Connect with us

Technology

Facebook job ads algorithm still discriminates on gender, LinkedIn not so much • The Register

Voice Of EU

Published

on

Two years after Facebook settled five lawsuits claiming that its employment, housing, and credit ads illegally discriminate, researchers with the University of Southern California have found that the company still serves job ads unfairly, based on gender.

In a paper titled “Auditing for Discrimination in Algorithms Delivering Job Ads,” scheduled to appear at The Web Conference later this month, Basileal Imana, a doctoral student at USC, Aleksandra Korolova, USC assistant professor of computer science, and John Heidemann, USC research professor of computer science, explore bias in algorithmic job ad delivery at Facebook and LinkedIn.

Korolova, in an email to The Register, explained that since US law allows for ad delivery to be differentiated on the basis of qualifications, she and her colleagues developed a way to test for bias while factoring out lawful qualification-based biasing.

“Even when controlling for job qualifications, Facebook introduces a delivery skew by gender for job ads with balanced targeting,” Korolova said, noting that this advances the argument that “Facebook’s ad delivery algorithms are not merely biased but actually discriminatory.”

“Interestingly, we did not find such an effect when auditing LinkedIn’s algorithms,” she added.

In 2019, Korolova was among a different set of academics who, shortly after Facebook settled the above mentioned civil rights lawsuits and announced changes to combat discriminatory advertising, found biased behavior in Facebook’s ad delivery attributable to ad budgets and ad content.

Time to check again

This time, Korolova and her colleagues have looked at how Facebook and LinkedIn’s algorithmic ad platforms skew job ads by delivering them to viewers identified as male and female – where that data is available – in a ratio that differs from the expected gender distribution for the job.

They managed this by comparing the performance of two ads in three different job categories – delivery driver, software engineer, and sales associate – with known differences in gender distribution. They then weighed the expected ratio against the actual gender ratio among Facebook and LinkedIn ad recipients.

One such ad pair consisted of an ad to be a delivery driver for Domino’s Pizza (98 per cent male) and an ad to be a delivery driver shuttling groceries for Instacart (more than 50 per cent female).

image of race with one running slowed by gender

Four women seek release from forced arbitration to sue Infosys for widespread gender discrimination

READ MORE

“The de facto gender distribution among drivers of these services is skewed male for Domino’s and skewed female for Instacart,” the paper explains.

“If a platform shows the Instacart ad to relatively more women than a Domino’s ad, we conclude that the platform’s algorithm is discriminatory, since both jobs have similar qualification requirements and thus a gender skew cannot be attributed to differences in qualifications across genders represented in the audience.”

The researchers found “a statistically significant gender skew on Facebook, and show no gender skew on LinkedIn.”

For software engineering job ads, the researchers choose recruitment pitches for Netflix (35 per cent female) and Nvidia (19 per cent female). They expected that an ad platform using an algorithm that learns and incorporates existing differences in employee demographics would show the Netflix job ad to more women than the Nvidia job ad and they were not surprised.

Facebook again skewed its ad distribution by gender; LinkedIn did not.

For sales associate positions, job ads for Reed Jewelers (jewelry sales being 62 per cent female per federal job statistics) and Leith Automotive (auto sales being 17.9 per cent female) were compared.

Again, this job ad category produced similar results to the previous ones: There was “statistically significant delivery skew between all jobs on Facebook but not for two of the three cases on LinkedIn.”

Facebook’s skewing of ads by gendering cannot be explained by differences in qualifications, the researchers argue, noting that their findings suggest “that Facebook’s algorithms may be responsible for unlawful discriminatory outcomes.”

According to Korolova, Facebook was informed of the researchers’ findings and has not responded. The Register asked Facebook for comment, but we’ve not heard back.

Something’s up

Asked why Facebook’s algorithm behaves differently from LinkedIn’s, Korolova proposed several possibilities.

