The UK’s liver transplant matching algorithm discriminated against young people

Here's how a 2003 BBC article starts:

The system for allocating most liver transplants on the NHS is causing younger patients to wait longer for surgery, figures show.

There is a shortage of liver donors, so a computer algorithm decides who to prioritise on the waiting list.

Younger people are currently waiting 156 days longer on average for a transplant than patients over 60.

But NHS Blood and Transplant (NHS BT) says difficult decisions have to be made and the system is saving lives. 

The AI Snake Oil blog discusses the algorithm (and other algorithms) in depth:

In November 2023, the Financial Times published a bombshell investigation about bias in the algorithm. It centers on a 31 year old patient, Sarah Meredith, with multiple genetic conditions including cystic fibrosis. It describes her accidental discovery that the Transplant Benefit Score algorithm even existed and would decide her fate; her struggle to understand how it worked; her liver doctors’ lack of even basic knowledge about the algorithm; and her realization that there was no physician override to the TBS score and no appeals process.

When she reached out to the National Health Service to ask for explanations, Meredith was repeatedly told she wouldn’t understand. It seems that the paternalism of health systems combined with the myth of the inscrutability of algorithms is a particularly toxic mix.

Meredith eventually landed on a web app that calculates the TBS, built by Professor Ewen Harrison and his team. He is a surgeon and data scientist who has studied the TBS, and is a co-author of a study of some of the failures of the algorithm. It is through this app that Meredith realized how biased the algorithm is. It also shows why the inscrutability of algorithmic decision making is a myth: even without understanding the internals, it is easy to understand the behavior of the system, especially given that a particular patient only cares about how the system behaves in one specific instance.

But this isn’t just one patient’s experience. From the Financial Times piece:

“If you’re below 45 years, no matter how ill, it is impossible for you to score high enough to be given priority scores on the list,” said Palak Trivedi, a consultant hepatologist at the University of Birmingham, which has one of the country’s largest liver transplant centres.

Finally, a 2024 study in The Lancet has confirmed that the algorithm has a severe bias against younger patients.

Those Columbus headlines today sound like pseudo science at best

You can read the BBC article and see how little it's based on, and per this write-up, it's based on even less than that. Perfect fodder for modern "history" channels on basic cable.  

Related:

In recent days, a number of news sites that rely heavily on aggregation have posted stories about [the Democrat vice presidential candidate] reporting “allegations” that he sexually assaulted a minor while working as a teacher and football coach.

The clearly false claims stem from the prolific work of one man, a Twitter conspiracy peddler [with a blue check]. 

...

In September, he promoted an obviously fake story about a “whistleblower” at ABC News anonymously claiming the presidential debate hosted by the channel had been biased in favor of Kamala Harris. To back up the claims, he published a purported affidavit by the whistleblower, a poorly formatted and typo-riddled document

"The firm run by a man dubbed 'Britain's kindest plumber' faked stories of helping people as it raised millions in donations"

BBC:

[He] has received letters of thanks from the late Queen and the King and a Pride of Manchester award. He has been a guest on Good Morning Britain, BBC Breakfast, The One Show, Sky News and the Russell Howard Show

...

[In one example, he] prevented one elderly woman from killing herself, the firm claimed. In fact, she had died years earlier.

He denied some of the BBC's allegations but admitted: "I've made mistakes."

Related:

The story of how Facebook became engulfed in Christian content can be told through the rise of one site, catholicfundamentalism.com. 

...

According to our data, provided by Newswhip’s real-time Facebook analytics, the vacuum of news content Meta created last year was quickly filled by religious content, or even just content that provoked vaguely religious reactions from users. 

...

As for why Catholic Fundamentalism is now Facebook’s top “news” site, it posts the exact kind of content that the platform’s news feed has been tailored to promote: blandly positive, removed from any news or current events, and arriving on a frequent and regular schedule. 

Spotify "removed tens of thousands of songs from artificial intelligence music start-up Boomy" (but apparently not because they're AI-generated)

Ars:

Boomy, which was launched two years ago, allows users to choose various styles or descriptors, such as “rap beats” or “rainy nights,” to create a machine-generated track. Users can then release the music to streaming services, where they will generate royalty payments. California-based Boomy says its users have created more than 14 million songs.

