Headline in Tech of the Week
Executive Vice President of the EU Commission Vestager talks about the Data protection and competition challenges
…her speech gives a good summary of what the Commission is thinking and where we are up to in the EU. Concluding that policymakers in the respective fields of competition, data protection and consumer protection needs to co-operate she pointed out the following:
Digital Markets Act
She said the ability to store data, combined with the reach and breadth created by powerful network effects, can enable large platforms to hunker down in their strongholds and, in worst case, to abuse their market power. Data can raise barriers to entry and it can also be used in anti-competitive ways. This is why the Digital Markets Act needed to be implemented, to deal with large global digital platforms.
- Amazon: A case was opened against it because it could use its business customer’s data to compete against the customers as a result of its dual role as platform and seller [For example, Amazon can assess what products sell well only to manufacture those products, and place them prominently on its eCommerce site].
- Meta/Facebook: A case was opened against because there was a concern that Facebook might make use of the data obtained from competing providers in the context of their advertising on Facebook’s social network, to help Facebook Marketplace outcompete them. Facebook could, for instance, receive precise information on users’ preferences from its competitors’ advertisement activities and use such data in order to adapt Facebook Marketplace.
The DMA is designed to regulate the way data can be collected and used by gatekeepers. For example, it bans the combination of data across services without consent and the use of tracking users and processing the data for ad targeting purposes without consent.
Digital Services Act, Data Act and Data Governance Act
- Digital Services Act: aim is to protect users’ rights. It includes the banning of use of sensitive personal data to target adverts, and granting users the right to opt out of content recommendations based on profiling.
- Data Act: establish harmonised rules on access for data generated by connected devices. [for example, users have the right to provide relevant data relating to a particular device – this could be useful when the user wants its device (eg. cars) to be repaired by an independent service provider]
- Data Governance Act: to facilitate data sharing by creating neutral data intermediaries. This will allow SMEs to access the data they need to innovate, and to enable the safe reuse of some types of public sector data.
Other competition law in play
- M&A: In the Google Fitbit case, the merger was cleared with some conditions, including a data silo separating the data collected from Fitbit users from any other Google data used for advertising.
- Privacy Sandbox: this is the phase out by Google Chrome of third party cookies that track users. The EU Commission is looking into who will have access to what data and whether the new practices could distort competition in the ad tech sector by favouring Google when it comes to having access to data [note that German publishers and marketing agencies, and others have complained to the EU commission that abolishment of third party cookies mean that third parties cannot use data to carry out targeted advertising although Google will be able to continue using first party data it has direct access to]
Artificial Intelligence
Amazon sued in Illinois and Washington for taking voice information captured by virtual assistant Alexa for ad targeted purposes despite stating it will not do so
…the Washington complaint refers to an article by the New York Times in 2018 titled “Hey, Alexa, What Can You Hear? And What Will You Do With It?” which disclosed a patent application Amazon filed for a “voice sniffer algorithm” that could be used to analyze audio on devices in almost real time and use the data gathered to target ads to the speakers. It notes that Amazon responded in a statement that it did “not use customers’ voice recordings for targeted advertising.”
Separately, Amazon sells a fitness band called Halo, which apparently can understand the tone of your voice, on the grounds that the tone of your voice plays a role in health.
BigTech/ Data / Platforms
Deutsche Telecom now installing Movius app to monitor its bankers
…Movius has been used in many other companies that belong to highly regulated industries such as banks and telecoms. However, they can evade the monitoring as the app is not installed on personal phones (although that would be against the firm’s code of conduct). Businesses have in the past been fined by regulatory bodies for failing to monitor communications with clients.
Meta facing proposed class action for collecting data from healthcare providers’ websites and apps
…According to the suit, “Facebook knows (or should have known) that its Pixel tracking tool is being improperly used on hospital websites resulting in the wrongful, contemporaneous, re-direction to Facebook of patient communications to register as a patient, sign-in or out of a supposedly “secure” patient portal, request or set appointments, or call their provider via their computing device. This unlawful collection of data is done without the knowledge or authorization of the patient…When a patient communicates with a health care provider’s website where the Facebook Pixel is present on the patient portal login page, the Facebook Pixel source code causes the exact content of the patient’s communication with their health care provider to be re-directed to Facebook in a fashion that identifies them as a patient”.
The plaintiff alleges that Facebook violates various US privacy laws, breach of contract, duty of good faith and fair dealing, negligent misrepresentation and unfair competition laws.
The complaint is more serious than the usual data collection cases (Facebook settled its privacy case in which users claimed that it had tracked browsing activity from logged off users) because it concerns such sensitive data.
