Start Up No.1476: might Dunning-Kruger be fake?, SolarWinds hack hits US court system, what LinkedIn is really good for, and more


Tesla has copied an element of the KITT Knight Rider car – unfortunately, the least ergonomic one. CC-licensed photo by Dan Thornton on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

A selection of 10 links for you. Mostly harmless. I’m @charlesarthur on Twitter. Observations and links welcome.

The Dunning-Kruger effect is probably not real • Office for Science and Society, McGill University

Jonathan Jarry:

»

I want the Dunning-Kruger effect to be real. First described in a seminal 1999 paper by David Dunning and Justin Kruger, this effect has been the darling of journalists who want to explain why dumb people don’t know they’re dumb. There’s even video of a fantastic pastiche of Turandot’s famous aria, Nessun dorma, explaining the Dunning-Kruger effect. “They don’t know,” the opera singer belts out at the climax, “that they don’t know.”

I was planning on writing a very short article about the Dunning-Kruger effect and it felt like shooting fish in a barrel. Here’s the effect, how it was discovered, what it means. End of story.

But as I double-checked the academic literature, doubt started to creep in. While trying to understand the criticism that had been leveled at the original study, I fell down a rabbit hole, spoke to a few statistics-minded people, corresponded with Dr. Dunning himself, and tried to understand if our brain really was biased to overstate our competence in activities at which we suck… or if the celebrated effect was just a mirage brought about by the peculiar way in which we can play with numbers.

…In 2016 and 2017, two papers were published in a mathematics journal called Numeracy. In them, the authors argued that the Dunning-Kruger effect was a mirage. And I tend to agree.

The two papers, by Dr. Ed Nuhfer and colleagues, argued that the Dunning-Kruger effect could be replicated by using random data.

«

😱😱😱

unique link to this extract


Feds want to talk to Tesla about the Model S ‘yoke’ steering wheel • Roadshow

Sean Szymkowski:

»

The National Highway Traffic Safety Administration told Roadshow on Friday it has reached out to Tesla following the news of its planned yoke-style steering wheel for the refreshed Model S electric sedan. The government agency did not say if the automaker has been in touch with regulators since it debuted the radical new wheel.

Tesla this week revealed the refreshed flagship sedan, along with a revamped Model X SUV, with the steering yoke grabbing eyeballs across the internet. Roadshow’s Editor-in-Chief Tim Stevens has laid out why the design is a likely safety risk, but US regulators will certainly get to the bottom of it. NHTSA told Roadshow that on first glance it “cannot determine if the steering wheel meets Federal Motor Vehicle Safety Standard.” If Tesla’s in violation of the standards, the yoke will have to go.

«

I heard this being discussed on the most recent Accidental Tech Podcast, not having seen the wheel (it’s basically a rectangle), and couldn’t believe that Tesla could have been so stupid. But apparently it has. In user interface terms, it’s, well, a car crash. There isn’t even a stalk for triggering the turn indicators, and the buttons that do it are now on the steering wheel on the same side for both right and left. Unbelievable that any of this could get through a usability test of any sort.
unique link to this extract


Russian hack brings changes, uncertainty to US court system • Associated Press

Maryclaire Dale:

»

Trial lawyer Robert Fisher is handling one of America’s most prominent counterintelligence cases, defending an MIT scientist charged with secretly helping China. But how he’ll handle the logistics of the case could feel old school: Under new court rules, he’ll have to print out any highly sensitive documents and hand-deliver them to the courthouse.

Until recently, even the most secretive material — about wiretaps, witnesses and national security concerns – could be filed electronically. But that changed after the massive Russian hacking campaign that breached the U.S. court system’s electronic case files and those of scores of other federal agencies and private companies.

The new rules for filing sensitive documents are one of the clearest ways the hack has affected the court system. But the full impact remains unknown. Hackers probably gained access to the vast trove of confidential information hidden in sealed documents, including trade secrets, espionage targets, whistleblower reports and arrest warrants. It could take years to learn what information was obtained and what hackers are doing with it.

It’s also not clear that the intrusion has been stopped, prompting the rules on paper filings. Those documents are now uploaded to a stand-alone computer at the courthouse — one not connected to the network or Internet. That means lawyers cannot access the documents from outside the courthouse.

«

This hack is putting the US back years. It’s amazing.
unique link to this extract


Abusing LinkedIn for better customer service • Terence Eden’s Blog

From 2019, but Eden says he’s still using it:

»

When I have a complaint about a company, and regular customer services just can’t fix it, I cheat. I send a connection request [on LinkedIn, not Twitter] to the CEO, or head of customer service, or anyone senior who looks like they might actually hold some sway.

After a month of my energy company sending me incorrect bills, and several hours on hold, I cracked and connected to someone senior there.

A few messages later, it was all sorted.

