Section 230: Congress Seeks Testimony, Ignores It

Section 230: Congress Seeks Testimony, Ignores It
By EJ Haselden, October 30, 2020

It’s a timeless trope from the era of afterschool specials: misbehaving children stand before Mom and Dad’s kitchen-table duumvirate to answer for their schoolyard shenanigans, but the pretense of discipline soon wears through and the scene devolves into a nasty argument between the grownups. The kids’ real punishment is that they are made pawns and captive audience to a painful display of parental dysfunction. So unfolded this week’s Senate hearing on social media regulation, rhetorically titled “Does Section 230’s Sweeping Immunity Enable Big Tech Bad Behavior?

Section 230 (47 U.S.C. § 230) is a part of the 1996 Communications Decency Act, and it is perhaps best known for shielding social media companies (among others) from liability for content that their users post:

“No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.”

The titular “bad behavior” and “sweeping immunity” that prompted this hearing, however, relate to another, lesser-known protection granted by Section 230, which shields platforms when they choose to filter, fact-check, or otherwise annotate content that they consider harmful and/or inaccurate:

“No provider or user of an interactive computer service shall be held liable on account of any action voluntarily taken in good faith to restrict access to or availability of material that the provider or user considers to be obscene, lewd, lascivious, filthy, excessively violent, harassing, or otherwise objectionable, whether or not such material is constitutionally protected”

The nominal debate here surrounds the “otherwise objectionable” material in that description. Social media companies have chosen to interpret it as any content of questionable origin or veracity that could result in public harm (most recently regarding health advisories, voter suppression, and influence campaigns orchestrated by foreign intelligence services). Their caution stems from lessons learned in the rapid spread of disinformation leading up to the 2016 election, as well as a once-in-a-century pandemic that has seen deadly irresponsible claims espoused by supposed authority figures. Republican lawmakers claim that this content moderation has disproportionately muted conservative voices on social media. Democratic lawmakers, meanwhile, argue that these companies not only have the right, but the responsibility, to assess content based on its potential consequences and without regard for its ideological bent. It should be noted that multiple independent studies and a Facebook internal audit failed to find the alleged anti-conservative bias, but the fact that right-leaning engagement actually dwarfs that of center and left-leaning sources means that flagging only a small fraction of it still provides ample anecdotal evidence of prejudice (which is obviously enough to prompt Congressional hearings).

The administration has called for an outright repeal of Section 230, despite the fact that this would almost certainly lead to more content restrictions as companies adapt to the increased threat of liability. The consensus on Capitol Hill and in Silicon Valley therefore appears to be some amount of targeted Section 230 reform, while keeping the basic framework intact.

Which brings us back to this week’s hearing (or spectacle, or charade, or sham, depending on whom you ask). The Senate Committee on Commerce, Science, and Transportation subpoenaed the CEOs of Google, Twitter, and Facebook, respectively, to testify on behalf of Social Media. Most commentators agree that the face time with Tech Actual was not spent productively. As with those quarrelling parents, it was never really about the kids.

Republicans’ line of soi-disant questioning focused almost entirely on what they consider censorship of conservatives (69 of 81 questions, per the New York Times), as they demanded examples of the same (loosely defined) censorship directed at liberal outlets. Senator Ron Johnson asked the witnesses about the ideological makeup of their respective workforces—rhetorically, because it would be illegal for them to maintain that sort of record—in an effort to prove anti-conservative bias by virtue of microcultural majority (which almost sounded like an argument for some variant of affirmative action).

Democrats, for their part, focused most of their attention on the legitimacy and impact of the hearing itself, expressing concern that it could serve to intimidate social media companies into relaxing moderation policies at a time when the nation is perhaps most vulnerable to manipulative media. The bulk of their more on-topic questioning concerned dis- and misinformation and what actions the companies were taking to combat it ahead of the election. Still, not that much about Section 230 reform.

In keeping with the scripted, postured non-discussion, the most meaningful witness testimony came in the form of prepared opening statements. In those, Pichai reasserted Google’s anti-bias philosophy and cautioned against reactionary changes to Section 230, Dorsey promoted increased transparency and user inclusion in Twitter’s decision-making processes, and Zuckerberg praised Section 230 while inviting a stricter and more explicit rewrite of its provisions (for which Facebook would gladly provide input). Their full statements are available on the committee’s hearing website.

The timing and tenor of this eleventh-hour pre-election partisan screed exchange never inspired much hope for substantive debate, but even so, there was a jarring lack of effort to better understand the pressing and complex problems that Section 230 is still, at this moment, expected to resolve. The reason this matters, the reason it’s so alarming that neither side was terribly interested in the companies’ offers of greater transparency—something we’d consider a win for democracy in saner times—is that our government has abdicated its responsibility of oversight on this topic except in cases where the threat of enforcement can be used as a political weapon.