Why would Facebook need health info?
The complaint explains:
Facebook monetizes the information it receives through the Facebook Pixel deployed on medical providers’ web properties by using it to generate highly-profitable targeted advertising on- and off-Facebook… The targeted advertising Facebook offers for sale includes the ability to target patients based on specific actions that a patient has taken on the medical providers’ websites… For example, Facebook could target ads to a patient who had (1) used the patient portal and (2) viewed a page about a specific condition, such as cancer…
The plaintiff alleges that Facebook violates various US privacy laws, breach of contract, duty of good faith and fair dealing, negligent misrepresentation and unfair competition laws.
The complaint is more serious than the usual data collection cases (Facebook settled its privacy case in which users claimed that it had tracked browsing activity from logged off users) because it concerns such sensitive data.
How does Pixel work?
Crudely (and as I understand it), there is a code which is embedded in the advertiser’s (eg. GAP, BMW etc) website, generated within the Facebook’s Ad Account. The advertiser can then track what users do in response to ads placed on Facebook or Instagram platforms. The advertisers can also get access to the profile of the user who has responded to the advert by clicking through (preferably to sale) the ads, so that Facebook or Instagram can look for other users with similar profile to display the same ads. This process maximises the chance of achieving higher return on investment (so-called targeted advertising) for the advertiser and Facebook/Instagram can justify charging more for their ad services.
Facebook settles algorithmic bias claim that was said to violate the Fair Housing Act with the US Department of Justice
…Facebook was alleged to have carried out ad targeting for housing based on user profiles comprised of protected categories such as race, religion, and sex. This meant that certain groups were excluded from being considered eligible to be exposed to certain housing ads. Facebook has committed not only to rebuild its housing ads targeting system, but to other areas such as social issues, elections or politics, credit and employment to ensure there is no bias. If more cases like this were to surface, tech companies could be looking at further regulatory pressures to make algorithms transparent.
Uber and Lyft sued by drivers for breaching competition law in California
…In the suit, the plaintiff drivers say that, despite asserting that their drivers are independent and not employees, they have excessive control over drivers unfairly driving down drivers’ profits. The main complaints are as follows:
• Unfairly fixing the prices that drivers can charge their customers which reduces competition because were the drivers able to decide their own prices they could earn more, and customers could pay less. For example, if drivers can decide its fees, it could charge less depending on how profitable the ride is estimated to be. The determination of the fee by Uber and Lyft is carried out by the respective undisclosed algorithms, and so lacks transparency.
• Practice of getting drivers to accept the ride without disclosing the customers’ desired destination is unfair. A driver can be made to drive long distances without being able to earn anything on the way back.
• Cost to drivers are on the rise as gas prices surge yet the companies’ takings appear to have increased.
• Exclusive commitment incentives mean drivers cannot in practice viably work for both Uber and Lyft.
EV/Connected Cars
Google’s licensing terms for the use of Google’s map services in vehicles to be scrutinised by the German competition authority
…it is assumed that powerful automotive companies in Germany have lobbied for this. The competition authority said “the fact that Google makes the use of its services in vehicle infotainment systems subject to very strict terms of use applicable to its “Google Automotive Services” could restrict competition even further”. At issue in particular is the restrictions on the combination of its own map services with third-party map services, for example when it comes to embedding Google Maps location data, the search function or Google Street View into maps not provided by Google.
Metaverse/NFT
Metaverse Standards Forum is launched
…to build an “open metaverse”, by promoting interoperability (that means – I assume – one metaverse can connect to another; for example, your avatar can transition to another, payment using the same currency across the metaverses, end –to-end encryption and other security and privacy measures).
Which Standard setting organisations will develop the interoperability standards?
Those that are mentioned are The Khronos Group, the World Wide Web Consortium, the Open Geospatial Consortium, the Open AR Cloud, the Spatial Web Foundation.
Areas of Focus
This will be on projects such as:
- implementation prototyping,
- hackathons
- plugfests,
- open-source tooling to accelerate the testing and adoption of metaverse standards
- developing consistent terminology and deployment guidelines.
Who is in?
Anyone can join. Founding members include: 0xSenses, Academy Software Foundation, Adobe, Alibaba, Autodesk, Avataar, Blackshark.ai, CalConnect, Cesium, Daly Realism, Disguise, the Enosema Foundation, Epic Games, the Express Language Foundation, Huawei, IKEA, John Peddie Research, Khronos, Lamina1, Maxon, Meta, Microsoft, NVIDIA, OpenAR Cloud, the Open Geospatial Consortium, Otoy, Perey Research and Consulting, Qualcomm Technologies, Ribose, Sony Interactive Entertainment, Spatial Web Foundation, Unity, VerseMaker, Wayfair, the Web3D Consortium, the World Wide Web Consortium, and the XR Association.
