Showing posts with label #MarkZuckerbergFAIL. Show all posts
Showing posts with label #MarkZuckerbergFAIL. Show all posts

Wednesday 3 June 2020

For years Facebook Inc. has known that its algorithms encourage and amplify antisocial behaviour like hate speech and extreme political bias


It seems that Facebook Inc. executives shut down efforts to make the site less divisive - because social and political division was increasing company profits by keeping certain categories of users engaged.

One has to wonder to what degree the company's decades of fostering poisonous online comment has contributed to the chaos that is American society in 2020.

Business Insider, 29 May 2020:
  • For years, Facebook has known that its algorithms encourage and amplify antisocial behaviour like hate speech and extreme political bias to keep users engaged, according to company documents reported in The Wall Street Journal.
  • When given proposals to make the platform better, executives often balked. They didn’t want to offend bad actors, and they didn’t want to release their hold on people’s attention. At Facebook attention equals money. 
  • So Facebook’s algorithms have been allowed to continue being sociopaths – pushing divisive content and exploiting people’s visceral reactions without a thought for the consequences or any remorse for their actions. 
  • Meanwhile, by letting bad actors on the platform do their thing, Facebook is feeding an inherent political bias into the algorithms themselves, and the company at large.
Facebook has always claimed that its mission is to bring people together, but a new report from The Wall Street Journal laid bare what many have suspected for some time: Its algorithms encourage and amplify harmful, antisocial behaviour for money. 

In other words, Facebook’s algorithms are by nature sociopaths. And company executives have been OK with that for some time. 

Here’s what we learned from Jeff Horowitz and Deepa Seetharaman at The Journal
  • A 2016 internal Facebook report showed “64% of all extremist group joins are due to our recommendation tools.” 
  • A 2018 internal report found that Facebook’s “algorithms exploit the human brain’s attraction to divisiveness” and warned that if left unchecked they would simply get nastier and nastier to attract more attention. 
  • An internal review also found that algorithms were amplifying users that spent 20 hours on the platform and posted the most inflammatory content (users that may not be people at all, but rather Russian bots, for example). 
  • Facebook executives, especially Mark Zuckerberg, time and time again ignored or watered down recommendations to fix these problems. Executives were afraid of looking biased against Republicans – who, according to internal reports, were posting the highest volume of antisocial content. 
  • And of course executives had to protect the company’s moneymaking, attention-seeking, antisocial algorithms – regardless of the damage they may be doing in society as a whole. Politics played into that as well. 
People who suffer from antisocial personality disorder – known in popular culture as “sociopaths” – engage in harmful, deceptive behaviour without regard for social norms. Sometimes this is done with superficial charm; other times this is done with violence and intimidation. These people never feel remorse for their behaviour, nor do they consider its long-term consequences. 

This is how Facebook’s algorithms behave. It’s how they hold on to users’ attention and how, ultimately, the company makes money. 

This runs contrary to what the company has been telling us about itself. After the bad rap it developed in the wake of the 2016 election, executives and the company’s marketing machine were telling us that Facebook was both financially and culturally committed to encouraging pro-social behaviour on the platform by doing things like removing violence and hate speech, making sure conspiracy theories and lies didn’t go viral, and cracking down on opioid sales. 

Now we know that that commitment was limited. Facebook would not kill the algorithms that laid the golden eggs despite their bias against these goals, or even clip their wings for that matter.....

Read the full article here.

Sunday 1 July 2018

So what has Facebook Inc been up to lately?


Everything from admitting to further data breaches, to altering images, to supressing legitimate content, to considering payment for access, to shareholder revolt, it seems......

The Herald Sun reported on 9 June 2018 at p.59:

Facebook is ­embroiled in another data privacy scandal, confirming a software bug led to the private posts of 14 million users being made public.

According to Facebook, the bug was active from May 18 to May 27 and changed the privacy settings of some users without telling them.

“Today we started letting the 14 million people affected know — and asking them to review any posts they made during that time,” Facebook chief privacy officer Erin Egan said.

“To be clear, this bug did not impact anything people had posted before, and they could still choose their audience just as they always have.” It was unclear yesterday how many Australian users were affected. Facebook said the bug occ­urred during the development of a new share function that ­allowed users to share featured items on their profile page, such as a photo.

“The problem has been fixed, and for anyone affected, we changed the audience back to what they’d been using before,” Ms Egan said.

Facebook has urged affected customers to review posts made between May 18 and May 27 to see if any private posts had been automatically made public.

The latest issue comes as Facebook chief Mark Zuckerberg faces the prospect of a public grilling before the Aus­tralian parliament’s intelligence and security committee.
Facebook admitted this week it had struck data partnerships — where it shares the personal data of people on the social media platform — with at least four Chinese electronics companies, including Huawei Technologies.

Huawei has been barred from a series of major projects in Australia over concerns about its close links to the Chinese government.

Members of the parliamentary intelligence and security committee want Mr Zuckerberg to come to Australia and answer questions about the data-sharing pact.

On 18 June 2018 The Sun reported that Facebook Inc had begun to manipulate images – effectively producing ‘fake images’ that were being passed off a real.

Then on 20 June 2018 Facebook Inc. declared its intention to charge certain private group users for participation on its platforms:

Today, we’re piloting subscriptions with a small number of groups to continue to support group admins who lead these communities.