In the end, it’s probably fitting that Congress used a social media hearing as a platform to amplify and disseminate entrenched views that they had no intention of rethinking.

 

Photo credits:

Can there truly be ethics in autonomous machine intelligence?

Can there truly be ethics in autonomous machine intelligence?
By Matt White, October 30, 2020

Some would say that we are in the infancy of the fourth industrial revolution, where artificial intelligence and the autonomy it is ushering in are positioned to become life-altering technologies. Most understand the impacts of autonomous technologies as it relates to jobs, they are concerned that autonomous vehicles and robotic assembly lines will relegate them to the unemployment line. But very little thought and conversely research has been done into the ethical implications of autonomous decision making that these systems are confronted with. Although there are far reaching ethical implications with AI and automation there are opposing views of who is truly responsible for the ethical decisions made by an autonomous system. Is it the designer? The programmer? The supplier of the training data? The operator? Or should the system itself should be responsible for any moral or ethical dilemmas and their outcomes.

Take for instance the incident with Uber’s self-driving car a few years ago, where one of its cars killed a pedestrian crossing the road in the middle of the night. The vehicle’s sensors collected data which revealed it was aware of a person crossing in front of its path, but the vehicle took no action and struck and killed the pedestrian. Who is ultimately responsible when an autonomous vehicle kills a person? In this case it was the supervising driver but what happens when there is no driver in the driver seat? What if the vehicle had to make a choice like in the trolley problem, between hitting a child or hitting a grown man? How would it make such a challenge moral decision?

A car parked on a city street


Image Source: Singularity Hub

The Moral Machine, a project from MIT’s Media Lab is tackling just this, developing a dataset on how people would react to particular moral and ethical decisions where it comes to driverless cars. Should you run over 1 disabled person and 1 child or 3 obese people, or should you crash yourself into a barrier and kill your 3 adult passengers to save two men and two women of a healthy weight pushing a baby? However, the thought that autonomous vehicles will base their decisions of morality on crowd-sourced datasets of varying moral perspectives seems absurd. Only those who participate in the process will have their opinions included, anyone can go online and contribute to the dataset without any form of validation, and not withstanding all of the opinions that are not included, there are various moral philosophy theories that could be applied to autonomous ethical decision making that would overrule rules derived from datasets. Does the system follow utilitarianism, Kantianism, virtue ethics, so forth? Although the Moral Machine is considered to be a study in its current incarnation, it uses a very primitive set of parameters (number of people, binary gender, weight, age, visible disability) to allow users to determine the value they place on human life. In real life, real people have more than these handful of dimensions like race, socio-economic status, non-binary gender, and so forth. Could adding these real-life dimensions create a bias that would further de-value people who might meet certain criteria and be in the way of an autonomous vehicle? Might the value placed on a homeless person by less than that of a Wall street stockbroker?

Graphical user interface, diagram


Image Source: Moral Machine

There is certainly a lot to unpack here, especially if we change contexts and look at armed unmanned autonomous vehicles (AUAVs) which are used in warfare to varying degrees. As we transition from remote pilots to fully autonomous war machines, who makes the decision whether to drop a bomb on a school containing 100 terrorists and 20 children? Does the operator absolve himself of any responsibility when the AUAV makes the decision to drop a bomb and kill innocent people? Does the programmer or the trainer of the system bear any responsibility?

As you can see the idea of ethical decision making by autonomous systems is highly problematic and presents some very serious challenges that require further research and exploration. Systems that are designed to have a moral compass will not be sufficient, as they will adopt the moral standpoint of its creators. Training data is likely to be short-sighted, shallow in dimensions and biased based on the ethical standpoints of its contributors. It is obvious that the issue of ethical decision making in autonomous system needs further discourse and research in order to ensure that future systems that we come to rely on can make ethical decisions in a manner that demonstrates no bias; or perhaps we may have to accept that in fact autonomous machines will not be able to make ethical decisions in an unbiased manner.

References:

The Looming Revolution of Online Advertising

The Looming Revolution of Online Advertising
By Anonymous, October 30, 2020

In the era of the internet, advertising is getting creepily accurate and powerful. Large ad networks like Google, Facebook, and more collect huge amounts of data, through which they can infer a wide range of user characteristics, from basic demographics like age, gender, education, and parental status to broader interest categories like purchasing plan, lifestyle, beliefs, and personality. With such powerful ad networks out there, users often feel like they are being spied on and chased around by ads.