“Facebook may have more sources of data about the users than LinkedIn, which enables them to better pick up on existing real-world skews,” she suggested. “Facebook may give higher weight to the engagement estimates in its ad delivery algorithms than LinkedIn. LinkedIn may make a deliberate effort in their algorithms to ensure fairness.”

She also allowed that the research methodology used might be insufficient to analyze LinkedIn’s algorithm, but noted that LinkedIn appears to have made a concerted effort to address algorithmic fairness.

Korolova said she and her colleagues would not presume to propose an optimal way to present job ads. “I think our assumption is that the gender of ad recipients should reflect the gender of the target population, as would occur from a naive algorithm of showing ads to all visitors,” she said.

Advertisers interested in increasing employee diversity, she said, “should be able to advertise to a balanced audience, rather than have decisions of who their ads go to be ‘overruled’ by Facebook.”

While acknowledging that outwardly neutral characteristics can reflect bias, citing the example of how home location data captures the bias of historical redlining, Korolova said it isn’t inevitable that Facebook’s ad delivery will be discriminatory.

“That their results are discriminatory in our data is particularly surprising given multiple prior observations that Facebook algorithms lead to skewed delivery and their statements that they were addressing it,” she said.

“With Facebook, these outcomes align with their business model of optimizing for advertiser and user ‘value’ (or engagement), suggesting the importance of external evaluation and potential regulation.”

Citing research costs approaching $5,000 in ad fees and a time investment of many months, the researchers argue that ad platforms like Facebook and LinkedIn should make it easier and more affordable to verify that ads comply with anti-discrimination laws. And since these platforms are unlikely to take such steps on their own, they suggest lawmakers should pass legislation to mandate access.

“We would like to make auditing by public interest researchers of Facebook’s ad delivery algorithms feasible, because we believe that will lead to greater transparency about skew and encourage addressing the problem,” said Korolova. “Facebook’s current transparency efforts fall far short of the feasibility goals.” ®

Source link

Technology

Amazon to let Prime users unsubscribe in two clicks to comply with EU rules

Voice Of EU

Published

on

There were complaints of ‘a large number of hurdles’ to unsubscribe from Amazon Prime such as complicated menus, skewed wording, confusing choices and warnings.

Amazon has committed to making it easier for users to cancel their Prime subscription to comply with EU rules.

The tech giant will now let consumers in the EU and EEA unsubscribe from Amazon Prime with just two clicks, using a prominent cancel button.

This came following a dialogue with the European Commission and national consumer protection authorities. Complaints had been issued to the Commission by the European Consumer Organisation, the Norwegian Consumer Council and the Transatlantic Consumer Dialogue.

These consumer authorities noted “a large number of hurdles” to unsubscribe from Amazon’s service, such as complicated navigation menus, skewed wording, confusing choices and repeated nudging.

Amazon made initial changes last year, labelling the cancel button more clearly and shortening the explanatory text. This text will now been reduced further so consumers don’t get distracted by warnings and deterred from cancelling.

“Consumers must be able to exercise their rights without any pressure from platforms,” said EU commissioner for justice Didier Reynders.

“Opting for an online subscription can be very handy for consumers as it is often a very straightforward process, but the reverse action of unsubscribing should be just as easy. One thing is clear: manipulative design or ‘dark patterns’ must be banned.”

Amazon has committed to implementing the new changes on all its EU websites and for all devices. The tech giant will be monitored by the European Commission and national authorities to ensure it complies with EU consumer law.

“Customer transparency and trust are top priorities for us,” an Amazon spokesperson said.

“By design we make it clear and simple for customers to both sign up for or cancel their Prime membership. We continually listen to feedback and look for ways to improve the customer experience, as we are doing here following constructive dialogue with the European Commission.”

Amazon has had a number of dealing with the European Commission over the years regarding its business practices. The tech giant was hit with a Statement of Objections in 2020 based on its use of marketplace seller data.

In 2017, an EU case led by competition commissioner Margrethe Vestager also accused Amazon of cutting an illegal deal with the Grand Duchy of Luxembourg to drastically lower its tax bill.