This world-wide social platform apparently expects that if it formally launches this access fee (reportedly up to $360 a year) then these costs to be passed on as subscription fees – with Facebook  letting administrators charge subscription fees from $4.99 to $29.99 each month to join premium subgroups containing exclusive posts.

Presumably, if the market responds in sufficient numbers then Facebook will change the rules and demand that private groups hand over a percentage of subscription fees collected.

The Guardian, 24 June 2018:

George Orwell wrote in his essay Politics and the English Language: “In our age there is no such thing as ‘keeping out of politics’. All issues are political issues.” 

When Facebook constructed a new archive of political advertising, had it thought a little more about this concept of what is “political”, it might have more accurately anticipated the subsequent Orwellian headache. As it is, journalists are finding their articles restricted from promotion because they are lumped in with campaigning materials from politicians, lobby groups and advocacy organisations.

The new archive of ads with political content, which Facebook made public last month, has become the latest contested piece of territory between platforms and publishers. The complaint from publishers is that Facebook is categorising posts in which they are promoting their own journalism (paying money to target particular groups of the audience) as “political ads”. Publishers have reacted furiously to what they see as toxic taxonomy.

Mark Thompson, the chief executive of the New York Times, has been the most vocal critic, describing Facebook’s practices as “a threat to democracy” and criticising the platform in a recent speech to the Open Markets Initiative in Washington DC. “When it comes to news, Facebook still doesn’t get it,” said Thompson. “In its effort to clear up one bad mess, it seems to be joining those who want to blur the line between reality-based journalism and propaganda.”

At a separate event at Columbia University, Thompson and Facebook’s head of news partnerships, Campbell Brown, fought openly about the initiative. Thompson showed examples of where New York Times articles, including recipes, had been wrongly flagged as political. Brown emphasised that the archive was being refined, but stood firm on the principle that promoted journalism ought to be flagged as “paid-for” political posts. “On this you are just wrong,” she told Thompson.

Publishers took to social platforms to question the labelling and representation of their work. One of the most egregious examples came from investigative journalism organisation Reveal. Last week, at the height of the scandal around the separation of undocumented migrant families crossing the US border, it published an exclusive story involving the alleged drugging of children at a centre housing immigrant minors. It was flagged in the Facebook system as containing political content, and as Reveal had not registered its promotion of the story, the promoted posts were stifled. Facebook did not remove the article, but rather stopped its paid circulation. Given the importance of paid promotion, it is not surprising that publishers see this as amounting to the same thing.

And trust issues can be found both inside and outside Facebook's castle walls.....

Business Insider, 24 June 2018:

A Survata study, seen exclusively by Business Insider, asked US consumers to rate big tech companies from one (most trusted) to five (least trusted). Survata surveyed more than 2,600 people in April and May. It’s the first time Survata has carried out the survey.

The results show that Facebook is nowhere near as trusted as Amazon, PayPal, or Microsoft – but that people do trust it more than Instagram. Instagram, of course, is owned by Facebook.

Here’s the top 15 in order of most to least trusted:
1 .Amazon
2. PayPal
3. Microsoft
4. Apple
5. IBM
6. Yahoo
7. Google
8. YouTube
9. eBay
10. Pandora
11. Facebook
12. LinkedIn
13. Spotify
14. AOL
15. Instagram

Business Insider, 26 June 2018:

Shareholders with nearly $US3 billion invested Facebook are trying to topple Mark Zuckerberg as chairman and tear up the company’s governance structure.

Business Insider has spoken with six prominent shareholders who said there was an unprecedented level of unrest among Facebook’s backers following a series of scandals.

They are in open revolt about Zuckerberg’s power base, which gives him the ability to swat away any shareholder proposal he disagrees with.

 One investor compared him to a robber baron, a derogatory term for 19th-century US tycoons who accumulated enormous wealth.

Facebook says its governance structure is “sound and effective” and splitting Zuckerberg’s duties as chairman and CEO would cause “uncertainty, confusion, and inefficiency.”

Finally, it was reported on 29 June 2018 by IT News that, you guessed it, yet another Facebook sponsored personality test was allowing data to be extracted without the users knowledge or informed consent:

A security researcher has found that a popular personality test app running on Facebook contained an easily exploitable flaw that could be used to expose sensitive information on tens of millions of users.

Belgian security researcher Inti De Ceukelaire joined Facebook's bug bounty program, set up by the giant social network after the Cambridge Analytica data leak scandal and tried out the NameTests.com's personality test app developed by Social Sweethearts.

De Ceukelaire discovered that when he loaded a personality test, NameTests.com fetched his personal data from Facebook and displayed it on a webpage.

He was shocked to see that users' personal data was wrapped in a Javascript file by NameTests.com, which could be accessed via a weblink over the plain text HTTP protocol.

This meant that any website that requested the file could access the personal information retrieved from users' Facebook accounts.

The security researcher tested this by setting up a website that connected to NameTests.com and was able to access Facebook posts, photos and friend lists belonging to visitors.

Information leaked included people's Facebook IDs, first and last names, languages used, gender, date of birth, profile pictures, cover photo, currency, devices used, and much more.

Worse, De Ceukelaire found that NameTests.com doesn't log off users which means the site would continue to leak user data even after the app was deleted.