Image credit: privateinternetaccess.com

How is this possible?
How did we leak so much data to these companies? The answer is through cross-site and app tracking. When you surf the internet, going from one page to another, trackers collect data on where you have been and what you do. According to one Wall Street Journal study, the top fifty Internet sites, from CNN to Yahoo to MSN, install an average of 64 trackers[1]. The tracking can be done by scripts, cookies, widgets, or invisible image pixels embedded on the sites you visit. You probably have seen the following social media sharing buttons. Those buttons, no matter you click them or not, can record your visits and send data back to the social platform.


Image credit: pcdn.co

A similar story is happening on mobile apps. App developers often link in SDKs from other companies, through which they can gain analytic insights or show ads. As you can imagine, those SDKs will also report data back to the companies and track your activities across apps.

Why is it problematic?
Cross-site or app tracking poses great privacy concerns. Firstly, the whole tracking process happens behind the scenes. Most users are not aware of it until they see some creepily accurate ads, and even if they are aware of it, the users often have no idea how the data is collected and used, and who owns it. Secondly, only very technically sophisticated people know how to prevent this tracking, which can involve tedious configuration or even installation of other software. To make things worse, even if we can prevent future tracking, there is no clue how to wipe out the already collected data.

In general, cross-site and app activities are collected, sold, and monetized in various ways with very limited user transparency and control. GDPR and CCPA have significantly improved this. Big trackers like Google, Facebook, and more provide dedicated ad setting pages (1, 2), which allow users to delete or correct their data, to choose how they want to be tracked, etc. Though GDPR and CCPA gave users more control, most users stay with the default options and cross-site tracking remains prevalent.

The looming revolution
With growing concerns of user privacy, Apple took a radical action to kill the cross-site and app tracking. Over the past couple of years, Apple gradually rolled out the feature of Safari Intelligent Tracking Prevention (ITP)[2], which curtailed companies’ ability to install third-party cookies. With Apple taking the lead, Firefox and Chrome browsers are also launching similar features as ITP. In the release of IOS 14, Apple brought a similar feature as ITP to Apps world.


Image credit: clearcode.com

While at the first glance this may sound like a long-overdue change to safeguard users’ privacy, when delving deeper, it could create backlashes. Firstly, internet companies collect data in exchange for their free services: products like Gmail, Maps, Facebook are all free of use. According to one study from VOX, in an ad-free internet, the user would need to pay $35 every month to compensate for ad revenue[3]. Some publishers even threatened to proactively stop working on Apple devices. Secondly, Apple’s ITP solution doesn’t give much chance for users to participate. Cross-site tracking can in general enable more personalized services, more accurate search results, better recommendations, etc. Some uses may choose to opt-in to allow cross-site tracking for this purpose. Thirdly, Apple’s ITP only disabled third party cookies, and there are many other ways to continue the tracking. For example, ad platforms can switch to device-id or “fingerprint” the users by combining IP address and Geolocation.

Other radical solutions were also proposed, such as Andrew Yang’s Data Dividend Project. With many ethical concerns and the whole ads industry at stake, it is very interesting to see how things play out and what other alternatives are proposed around cross-site and app tracking.

 

References

We see only shadows

We see only shadows
By David Linnard Wheeler, October 30, 2020

After the space shuttle Challenger disaster (Figure 1) on January 28th, 1986, most people agreed on the cause of the incident – the O-rings that sealed the joints on the right solid rocket booster failed under cold conditions (Lewis, 1988). What most failed to recognize, however, was a more fundamental problem. The casual disregard of outliers, in this case from a data set used by scientists and engineers involved in the flight to justify the launch in cold conditions, can yield catastrophic consequences. The purpose of this essay is to show that a routine procedure for analysts and scientists – outlier removal – not only introduces biases but, under some circumstances, can actually lead to lethal repercussions. This observation raises important moral questions for data scientists.

Figure 1. Space shuttle Challenger disaster. Source: U.S. NEWS & WORLD REPORT

The night before the launch of the space shuttle Challenger, executives and engineers from NASA and Morton Thiokol, the manufacturer of the solid rocket boosters, met to discuss the scheduled launch over a teleconference call (Dalal et al. 1989). The subject of conversation was the sensitivity of O-rings (Figure 2) on the solid rocket boosters to the cold temperatures forecasted for the next morning.

Figure 2. Space shuttle Challenger O-rings on solid rocket boosters. Source: medium.com/rocket-science-falcon-9-and-spacex/space-shuttle-challenger-disaster-1986-7e05fbb03e43

Some of the engineers at Thiokol opposed the planned launch. The performance of the O-rings during the previous 23 test flights, they argued, suggested that temperature was influential (Table 1). When temperatures were low, for example between 53 and 65∘F, more O-rings failed than when temperatures were higher.