The country was ordered to recoup €250m in back taxes. However, Amazon won its appeal against this ruling last year, as the EU’s general court said the European Commission didn’t provide the “requisite legal standard” to prove Amazon received favour from tax authorities.

10 things you need to know direct to your inbox every weekday. Sign up for the Daily Brief, Silicon Republic’s digest of essential sci-tech news.

Source link

Continue Reading

Technology

How to read: a guide to getting more out of the experience | Books

Voice Of EU

Published

on

Why read books, in this day and age? “Haven’t we all secretly sort of come to an agreement, in the last year or two or three, that novels belonged to the age of newspapers and are going the way of newspapers, only faster?” wrote Jonathan Franzen, tongue firmly in cheek, in a 2010 essay. The comment feels trenchant a decade later, in an era marked by a saturation of streaming platforms, short-form video, podcasts and screen adaptations of said podcasts.

The proportion of non-readers in Australia has grown in recent years: results of the 2021 National Reading Survey found that 25% of people reported not reading a single book in the previous year – up from 8% in a 2017 survey.

Any bibliophile can easily rattle off a list of reasons for reading. Books enlighten and challenge us, they transport us to different worlds, they reflect essential truths about the human condition.

“People who read well and read a lot learn more, pick up more general knowledge … and can then be better critical consumers of what they read,” says Prof Pamela Snow, co-director of the Science of Language and Reading lab at La Trobe University.

So, within our busy lives, how do we better find the time for books? How can we get more out of the reading experience?

Skim/deep

We commonly interact with texts in different modes. In skimming through an article, taking in a few lines – a headline and subheadings, for example – we might gain a general but shallow understanding of its meaning. We also scan texts for specific numbers, names, or ideas – a quantity in a recipe, say.

Then there’s deep reading, what the scholars Dr Maryanne Wolf and Dr Mirit Barzillai define as “the array of sophisticated processes that propel comprehension and that include inferential and deductive reasoning, analogical skills, critical analysis, reflection, and insight. The expert reader needs milliseconds to execute these processes; the young brain needs years to develop them.”

Reading on screens has turned us into adept text skimmers. An influential 2005 study that analysed how reading behaviour had changed over the previous decade – coinciding with the global rise of the internet – found that online reading was characterised by “more time spent on browsing and scanning, keyword spotting, [and] one-time reading … while less time is spent on in-depth reading, and concentrated reading”.

Wolf has advocated for the need to cultivate a “bi-literate” reading brain, one capable of both deep reading processes and the skim reading more commonly associated with screens.

“Readers must engage in an active construction of meaning, in which they grapple with the text and apply their earlier knowledge as they question, analyse, and probe,” she and Barzillai have suggested. One technique for in-depth reading of narrative texts is RIDA: to Read, Imagine the scene, Describe it to yourself, and Add more mental detail by noting powerful imagery or salient passages.

Woman reading textbooks in library
Our brains should ideally be ‘capable of both deep reading processes and the skim reading more commonly associated with screens’. Photograph: Jacobs Stock Photography Ltd/Getty Images

Physical books, rather than devices like smartphones, tend to support more focused reading, says Baron, though she says the choice of medium is ultimately a matter of personal preference.

Screens themselves are not inherently detrimental to our ability to focus, says the head of the visual and cognitive neuroscience laboratory at the University of Melbourne, Prof Trichur Vidyasagar.

“People often have the belief, particularly concerned parents, that if you spend too much time on screen devices your concentration may get poorer. That’s not necessarily true,” he says. “If used correctly and not at the cost of other useful activities, they can greatly benefit learning.”

The key is the internet’s boundless potential for distraction. “When you use the screen, there are so many hyperlinks, so many sites, stories, and rabbit holes to go into,” Vidyasagar says. The temptation to multitask – “an illusory myth,” he says – can be hard to resist. “If you think you’re multitasking, what you’re actually doing is switching between two tasks at a rapid rate, and your performance in both goes down.”

“When you read a [physical] book it’s quite different – you can’t get distracted as easily.”

