/01/make-it-your-new-years-resolution-not-to-share-misinformation/

Not profound but worth sharing with the less tech-savvy.


Inside the Feds' Battle Against Huawei (WiReD)

Gabe Goldberg <gabe@gabegold.com>
Fri, 17 Jan 2020 11:50:03 -0500

https://www.wired.com/story/us-feds-battle-against-huawei/

Long, interesting…


Apple Is Bullying a Security Company with a Dangerous DMCA Lawsuit (iFixit)

Gabe Goldberg <gabe@gabegold.com>
Mon, 6 Jan 2020 19:57:42 -0500

https://www.ifixit.com/News/apple-is-bullying-a-security-company-with-a-dangerous-dmca-lawsuit


How to Protect Yourself From Real Estate Scams (NYTimes)

Gabe Goldberg <gabe@gabegold.com>
Mon, 6 Jan 2020 19:58:52 -0500

https://www.nytimes.com/2020/01/03/realestate/how-to-protect-yourself-from-real-estate-scams.html

Not entirely new, but worth reading how it works, what to do and not to.


Dutch Artists Celebrate George Orwell's Birthday By Putting Party Hats On Surveillance Cameras (BuzzFeed News)

Gabe Goldberg <gabe@gabegold.com>
Fri, 17 Jan 2020 10:14:25 -0500

https://www.buzzfeednews.com/article/ellievhall/dutch-artists-celebrate-george-orwells-birthday-by-adorning


Re: reliability of computers (RISKS-31.53)

Chris Drewe <e767pmk@yahoo.co.uk>
Mon, 06 Jan 2020 20:27:28 +0000

This brought back memories from a guy at the company where I used to work, as he told of being called in as an expert witness on something very similar back in the 1990s. As I recall, he said that two banks or building societies (mortgage providers) had merged; they had totally different computer systems, but the new managers simply fired one of the support teams and expected the other to cope with both systems, which they struggled to do. His expert opinion was that security on the unsupported system was a disaster area, with security features not enabled, passwords and log-ins left with default settings, etc. As mentioned, he felt sympathy for the police officer, who queried some transactions on his account and ended up being charged with attempting to obtain money by deception. The geographical location for the case was Woodbridge, Suffolk.

By the way, there was a similar “our computers are never wrong” item on a BBC radio programme covering consumer affairs a couple of months ago. This featured a woman with a regular Chip&PIN credit/debit card, which had expired and been routinely replaced by the card provider. She was told to cut up the old one but forgot to do this, however she expected it to be cancelled anyway so wasn't concerned. Quite some time later she found unexpected transactions on the account and was told “the security with these cards has never failed so it must have been stolen”, which she knew was untrue as she still had it in her hands. After much argument it turned out that the old card had *not* been cancelled, so the woman went through normal life unknowingly having a pair of duplicate cards, then didn't notice when one was stolen…

Please report problems with the web pages to the maintainer

Top