Table 1: Previous flight number, temperature, pressure, number of failed O-rings, and number of total O-rings

Some personnel at both agencies did not see this trend. They focused only on the flights where at least one O-ring had failed. That is, they ignored outlying cases where no O-rings failed because, from their perspective, they did not contribute any information (Presidential Commission on the space shuttle Challenger Accident, 1986). Their conclusion, upon inspection of data from Figure 3, was that “temperature data [are] not conclusive on predicting primary O-ring blowby” (Presidential Commission on the space shuttle Challenger Accident, 1986). Hence, they asked Thiokol for an official recommendation to launch. It was granted.

Figure 3. O-ring failure as a function of temperature

The next morning the Challenger launched and 7 people died.

After the incident, President Regan ordered William Rogers, former Secretary of State, to lead a commission to determine the cause of the explosion. The O-rings, the Commission found, became stiff and brittle in response to cold temperatures, thereby unable to maintain the seal between the joints of the solid rocket boosters. The case was solved. But a more fundamental lesson was missed.

Outliers and their removal from data sets can introduce consequential biases. Although this may seem obvious, it is not. Some practitioners of data science essentially promote cavalier removal of observations that are different from the rest. They focus instead on the biases that can be introduced when certain outliers are included in analyses.

This practice is hubristic for at least one reason. We, as observers, do not, in most cases, completely understand the processes by which the data we collect are generated. To use Plato’s allegory of the cave, we just see the shadows, not the actual objects. Indeed, this is one motivation to collect data. To remove data without defensible justification (e.g measurement or execution error) is to claim, even if implicitly, that we know how the data should be distributed. If true, then why collect data at all?

To be clear, I am not arguing that outlier removal is indefensible under any condition. Instead, I am arguing that we should exercise caution and awareness of the consequences of our actions, both when classifying observations as outliers and ignoring or removing them. This point was acknowledged by the Rogers Commission in the statement: “a careful analysis of the flight history of O-ring performance would have revealed the correlation in O-ring performance in low temperature[s]” (Presidential Commission on the space shuttle Challenger Accident, 1986).

Unlike other issues in fields like data science, the solution here may not be technical. That is, a new diagnostic technique or test will likely not emancipate us from our moral obligations to others. Instead, we may need to iteratively update our philosophies of data analysis to maximize benefits, minimize harms, and satisfy our fiduciary responsibilities to society.

 

References:

  • Dalal, S.R., Fowlkes, E.B., Hoadley, B. 1989. Risk analysis of the space shuttle: Pre-Challenger prediction of failure. Journal of the American Statistical Association.
  • Lewis, S. R. 1988. Challenger The Final Voyage. New York: Columbia University Press.
  • United States. 1986. Report to the President. Washington, D.C.: Presidential Commission on the Space Shuttle Challenger Accident.

A Short Case for a Data Marketplace

A Short Case for a Data Marketplace
By Linda Dong, October 23, 2020

In today’s digital, internet age, data is power. Using data, Netflix can generate recommendations, Facebook can tailor advertisements, and Visa can detect fraud. Google can predict your search phrase, Alexa can prompt you to restock household products, and Wealthfront can create your personalized retirement path, taking into account individual savings, spending, and investment goals.

Not only are data products powerful, but they also tend to be lucrative. Data products tend to be high-margin because the cost of goods sold is so low: companies generally do not pay users to collect their data. Whether companies are channeling these lucrative products into customer savings (by making other services free) or purely amassing these gains as company profits, the central question remains: should data collection be free?

– – – –


Image Source: Robinhood

Just like oil, labor, and water, data is a commodity. True – it happens to be a non-finite commodity that humans can create; however, it is also a raw material used to create sold products. Just as a bar of chocolate is made from many cacao beans, so is a web marketing analytics insight crafted from many individual browser interactions.

If you’re a chocolate maker, you’ll likely have a handful of cocoa suppliers. If you’re a web analytics company, you’ll likely have millions of users providing a little data each. However, the simple facts that your suppliers are: (i) distributed, and (ii) orders-of-magnitude more numerous do not constitute adequate justification for not compensating them.

The logistics might be simpler than you think. The idea of web-based microtransactions is not new; little known to most people, the HTTP status code of 402 [2] has been reserved for “Payment Required” use-cases for a while. While this was meant to power the opposite flow (for a requestor to present payment to access content, rather than a content provider to pay a visitor for data gathered during an interaction), this nevertheless brings us one step closer to a future where browsers might contain native wallets that can enable hundreds of microtransactions per hour.


Image Source: Mozilla Foundation

– – – –

Regulation lags behind innovation. While privacy concerns have culminated in new statutes regulating how entities should collect and use data, most protections today concern only data subjects’ rights and obligations. They have not yet evolved to address questions of compensation and profit-sharing.