Similarly, when an employer’s payroll company started messing me around, I went straight to the top. In this case, an executive had posted several times about their “award winning” team. So I left comments on their post asking if my poor experience with their company was typical of their service. Within a moment, I had a response.

«

Smart understanding of the real dynamics of the social networks.
unique link to this extract


Analysis: Robinhood and Reddit protected from lawsuits by user agreement, Congress • Reuters

Tom Hals:

»

Robinhood is not legally bound to carry out every trade and the lawsuits will not succeed without evidence the company restricted trading for an improper reason, such as to favor certain investors, according to several legal experts.

The user agreement on Robinhood’s website says it “may at any time, in its sole discretion and without prior notice to Me, prohibit or restrict My ability to trade securities.”

Adam Pritchard, a professor at the University of Michigan Law School, said the lawsuits are very unlikely to gain traction.

“The contract says they can do it,” Pritchard said of the company’s decision to restrict trading. “That seems to be a big stumbling block to the breach of contract claim.”

«

You knew it would be in there somewhere, and so it is.
unique link to this extract


How I learned to stop worrying and love Twitter’s Trump ban • The New York Times

Farhad Manjoo:

»

instead of worrying about the precedent Twitter set by “permanently suspending” the president, I lament all that might have been if Trump had been banned sooner. And I believe there’s only one lesson Silicon Valley’s luminaries should take away from Trump’s calamitous time on Twitter — it is a cautionary tale about all that can go wrong when digital innovation is stretched far beyond its original purpose.

Twitter was never meant to be used the way Trump used it, as an all-purpose bullhorn for governing a superpower, and as a result the company often found itself having to create special rules just for the president. Ordinary users might be punished for harassment or threats of violence, for instance, but Twitter gave “world leaders” like Trump special permission to behave badly. The company reversed course only after Trump’s supporters stormed the Capitol; Trump’s tweets following the riot risked “further incitement of violence,” it said.

Trump’s case may be unique, for now, but why risk a replay? If I ran the company I’d push to make sure Twitter can never be used the way Trump used it.

The rule need not be partisan. Something far simpler would suffice: heads of state should not be allowed to tweet. The most powerful person in the world’s most visible lever on power should not be 280-character chunks of tossed-off thoughts published instantly, without review, on a medium run by a private company whose secret algorithms are designed to encourage outrage and reward cheap dunks.

«

Trump was an exception, but not a lone one. The benefit of this is that at least it doesn’t require any judgement by Twitter. Feels a bit brutal to lump Jacinda Ardern in with Trump, though.
unique link to this extract


We spoke to a guy who got his dick locked in a cage by a hacker • Vice

Lorenzo Franceschi-Bicchierai:

»

Sam Summers was sitting at home with his penis wrapped in an internet-connected chastity cage when he got a weird message on the app that connects to the device. Someone told him they had taken control and they wanted around $1,000 in Bitcoin to give control back to Summers. 

“Initially, I thought it was my partner doing that,” Summers told Motherboard in a phone call. “It sounds silly, but I got a bit excited by it.” 

But when Summers called his partner, she told him it wasn’t her, even after he told her their safe word. That’s when he realized he had gotten hacked. His penis was locked in the cage, and he had no way out.

“Oh, shit, it’s real,” Summers said. “I started looking at the thing. There’s no manual override at all. It’s a chastity belt, I guess it kind of shouldn’t [have an override.] But when it’s a digital thing like that, it should have a key or something. But it obviously didn’t.”

«

A topic that I covered a while back, but here’s the interview with a victim. Appropriate that it should appear in Vice.
unique link to this extract


‘Carbon-neutrality is a fairy tale’: how the race for renewables is burning Europe’s forests • The Guardian

Hazel Sheffield:

»

To investigate the subsidised European pellet trade and its impact on Baltic forests, we uploaded boundary files for Estonia’s Natura 2000 zones to Global Forest Watch, an online platform for monitoring forests, and found that per-hectare tree cover loss (the removal of the tree canopy rather than outright deforestation) in these areas accelerated after 2015. That was when the government adjusted park conservation rules to allow clear-cutting of up to one hectare at a time in some nature reserves.

Across Estonia, between 2001 and 2019, Natura 2000 areas lost more than 15,000 hectares (37,000 acres) of forest cover, an area more than twice the size of Manhattan. The last five years account for 80% of that loss. Further alterations to rules in other Estonian national parks are planned.

This acceleration appears to be taking a toll on bird species like the black grouse, woodlark and others. Woodland birds are declining at a rate of 50,000 breeding pairs a year, according to national records.

The clearances are also damaging the ability of Baltic forests to store carbon, and could be undermining climate goals by reducing the chance for Estonia and Latvia to achieve net-zero greenhouse gas emissions.