...

Spotify, the largest audio streaming business, recently took down about 7 percent of the tracks that had been uploaded by Boomy

...

The Boomy songs were removed because of suspected “artificial streaming”—online bots posing as human listeners to inflate the audience numbers for certain songs.

...

Boomy at the weekend resumed submitting new tracks to Spotify. The two sides are negotiating

Algorithm Heaven: Feel the Rhythm & Feed the Algorithm

Another good Famicase 20203 design:

What is this game about? Rhythm? Strategy? It’s not clear. We only know two things:

1. The algorithm is hungry.
2. We don’t want to know what happens if we don’t feed the algorithm.

Don’t stop dancing now, but this could be a survival horror game.#famicase pic.twitter.com/NJst1tOvZH

— Sebastià López (@daruma_studio) April 28, 2023

"Facebook's Ad System Went Nuts and Ripped Off Customers"

Gizmodo has one write-up, and here's another. (Not super clear to me how many people were defrauded, or by how much, or whether they'll get their money back.)

Logo for the European Centre for Algorithmic Transparency

🔚 Enhorabuena a @EU_ScienceHub y la @ComisionEuropea por el Centro 🇪🇺 de Transparencia Algorítmica #ECAT. Y gracias por ubicar en 🇪🇸 una sede fundamental en la implantación de la #DigitalServicesAct.

Let’s open the black box 🔓 pic.twitter.com/FGTaMmKTsw

— S.E. Digitalización e Inteligencia Artificial (@SEDIAgob) April 18, 2023

ECAT:

Algorithmic systems determine many aspects of our online experience, for example, a music streaming app may use algorithms to suggest songs or bands to its users. With the ever-increasing societal impact of online platforms such as social networks, online marketplaces, and search engines, there is an urgent need for public oversight of the processes at the core of their business. This includes in particular how these platforms and search engines moderate content and how they curate information for their users.

Mission

The European Centre for Algorithmic Transparency (ECAT) will contribute to a safer, more predictable and trusted online environment for people and business.

How algorithmic systems shape the visibility and promotion of content, and its societal and ethical impact, is an area of growing concern. Measures adopted under the Digital Services Act (DSA) call for algorithmic accountability and transparency audits.

The ECAT is part of the European Commission, hosted by the Joint Research Centre (JRC) -the Commission’s in-house science and knowledge service- in close cooperation with the Directorate General Communications Networks, Content and Technology (DG CONNECT). It contributes scientific and technical expertise to the Commission's exclusive supervisory and enforcement role of the systemic obligations on Very Large Online Platforms (VLOPs) and Very Large Online Search Engines (VLOSEs) provided for under the DSA.

Related, in Los Angeles we have USC's "meme team":

I like the different strategies websites are using to pad out their articles about the escaped wolf being recaptured

They have no information other than a wolf "escaped" and was recaptured at the zoo, but have to pump up that word count.

WaPo reposting the AP:

The zoo’s website says the Mexican gray wolf, canis lupus baileyi, is the smallest of the gray wolf subspecies, 54 to 66 inches (137 to 167 centimeters) in length from snout to tail and weighing 50 to 90 pounds (22 to 40 kilograms). Their coat is a varied mixture of tan, red, white, and black fur, with darker colors on their head, shoulders and back.

Cleveland 19:

The Zoo’s wolves, Catori, Sarita, Una, Aprecia, Nancita and Mitzi, are all sisters. They came to Cleveland from the Wild Canid Survival and Research Center in Missouri in 2009.

The Wolf Lodge is also home to the Zoo’s bald eagle, beaver and several species of North American fish.