Some of this is due to a lack of pressure from the general public, which, in turn, results from a lack of awareness regarding the value of data, as well as opacity regarding how companies collect and use data. Some of this is due to coercive user policies that foist consent of data collection. And some of it is due to the lack of a clear solution and path forward.

What if we reimagined the concept of privacy in an economic, rather than rights-based, context? Could browsers compete for users by providing more sophisticated privacy customizations? Could they better enable user control to select and disclose limited and specific data in exchange for monetary earnings? Could they auto-respond to pesky cookie preference pop-ups? Could they broker a new type of data marketplace between companies who want to buy data and users who want to sell data? Are these features valuable enough for them to charge users a fee, and would the public pay?

I, for one, would.

 

[1] learn.robinhood.com/articles/626haurrOd1BFJ3CkfH7xq/what-is-a-commodity/
[2] developer.mozilla.org/en-US/docs/Web/HTTP/Status/402

All about Grandma

All about Grandma
By Anonymous, October 23, 2020

My grandma Diane lives in Tulsa, OK on a small farm with one of my aunts, Heather, my uncle Carl, my two cousins Carl III and Toby, and my uncle Carl’s mom Bethanne. They raise goats and fowl, have a couple house dogs and some cats that come and go as they are wont to. The farm has a pond that the dogs swim in sometimes. These are things that I know because they’re my family. I’ve spent countless Thanksgivings and Christmases and been to several weddings with them.

What I didn’t know until today was that grandma is a registered Republican and Heather and Carl are registered Democrats. I didn’t intend to find this information. Rather with the 2020 election on the mind and news media covering early voting, I decided to do a cursory search about what voting information exists in the public domain. It took less than a minute to stumble onto grandma’s voter registration on the data aggregator: voterrecords.com, where voter registration records are available in searchable form for 16 states, Oklahoma included.

 

Of course, voter registration records have been public for a long time, but before sites like voterrecords.com it took real effort to go peruse voter rolls. While the process differed from state to state, you typically had to go to the local county office or the secretary of state’s office to formally request access. These barriers meant only the most interested of actors, like political parties or investigative journalists, took the time to do it. Now, this information is available almost accidentally to anyone with an internet connection anywhere in the world.

While presence of the internet makes access to voter records fundamentally different than in the past, what makes it concerning now is the degree to which political affiliation has become enmeshed with personal identity, particularly for more extreme actors on both ends of the political spectrum, some of which threaten violence.

To make matters much worse, voterrecords.com connects voter registration information to sites that conduct extensive background searches – truthfinder.com and beenverified.com – all without transparent labeling that prominently displayed buttons will trigger a background search.

Truthfinder conducts a search of property records, criminal records, bankruptcy records, social media accounts, etc. While truthfinder exploits public records databases for much of this information, its site is set up to make use of users’ interactions to reinforce algorithmic conclusions about which records are related to the actual person in question. Presenting follow-on questions in a way that most users are likely to think that the site is trying to isolate a particular individuals’ records, the questions ask users to confirm or deny algorithmically generated relationships with other records it has come across, thereby strengthening the person-matching algorithms that form the core of those sites.

After asking several such questions the site prompts users to search for more people – including people with which the person likely has no personal connection such as ‘celebrities’. Truthfinder’s charges for its services, and its model invites people to conduct ‘unlimited’ searches over a month, rather than purchase individual reports. Furthermore, the generated report contains information not just about the person you’ve gone down a rabbit hole searching for but also about several people that truthfinder has determined are related to the person you’ve searched for.

It is through this that I learned, despite having known grandma all my life, that a lien was put on the farm last year, that she received her social security number and card around the time she turned 18 rather than at birth, and the VIN number on her Toyota Sequoia. While she doesn’t have a criminal record, several people in neighboring states with similar names do. While I know those people aren’t her, someone who doesn’t know her as well may not and might mistakenly come to the conclusion that my grandma has a problem with shoplifting. Truthfinder’s presentation of this information makes this outcome more likely by exaggerating and not disclaiming that the information may not be linked to the right person, as happened in this case. This is all in addition to a litany of phone numbers, email addresses, social media accounts, amazon wish lists, and the addresses she has lived at or co-signed for going back decades. A couple more clicks yields similar information about all of my Oklahoma relatives over the age of 18.

While voter registration records and for that matter each of the other sets of public records used by these sites historically may have had valid reasons for being in the public domain, the internet has enabled aggregation across these datasets in a way that it literally takes less than 10 minutes to stumble unintentionally from a person’s voter record to knowing some of the most personal aspects of their lives like bankruptcy and criminal records, and not much longer to unearth similar information about nearly everyone they are related to.