In a country where the overwhelming majority of people say they regard nature as sacred, logging has led to protests or what the Estonian media calls the “forest war”. Residents of Saku, a small town 16 miles south of Tallinn, successfully fought to save an area of forest that was scheduled to be cut down this year by RMK, the state forest management company, which manages around half of Estonian forests

«

unique link to this extract


Here’s a way to learn if facial recognition systems used your photos • The New York Times

Cade Metz and Kashmir Hill:

»

When tech companies created the facial recognition systems that are rapidly remaking government surveillance and chipping away at personal privacy, they may have received help from an unexpected source: your face.

Companies, universities and government labs have used millions of images collected from a hodgepodge of online sources to develop the technology. Now, researchers have built an online tool, Exposing.AI, that lets people search many of these image collections for their old photos.

The tool, which matches images from the Flickr online photo-sharing service, offers a window onto the vast amounts of data needed to build a wide variety of A.I technologies, from facial recognition to online “chatbots.”

“People need to realize that some of their most intimate moments have been weaponized,” said one of its creators, Liz O’Sullivan, the technology director at the Surveillance Technology Oversight Project, a privacy and civil rights group. She helped create Exposing.AI with Adam Harvey, a researcher and artist in Berlin.

«

unique link to this extract


Why webcams aren’t good enough • Reincubate

Jeff Carlson:

»

Primarily this is about size: webcams are designed as small devices that need to fit onto existing monitors or laptop lids, so they use small camera modules with tiny image sensors. These modules have been good enough for years, generating accolades, so there’s little incentive to change. The StreamCam appears to have a better camera and sensor, with an aperture of f/2.0; aperture isn’t listed for the other cameras.

Contrast this technology with the iPhone, which also includes small camera modules by necessity to fit them into a phone form factor. Apple includes better components, but just as important, incorporates dedicated hardware and software solely to the task of creating images. When you’re taking a photo or video with an iOS device, it’s processing the raw data and outputting an edited version of the scene.

Originally, Logitech’s higher-end webcams, such as the C920, also included dedicated MPEG processing hardware to decode the video signal, but removed it at some point (without changing the model numbers or otherwise indicating the change except for an undated blog post). The company justified the change because of the power of modern computers, stating, “there is no longer a need for in-camera encoding in today’s computers,” but that just shifts the processing burden to the computer’s CPU, which must decode raw video instead of an optimized stream. It’s equally likely Logitech made the change to reduce component costs and no longer pay to license the H.264 codec from MPEG LA, the group that owns MPEG patents.

That brings us to the other factor keeping webcam innovation restrained: manufacturers aren’t as invested in what has been a low margin business catering to a relatively small niche of customers.

«

Easy to forget that for most PC companies, making PCs isn’t actually a high-margin business: when I looked into it in detail in 2014, the top five PC companies were making operating margins of about $15 per PC sold. No hurry to spend more than that extra on the webcam. (Link via John Naughton.)
unique link to this extract


Errata, corrigenda and ai no corrida: none notified

4 thoughts on “Start Up No.1476: might Dunning-Kruger be fake?, SolarWinds hack hits US court system, what LinkedIn is really good for, and more

  1. Wait, isn’t the statistical discussion in fact giving more evidence that the Dunning-Kruger effect is *real*, by showing it directly falls out of a very simple mathematical model? Consider two statements:

    1) People who are terrible at something often think they are much better at it than they really are.

    2) People’s assessment of their skill at something is indistinguishable from taking their “real” skill plus a large *completely random* factor.

    Why is 2 considered to be refuting 1? It seems to me more at proving it.

    If someone is terrible at a skill, and self-assessment value is (“real” plus large random), then the bottom of the barrel has nowhere to go overall but much higher than true value.

    The psychological issue is why the terrible at a skill group uses the large random factor, i.e. why don’t they know they are so bad at it?

    “For an effect of human psychology to be real, it cannot be rigorously replicated using random noise.” – err, why not? If the effect is that one’s estimate about something is no better than random noise, then this is begging the question.

  2. Was wondering how such an apparently counter-intuitive steering wheel and indicator controls as the new Tesla’s could get past safety regulators – maybe it won’t.

    Yet there are many examples of bad design that have already been allowed – the preponderance of touch controls in modern cars seems a clear backwards step. And for years I’ve wondered how less obvious indicator lights on cars were ever allowed (in the UK & Ireland at least) for seemingly nothing more than aesthetic reasons.

    Re: Dunning-Kruger – what seems dumb to me (and I know Dr Dunning specifically said the model was about all of us – not just the ‘dumb’) is the notion of there being any value in creating a model for something so self-evident; that people with little knowledge in a subject will likely be unaware of their lack of knowledge and, given human psychology, be loath to recognise that.
    But I may well be proving the model with my opinion 😃

Leave a Reply to Seth Finkelstein Cancel reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.