Here are magic incantations you can try to thwart Twitter's recommended tweet algorithm

Thank you! I typed these out if anyone wants to copy:
suggest_recycled_tweet_inline
suggest_activity_tweet
suggest_who_to_follow
suggest_recap
suggest_recycled_tweet
suggest_pyle_tweet
suggest_ranked_timeline_tweet

— mitch (@p2isonline) February 6, 2022

Funny (in a nightmarish way) thread describing an attempt to let the Disney Genie app plan a day at Disneyland

Like Amazon search results, Instagram feeds, and Spotify playlists, adversarial to the user's stated desire. Thread starts here.

Genie has finally noticed that Pinocchio’s Daring Journey is temporarily closed and revised my itinerary to direct me to Casey Jr., another ride that is temporarily closed. You simply cannot make this up. Battin’ .1000, Genie!! pic.twitter.com/fgr60OHHax

— darren, a dog liker (@tai_nugget) December 12, 2021

I can’t be the only one who finds it a little gross that Genie is adding two upsells to my itinerary that I not only didn’t reserve but that also involve dropping serious coin. pic.twitter.com/5JXDwLduPw

— darren, a dog liker (@tai_nugget) December 12, 2021

After picking up my mobile order and refreshing, the Rise suggestion is now nowhere to be seen (perhaps Genie thinks I did it already?), and my midday slate is ready. I have been in the park 4 hours and I have not yet gone on one of the 5 rides I told Genie I was interested in. pic.twitter.com/n8gpJwtfzA

— darren, a dog liker (@tai_nugget) December 12, 2021

Today's funny posts

I love these swooping magpie signs. 'Try a different route next time'. Like they're basically saying 'yeah nah this area's stuffed, belongs to the magpies, better abandon it'. pic.twitter.com/qABOu1XIDO

— Kate Walton (@waltonkate) October 29, 2021

The "mamasphere is a series of interactions between human and non-human actors"

Anne Helen Petersen's latest newsletter features an interview with Kathryn Jezer-Morton, who is writing a doctoral dissertation on mombloggers:

I’ve learned that a lot of moms are increasingly planning their content around what the algo rewards. It brings to mind actor-network theory -- the idea that the mamasphere is a series of interactions between human and non-human actors. Time to read more Donna Haraway! The algorithm is absolutely an actor here. Momfluencers have to try to decipher its preferences and react accordingly. Which is part of why sometimes everyone in the mamasphere appears to be doing exactly the same thing, like a bunch of tall-boot-wearing synchronized swimmers. Like using letterboards, for example. They are doing something that they have found to be effective for the success of their accounts, but probably won’t be effective for long.

...

My favorite tropes, oh boy. I am fascinated by momfluencers’ home interiors, the constant upgrading and renovating and reorganizing of nurseries and bedrooms and the requisite “reveals.” A few years ago Amber Fillerup Clark revealed her new laundry room and it contained TWO washing machines side by side and my soul temporarily left my body. Taza of Love Taza has a row of fridges with, if I am not mistaken, glass doors like a bodega beer fridge. I am riveted by this. Interiors were really not part of mommyblogs, but of course they are Instagram content mines. There does seem to be an interiors arms race, which mirrors the general pressure within capitalism to continuously “level up,” and when we bring that pressure into the home, it’s worth paying attention to what happens next. Also, kids grow up and sometimes don’t want to be photographed anymore, but interiors are always fair game. I think as momfluencers work to diversify their content in preparation for their children growing up — which is happening more and more, as the first big wave of momfluencers’ kids reach adolescence — we might see more emphasis on home decor.

1602 Spider-Man cosplay; Grant Morrison on internet trolls; Tiktok's algorithm basically produces reality television

Hey! My design for Spider-man 1602 has real image in cosplay!#spiderman1602 @Marvel pic.twitter.com/VE6OyxelLo

— Ramon Rosanas (@RamonRosanas) August 17, 2021

#ncbdspoilers Genuinely crying with laughter over “Grant Morrison riffs on the Internet” pic.twitter.com/wkQ4rvZ4X7

— Magneto’s Best Pal, Peepers (@excel522) August 17, 2021

It's still amazing to me that the TikTok algorithm essentially produced and distributed an entire reality TV showhttps://t.co/EFbXTnkQfo

— Ranjan (@ranjanxroy) August 17, 2021

it created the drama, guided the rushees to understand exactly what kind of content to generate, made sure it found an audience, promoted the reaction videos - all without (I'm assuming) any human intervention.