Research in university students has found that comprehension is generally higher for print reading. “There is something about reading digitally that seemingly increases the speed at which students move through the text and this processing time translates into reduced comprehension,” one study found. “The findings are especially true when you’re talking about longer materials,” Baron says, adding as a caveat that research tends to focus on academic rather than leisure reading.

Results seem to differ slightly for dedicated e-reader devices. One study, in which participants read a 28-page mystery story by Elizabeth George either in print or on a Kindle, found no differences in most standard comprehension measures. The print readers, however, were better at reconstructing the plot and chronological aspects of the story – potentially because “the physical substrate of paper provides material placeholders” for events within the story.

Rediscovering joy and meaning

Dr Judith Seaboyer, formerly a senior lecturer in literary studies at the University of Queensland, who retired last May, recently went through a fiction dry spell. “There’s so much good stuff to listen to [on the radio], so much good journalism out there to read, and I was finding that I wasn’t reading novels any more.”

“As somebody … who’s done a PhD in contemporary literary fiction, and taught it for over 20 years – you think I’d know [reading books] is worth doing.”

What broke Seaboyer out of her slump was reading new work by an author she loves – Ali Smith’s Companion Piece. Synthesising ideas and making comparisons across multiple texts is also a known strategy for deepening reading comprehension, so some might find it helpful to dig into multiple books by the same author.

Seaboyer’s advice is to read with curiosity and to carefully consider an author’s choices, which can lead to a deeper understanding of language, characters and plot. “Jot things down, annotate your book, write things in the margin,” she says. “Some publishers are putting out reading guides now – that’s often quite useful.”

Nabokov believed that “One cannot read a book: one can only reread it”. For him, revisiting books – like the process of regarding a painting – meant the mind first “takes in the whole picture and then can enjoy its details”.

“You [might] remember that you really loved reading Austen,” Seaboyer says. “It’s interesting to be thinking as you read … now that I’m older and wiser, am I seeing any of this any differently than I did when I was 18?”

“There are ways to be kind to yourself, to allow yourself the opportunity not to understand something the first time through, or to say … maybe there’s a different book I should read first,” Baron says. “It’s like reading James Joyce: if you want to start with Ulysses, good luck. If you start with A Portrait of the Artist as a Young Man, you’ll have a better shot at working your way in.”

Copy of Ulysses by James Joyce
‘It’s like reading James Joyce: if you want to start with Ulysses, good luck.’ Photograph: Martin Argles/The Guardian

If reading solely for pleasure, abandoning books that are not bringing enjoyment could, in fact, increase reading time. Of frequent readers surveyed in 2021 – those who consumed at least one book a month – 54% reported not finishing a book if they disliked it. As a result, they “move[d] on more quickly to the next book for greater enjoyment … and have fewer and shorter gaps between books”.

For those wanting to read more – for relaxation or self-improvement – Baron suggests committing to short but regular periods of reading, similar to time set aside for exercise or meditation.

The speed question

Some people are naturally fast readers – celebrated academic Harold Bloom claimed to be able to read 1,000 pages an hour in his prime. Most adults, according to 2019 analysis, read English nonfiction silently at a rate of between 175 and 300 words a minute, and fiction at a rate of 200 to 320.

While speed reading techniques or apps may seem alluring for the time poor, they’re unlikely to work without compromising understanding.

“Fast readers are not necessarily better at reading comprehension,” Vidyasagar says.

There are no shortcuts to reading faster. Becoming a better reader requires persistence and “dealing with the frustration at not seeing overnight results”, Snow says. “It’s like any skill – learning a musical instrument, learning to drive a car.”

A 2016 review of the science of reading found that reading can be improved in the same way all other skills are developed: through practice. “The way to maintain high comprehension and get through text faster is to practise reading and to become a more skilled language user.”

“If two goals of reading might be to learn for the long haul, and to think – that may be part of enjoyment, that may be part of learning – then what’s the hurry?” Baron says. “Why are we feeling like the White Rabbit?”