This is made all the more troubling by the devolution in public discourse and increase in othering as personal identities of all sorts and stripes are increasingly coalescing into constellations around bipolar political affiliations. This is all paired with increasing rhetoric of political violence. Americans should consider carefully what information is put into the public domain, and should advocate to their state legislatures to curtail the publication and aggregation of such data sources.

Clearview AI: The startup that is threatening privacy

Clearview AI: The startup that is threatening privacy
By Stefania Halac, October 16, 2020

Imagine walking down the street, a stranger points their camera at you and can immediately pull up all your pictures from across the internet; they may see your instagram posts, your friends’ posts, any picture that you appear in, some which you may have never seen before. This stranger could now ascertain where you live, where you work, where you went to school, whether you’re married, who your children are… This is one of many compromising scenarios that may become part of our normal life if facial recognition software is widely available.

Clearview AI, a private technology company, offers facial recognition software that can effectively identify any individual. Facial recognition technology is intrinsically controversial, so much so that certain companies like Google don’t offer facial recognition APIs due to ethical concerns. And while some large tech companies like Amazon and Microsoft do sell facial recognition APIs, there is an important distinction between Clearview’s offering and that of the other tech giants. Amazon and Microsoft only allow you to search for faces from a private database of pictures supplied by the customer. Clearview instead allows for recognition of individuals in the public domain — practically anyone can be recognized. What sets Clearview apart is not its technology, but rather the database it assembled of over three billion pictures scraped from the public internet and social media. Clearview AI did not obtain consent from individuals to scrape these pictures, and has been sent cease and desist orders from major tech companies like Twitter, Facebook and Youtube over its practices due to policy violations.

In the wake of the Black Lives Matter protests earlier this year, IBM, Microsoft and Amazon updated their policies to restrict the sale of their facial recognition software to law enforcement agencies. On the other hand, Clearview AI not only sells to law enforcement and government agencies, but until May of this year was also selling to private companies, and has even been reported to have granted access to high net-worth individuals.

So what are the risks? One on hand, the algorithms that feed these technologies are known to be heavily biased and perform more poorly on certain minority populations such as women and African Americans. In a recent study, Amazon’s Rekognition was found to misclassify women as men 19% of times, and darker-skinned women for men 31% of time. If this technology were to be used in the criminal justice system, one implication here is that dark-skinned people would be more likely to be wrongfully identified and convicted.

Another major harm is that this technology essentially provides its users the ability to find anyone. Clearview’s technology would enable surveillance at protests, AA meetings and religious gatherings. Attending any one of these events or locations would become a matter of public record. In the wrong hands, such as those of a former abusive partner or a white supremacist organization, this surveillance technology could even be life-threatening for vulnerable populations.

In response, the ACLU filed a lawsuit against Clearview AI in May for violation of the Illinois Biometric Information Privacy Act (BIPA), alleging the company illegally collected and stored data on Illinois citizens without their knowledge or consent and then sold access to its technology to law enforcement and private companies. While some cities like San Francisco and Portland have enacted facial recognition bans, there is no overarching national law protecting civilian privacy from these blatant privacy violations. With no such law in sight, this may be the end of privacy as we know it.

References:

www.aclu.org/news/privacy-technology/were-taking-clearview-ai-to-court-to-end-its-privacy-destroying-face-surveillance-activities/

The Gender Square: A Different Way to Encode Gender

The Gender Square: A Different Way to Encode Gender
By Emma Tebbe, October 16, 2020


Image: square with two axes, the horizontal reading Masculine and Feminine and the vertical reading Low Gender Association / Agender and Strong Gender Association

As non-gender-conforming and transgender folks become more visible and normalized, the standard male / female / other gender selections we all encounter in forms and surveys become more tired and outdated. First of all, the terms “male” and “female” generally refer to sex, or someones biological configuration, “including chromosomes, gene expression, hormone levels and function, and reproductive/sexual anatomy.” Male and female are not considered the correct terms for gender orientation, which “refers to socially constructed roles, behaviours, expressions and identities of girls, women, boys, men, and gender diverse people.” Although sex exists on a spectrum which includes intersex people, gender has a wide range of identities, including agender, bigender, and genderqueer. This gender square method of encoding gender aims to encompass more of the gender spectrum than a simple male / female / other selection.


Image: triangle defining sex, gender expression, gender attribution, and gender identity

Upon encountering this square in a form or survey, the user would drag the marker to the spot on the square that most accurately represents their gender identity. This location would then be recorded as a coordinate pair, where (0, 0) is the center of the square. The entity gathering the data would then likely use those coordinates to categorize respondents. However, using continuous variables to represent gender identity allows for many methods of categorization. The square could be divided into quadrants, as pictured above, vertical halves (or thirds, or quarters), or horizontal sections. This simultaneously allows for flexibility in how to categorize gender and reproducibility of results by other entities. Other analysts would be able to reproduce results if they are given respondents’ coordinates and the categorization methodology used. Coordinate data could even be used as it was recorded, turning gender from a categorical variable into a continuous one.