— Ranjan (@ranjanxroy) August 18, 2021

Virgin Denis Villeneuve vs Chad Abel Ferrara https://t.co/XteTEjYQt0 pic.twitter.com/znb7bU50LR

— John Frankensteiner (@JFrankensteiner) August 18, 2021

The hot algorithm trick is tweeting "logo for my business" to attract attention

Ryan Broderick's GarbageDay:

On Wednesday, I wrote about Twitter users having a botnet post gibberish in the replies of a tweet to artificially boost its engagement. I wrote that I assumed the tweet was targeted by bots to make it appear more prominently in Twitter’s trending content widgets. But a few readers messaged me to say it’s actually a bit more complicated than that.

I was told me that what was actually happening is that users are posting things like “logo for my business” to attract the attention of bots that would then boost the engagement of the thread.

So I tried it out as an experiment, I tweeted that I needed a logo for my business. I received a BUNCH of replies and because I have open DMs, a bunch of messages, as well. I actually messaged one of the “bots” back and discovered they weren’t a bot at all, actually.

...

What actually may have happened was users wanted to boost the visibility of a tweet, so they filled up the replies with requests for a “logo” or “essay,” which then pinged thousands of anonymous freelancers looking for work, who then replied to these requests, thus boosting the tweet’s placement in Twitter’s trending algorithm.

TikTokers spread a baseless rumor that a company selling glowsticks on Amazon was involved in human trafficking

Mashable:

The conspiracy theorists posed this question to their audience: Who would buy a party hat or glow stick or bubble gun for thousands of dollars?

...

These TikTok video creators missed a crucial detail on these Amazon listings. These were prices for bulk orders — and the pricing would change for each listing based on the quantity a user selected. 

...

 "I thought, it's TikTok, let's just kind of let this all blow over," said Momin, recounting his reaction when the messages first started coming in. "Then that one user got millions of views. It's obvious he was depicting the product incorrectly and just putting conspiracies out there." 

...

"We had seven police officers come [to our warehouse] in a span of three days, back to back to back," Momin tells me. "I told the officers, 'Look, you can come in any time and just walk around in the warehouse and do what you've got to do, but just don't bother us anymore.' It was getting to a point where it was ridiculous. And then we had the commissioner come and he apologized. They said they received calls."

Ghost kitchen companies are now competing to game the algorithm on food delivery services

Long article at the Verge about restaurants that sometimes house 12 different brands, and the various companies competing to dominate delivery service orders:

“With the virtual brands thus far, a lot of it really boils down to search optimization,” says Melissa Wilson, a principal at the food service consulting company Technomic. Just as the incentives and constraints of Google search or Facebook’s News Feed gave rise to certain emergent styles — the keyword-crammed headline, the clickbait tease — digital brands are converging on a distinctive form. Because people search for food on delivery apps in much the same way they search for anything else online — by product type rather than brand — specific restaurant names like The Bergen, named after the street it’s on, or even Denny’s and Red Robin, are too opaque. 

...

Soon, he was selling burgers under the names of Chef Burger, Burger Mansion, Hey Burger, and MrBeast Burger. Wings were sold under the names CHICKS, Wild Wild Wings, Crispy Wings, Killer Wings, Firebelly Wings, and The Wing Dynasty.

...

“Instead of saying buffalo wings, they might say classic wings,” he goes on. “It’s the same exact wing, man. 

The gross viral food videos are connected to a magician that's skilled at Facebook's algorithm

Ryan Broderick for Eater:

It is, at this point, statistically impossible to not half-recognize magician Rick Lax. He is essentially the face of Facebook’s Watch program. Launched in the U.S. in 2017, Watch was meant to be Facebook’s Netflix or YouTube competitor. 

...

Satire or not, the main refrain when content from Lax’s collaborators goes viral is based on some joke about how white women don’t know how to cook. It’s an undeniable part of the appeal of hate-sharing these videos. And it’s something Lax said he’s aware of.

...