For Seaboyer, reading a good book is akin to a meditative experience . The “wonderful, immersive process that is deep reading” reliably brings her pleasure. “Something else is picking you up, and moving your mind and body and soul into a different space so you can think about the world differently.”

Source link

Continue Reading

Technology

Is a lack of standards holding immersion cooling back? • The Register

Voice Of EU

Published

on

Comment Liquid and immersion cooling have undergone something of a renaissance in the datacenter in recent years as components have grown ever hotter.

This trend has only accelerated over the past few months as we’ve seen a fervor of innovation and development around everything from liquid-cooled servers and components for vendors that believe the only way to cool these systems long term is to drench them in a vat of refrigerants.

Liquid and immersion cooling are by no means new technologies. They’ve had a storied history in the high-performance computing space, in systems like HPE’s Apollo, Cray, and Lenovo’s Neptune to name just a handful.

A major factor driving the adoption of this tech in traditional datacenters is a combination of more powerful chips and a general desire to cut operating costs by curbing energy consumption.

One of the challenges, however, is many of these systems employ radically different form factors than are typical in air-cooled datacenters. Some systems only require modest changes to the existing rack infrastructure, while others ditch that convention entirely in favor of massive tubs into which servers are vertically slotted.

The ways these technologies are being implemented is a mixed bag to say the least.

Immersion cooling meets rack mount

This challenge was on full display this week at HPE Discover, where the IT goliath announced a collaboration with Intel and Iceotope to bring immersion-cooling tech to HPE’s enterprise-focused Proliant server line.

The systems can now be provisioned with Iceotope’s Ku:l immersion and liquid-cooling technology, via HPE’s channel partners with support provided by distributor Avnet Integrated. Iceotope’s designs meld elements of immersion cooling and closed-loop liquid cooling to enable this technology to be deployed in rack environments with minimal changes to the existing infrastructure.

Ice’s chassis-level immersion-cooling platform effectively uses the server’s case as a reservoir and then pumps coolant throughout to hotspots like the CPU, GPU, or memory. The company also offers a 3U conversion kit for adapting air-cooled servers to liquid cooling.

Both designs utilize a liquid-to-liquid heat exchanger toward the back of the chassis, where deionized water is pumped in and heat is removed from the system using an external dry cooler.

This is a stark departure from the approach used by rival immersion-cooling vendors, such as LiquidStack or Submer, which favor submerging multiple systems in a tub full of coolant — commonly a two-phase refrigerant or specialized oil.

While this approach has shown promise, and has even been deployed in Microsoft’s Azure datacenters, the unique form factors may require special consideration from building operators. Weight distribution is among operators’ primary concerns, Dell’Oro analyst Lucas Beran told The Register in an earlier interview.

Standardized reference designs in the works

The lack of a standardized form factor for deploying and implementing these technologies is one of several challenges Intel hopes to address with its $700 million Oregon liquid and immersion cooling lab.

Announced in late May, the 200,000-square-foot facility, located about 20 miles west of Portland at its Hillsboro campus in the US, will qualify, test, and demo its expansive datacenter portfolio using a variety of cooling tech. The chipmaker is also said to be working on an open reference design for an immersion-cooling system that’s being developed by Intel Taiwan.

Intel plans to bring other Taiwanese manufacturers into the fold before rolling out the reference design globally. Whether the x86 giant will be able to bring any consistency to the way immersion cooling will be deployed in datacenters going forward remains to be seen, however.

Even if Intel’s reference design never pans out, there are still other initiatives pursuing similar goals, including the Open Compute Project’s advanced cooling solutions sub project, launched in 2018.

It aims to establish an ecosystem of servers, storage, and networking gear built around common standards for direct contact, immersion, and other cooling tech.

In the meantime, the industry will carry on chilling the best ways it can. ®

Source link

Continue Reading

Trending

Subscribe To Our Newsletter

Join our mailing list to receive the latest news and updates 
directly on your inbox.

You have Successfully Subscribed!