Although this encoding of gender encompasses more dimensions, namely representing gender as a spectrum which includes agender identities, it still comes with its own problems. First of all, the gender square still does not leave room for flexible gender identities including those whose gender is in flux or those who identify as genderfluid or bigender. There are a few potential solutions for this misrepresentation on the UI side, but these create new problems with data encoding. Genderfluid folks could perhaps draw an enclosed area in which their gender generally exists, but recording this data is much more complex than a simple coordinate pair, and would become an array of values rather than a coordinate pair. People who identify as bigender could potentially place two markers, one for each of the genders they experience. Both this approach and an area selection approach make the process of categorization more complex – if an individual’s gender identity spans two categories, would they be labeled twice? Or would there be another category for people who fall into multiple categories?


Image: a gender spectrum defining maximum femininity as “Barbie” and maximum masculinity as “G.I. Joe”

Another issue might arise with users who haven’t questioned their gender identity along either of these axes, and may not understand the axes (particularly the Highly Gendered / Agender axis) enough to accurately use the gender square. When implemented, the gender square would likely need an explanation, definitions, and potentially suggestions. Suggestions could include examples such as “If you identify as a man and were assigned that gender at birth, you may belong in the upper left quadrant.” Another option may be to include examples such as in the somewhat problematic illustration above.

This encoding of gender would likely first be adopted by groups occupying primarily queer spaces, where concepts of masculinity, femininity, and agender identities are more prominent and considered. If used in places where data on sex and transgender status is vital information, such as at a doctor’s office, then the gender square would need to be supplemented by questions obtaining that necessary information. Otherwise, it is intended for use in spaces where a person’s sex is irrelevant information (which is most situations where gender information is requested).

Although still imperfect, representation and identification of gender along two axes represents more of the gender spectrum than a simple binary, and still allows for categorization, which is necessary for data processing and analytics. With potential weaknesses in misunderstanding and inflexibility, it finds its strength in allowing individuals to more accurately and easily represent their own identities.

References:
cihr-irsc.gc.ca/e/48642.html
www.glsen.org/activity/gender-terminology
journals.sagepub.com/doi/full/10.1177/2053951720933286
Valentine, David. The Categories Themselves. GLQ: A Journal of Lesbian and Gay Studies, Volume 10, Number 2, 2004, pp. 215-220
www.spectator.co.uk/article/don-t-tell-the-parents for image only

 

When Algorithms Are Too Accurate

When Algorithms Are Too Accurate
By Jill Cheney, October 16, 2020

An annual rite of passage every Spring for innumerable students is college entrance exams. Regardless of their name, the end result is the same: to influence admission applications. When the Covid-19 pandemic swept the globe in 2020, this milestone changed overnight. Examinations were cancelled, leaving students and universities with no traditional way to evaluate admission. Alternative solutions emerged with varying degrees of veracity.

In England, the solution used to replace their A-level exams involved developing a computer algorithm to predict student performance. In the spirit of a parsimonious model, two parameters were used: the student’s current grades and the historical test record of the attending school. The outcome elicited nationwide ire by highlighting inherent testing realities.

Overall, the predicted exam scores were higher – more students did better than on any previous resident exam with 28% getting top scores in England, Wales and Northern Ireland. However, incorporating the school’s previous test performance into the algorithm created a self-fulfilling reality. Students at historically high performing schools had inflated scores; conversely, students from less performing schools had deflated ones. Immediate cries of AI bias erupted. However, the data wasn’t wrong – the algorithm simply highlighted the inherent biases and disparity in the actual data modeled.

Reference points did exist for the predicted exam scores. One was from teachers since they provide a prediction on student performance. The other was from student scores on previous ‘mock’ exams. Around 40 percent of students received a predicted score that was one step lower than their teachers’ predictions. Not surprisingly, the largest downturn in predictions occurred amongst poorer students. Many others had predicted scores below their ‘mock’ exam scores. Mock exam results support initial university acceptance; however, they must be followed-up with commensurate official exam scores. For many
students, the disparity between their predicted and ‘mock’ exam scores jeopardized their university admission.

Attempting to rectify the disparities came with its own challenges. Opting to use teacher predicted scores required accepting that not all teachers provided meticulous student predictions. Based on teacher predictions alone, 38% of predicted scores would have been at the highest levels: A*s and As. Other alternatives included permitting students to retake the exam in the Fall or allowing the ‘mock’ exam scores to stand-in should they be higher than the predicted ones. No easy answers existed when attempting to navigate an equitable national response.