But whether he and his creators are doing it on purpose or not, the videos Lax is sharing across his absolutely massive pages all present a very specific middle-class, suburban portrait of the world, where beautiful, young newlyweds play sexy pranks and do nice things for older people. Where construction workers, who are played by people of color, defend white women from street harassment. They’re the kind of videos that do well with Facebook’s overwhelmingly conservative user base.

The people posting slander online and the people who help remove it are often one and the same

From a long article at the NYTimes, detailing the shady characters contacted.

The businesses that say they can remove the libelous posts are sometimes owned by the same business owners hosting the posts. Other times it's arms-length--the removal service pays to advertise on the libel-hosting site and pays the site to remove the libel. And paying a removal service might just tell them that if they repost the libel somewhere, they can be confident you'll pay more to take it down. It all works because Google and Bing highly promote the sites in search results:

For about one-third of the people, the nasty posts appeared on the first pages of their results. For more than half, the gripe sites showed up at the top of their image results.

...

Sometimes search engines go a step further than simply listing links; they display what they consider the most relevant phrases about whatever you’re searching for.

One woman in Ohio was the subject of so many negative posts that Bing declared in bold at the top of her search results that she “is a liar and a cheater” — the same way it states that Barack Obama was the 44th president of the United States. For roughly 500 of the 6,000 people we searched for, Google suggested adding the phrase “cheater” to a search of their names.

There is some hope of getting such results removed:

There is another way to lessen the posts’ impact. In certain circumstances, Google will remove harmful content from individuals’ search results, including links to “sites with exploitative removal practices.” If a site charges to remove posts, you can ask Google not to list it.

Google didn’t advertise this policy widely, and few victims of online slander seem aware that it’s an option. That’s in part because when you Google ways to clean up your search results, Google’s solution is buried under ads for reputation-management services like RepZe.

...

Other people who have used Google’s form reported similar experiences: It mostly works, but is less effective for images. And if you have an attacker who won’t stop writing posts about you, it’s almost useless. 

Tiktok's automatic translator transforms nonsense syllables into disturbing phrases

This is legit so weird and terrifying can someone who is an expert explain how this works? https://t.co/4CqOLFtLo9 via @alisedaandrea pic.twitter.com/4ByCp7rk1U

— Taylor Lorenz (@TaylorLorenz) January 3, 2021

Here's a 2018 article from Vice, when Google Translate was observed doing something similar:

Type the word “dog” into Google Translate 19 times, request that the nonsensical message be flipped from Maori into English, and out pops what appears to be a garbled religious prophecy.

“Doomsday Clock is three minutes at twelve,” it reads. “We are experiencing characters and a dramatic developments in the world, which indicate that we are increasingly approaching the end times and Jesus' return.”

...

a senior scientist at BBN Technologies who works on machine translation, agreed that strange outputs are probably due to Google Translate’s algorithm looking for order in chaos. He also pointed out that the languages that generate the strangest results—Somali, Hawaiian and Maori—have smaller bodies of translated text than more widely spoken languages like English or Chinese. As a result, he said, it’s possible that Google used religious texts like the Bible, which has been translated into many languages, to train its model in those languages, resulting in the religious content.

Instagram’s algorithm will either reward or punish based on how comprehensively you use the app

Here's one artist describing what they believe you need to do to get Facebook to boost your Instagram account:

Instagram likes it when you are spending all your time on the app :) and using all of their available features aside from normal feed posts :) They like it when you post a combination of normal posts, reels, IGTV, stories, shoppable posts, etc etc. You know those little events they do in stories like the “I Voted” sticker? participate in that school spirit kinda junk. The algorithm likes it when you use the in app camera and filters, geolocation tagging, messaging, story buttons, all that stuff. 

...

• 3 feed posts per week (reels & IGTV you choose to post to feed count here)

• 8-10 stories per week, preferably 2+ per day

• 4-7 reels per week

• 1-3 IGTV a week

Basically,

Man IG really said "we'll give you free exposure if you work for us full time" and we fell for it

— Jen Bartel (@heyjenbartel) December 22, 2020