As designed, the computer model assessed the past performance of a school over student performance. Individual grades could not offset the influence of a school’s testing record. It also clearly discounted more qualitative variables, such as test performance skills. In the face of a computer-generated scoring model, a feeling of powerlessness emerged. No longer did students feel they possessed control over their future and schooling opportunities.

Ultimately, the predictive model simply exposed the underlying societal realities and quantified how wide the gap actually is. In the absence of the pandemic, testing would have continued on the status quo. Affluent schools would have received higher scores on average than fiscally limited schools. Many students from disadvantaged schools would have individually succeeded and gained university admission. The public outcry this predictive algorithm generated underscores how the guise of traditional test conditions assuages our concerns about the realities of standardized testing.

Sources:
www.theverge.com/2020/8/17/21372045/uk-a-level-results-algorithm-biased-coronavirus-covid-19-pandemic-university-applications

www.bbc.com/news/education-53764313

Data as Taxation

Data as Taxation
By Anonymous, October 16, 2020

Data is often analogized with transaction. We formulate our interactions with tech companies as an exchange of our data as payment for services, which in turn allow for the continued provision of those services.

Metaphors like these can be useful in that they allow us to port developed intuitions from a well-trodded domain (transactions) to help us navigate more less familiar waters (data). In this spirit, I wanted to further develop this “data collection = economic transaction” metaphor, and explore how our perceptions of data collection change with a slight tweak: “data collection = taxation”


In the context of data collection, the following quote from Supreme Court Justice Oliver Wendall Holmes might give one pause. Is this applicable, or entirely irrelevant?

Here’s what I mean: with taxation, government bodies mandate that citizens contribute a certain amount of resources to fund public services. The same goes for data – while Google, Facebook, and Amazon are not governments, they also create and maintain enormous ecosystems that facilitate otherwise impossible interactions. Governments allow for a coordination around national security, education, and supply chains, and Big Tech provides the digital analogues. Taxation and ad revenue allow for the perpetual creation of this value. Both can embody some (deeply imperfect) notion of “consent of the governed” through voter and consumer choice, although neither provides an easy way to “opt out.”

Is this metaphor perfect? Not at all, but there is still value in making the comparison. We can recycle centuries of bickering over fairness in taxation.

For instance, one might ask “when is taxation / data collection exploitative?” On one end, some maintain that “all taxation is theft,” a process by which private property is coercively stripped. Some may feel a similar sense of violation as their personal information is harvested – for them, perhaps the amorphous concept of “data” latches onto the familiar notion of “private property,” which might in turn suggest the need for some kind of remuneration.

At the other extreme, some argue that taxation cannot be the theft of private property, because the property was never private to begin with. Governments create the institutions and infrastructure that allows the concept of “ownership” to even exist, and thus all property is on loan. One privacy analogue could be that the generation of data is impossible and worthless without the scaffolding of Big Tech, and thus users have a similarly tenuous claim on their digital trails.

The philosophy of just taxation has provided me an off-the-shelf frame by which to parse a less familiar space. Had I stayed with the “data collection = economic transaction” metaphor, I would have never thought about data from this angle. As is often the case, a different metaphor illuminates different dimensions of the issue.

Insights can flow the other way as well. For example, in data circles there is a developing sophistication around what it means to be an “informed consumer.” It is recognized by many that merely checking the “I agree” box does not constitute a philosophically meaningful notion of consent, as the quantity and complexity of relevant information is too much to expect from any one consumer. Policies and discussions around the “right to be forgotten”, user control of data, or the right to certain types of transparency acknowledge the moral tensions inherent in the space.

These discussions are directly relevant to justifications often given for a government’s right to tax, like the “social contract” or the “consent of the governed.” Both often have some notion of informed consent, but this sits on similarly shaky ground. How many voters know how their tax dollars are being spent? While government budgets are publicly available, how many are willing to sift through reams of legalese? How many voters can tell you what military spending is within an even order of magnitude? Probably as many as who know exactly how their data is packaged and sold. The data world and its critics have much to contribute to the question of how to promote informed decision-making in a world of increasing complexity.


Linguists George Lakoff and Mark Johnson suggest that metaphors are central to our cognitive processes.

Of course, all of these comparisons are deeply imperfect, and require much more space to elaborate. My main interest in writing this was exploring how this analogical shift led to different questions and frames. The metaphors we use have a deep impact on our ability to think through novel concepts, particularly when navigating the abstract. They shape the questions we ask, the connections we make, and even the conversations we can have. To the extent that that’s true, metaphors can profoundly reroute society’s direction on issues of privacy, consent, autonomy, and property, and are thus well-worth exploring.