Worth noting the lack of Alphabet, Amazon, Apple, Roblox and Samsung says the FT. And about time, said the Chinese government led metaverse industry group, The Metaverse Industry Committee, probably.
Anonymous cryptohacker gets served NFT restraining order
…following the New York Supreme Court’s approval of the method of service. The case was brought by LCX, a fintech company. It has been able to track down and freeze around 60% of the stolen funds thus far, using “algorithmic forensic analysis” and support from Circle, the issuer of USDC (digital stablecoin pegged to the dollar) and Coinbase (cryptocurrency exchange – incidentally, it has a stake in Circle).
LCX was able to airdrop its “service token” to the wallet the investigation identified as being in possession of the stolen crypto.
In the Spotlight
US users’ data collected through the TikTok app to be transferred to Oracle’s data servers
…this development follows a report that US users’ data had been accessible by TikTok’s staff in China, despite stating that the data will be localised in the US. The Trump administration had ordered a forced sale of TikTok to a US entity (Microsoft and Oracle being contenders) but that order was reversed by the Biden administration instead, calling for a broader review of national security risks posed by foreign controlled apps. TikTok could compromise US security if it had US user data –for example, it could blackmail US military personnel by disclosing their (or their children’s) likes/ sexual orientation / religious beliefs / geolocation of homes etc which they would know because TikTok will have granular knowledge on users’ predilections and usage. In India, TikTok is banned in the interests of national security.
The secret of TikTok’s success
Not only has TikTok experienced a huge rise in popularity in the last few years, it has managed to maintain its appeal to the fickle masses (ie: Gen-Z). It streams short user generated videos, one after another. Facebook users’ limited eyeball time have been taken away by TikTok, Facebook has complained. Whilst sceptics have suggested that Facebook’s comments were made to appeal to the antitrust watchdogs that it is not as dominant as they think, there appears to be no doubt that TikTok has created a significant dent in Facebook’s engagement rate. TikTok’s new concept was to play into the inherent laziness of us human beings. It was a hit. Through machine learning, TikTok works out what contents that particular user might like to see meaning users don’t have to actively search for contents of his or her interest. In order to compete, Facebook launched a similar service called Reels, but it has not caught up. It has been said that users treat Facebook as a space to catch up with friends and family and so a social media service similar to TikTok was not quite what the users were looking for.
Then, is YouTube a strong contender?
YouTube, like TikTok, is in the business of displaying user generated content usually unconnected to the user’s friends and family sphere. It has also released a similar service to TikTok, called YouTube shorts. One might expect YouTube shorts to become equally as successful as TikTok. However, as explained in a YouTube clip (contents are in Japanese), this is not so because the critical algorithm isn’t the same. What YouTube does is profile the user and its browsing history, the channels subscribed to and works out what other people of similar profile have enjoyed watching. A user who likes watching others play first person shooter games might be introduced to other top players playing the same or similar first person shooter games. The key here is that YouTube shows you similar material which has a good track record or a high number of views. YouTube, owned by Google, is a master of indexing, categorising and ranking content, so it knows to provide the user with a highly ranked, relevant content. Not TikTok. What TikTok does is to take any newly uploaded clip, release it to a number of viewers and see whether there is engagement. If there is engagement, they analyse the profile of those users and the same clip is shown to people of similar profile. At which point the algorithm might be even further refined. Thereby, TikTok can provide users with surprise encounters and avoids users getting bored. It is also much easier for new content creators to break the TikTok userbase, because they are guaranteed to be put in front of a certain number of users which means long time TikTokkers cannot rest on their laurels – as may be possible on YouTube. Every content creator on TikTok have to keep inventing and evolving to keep users engaged, and not be defeated by the daily wave of newcomers. As the clips on TikTok have historically been shorter than YouTube, it also has the advantage of clocking up more data points for any particular user over the same period. The TikTok algorithm is therefore in theory, better at working out which clips will appeal most to any particular user.
But actually, TikTok and YouTube co-exist symbiotically – at the moment. Content creators can get paid well by YouTube, but not so much from TikTok (sponsors of influencers (eg. wearing clothes from a particular brand) might pay them if they perform well on TikTok, but TikTok itself has not historically rewarded content creators as much, as I understand it). Therefore, content creators aim to be discovered on TikTok by uploading a short and impactful clip and get users to check them out on YouTube where they will upload a much longer video (and so more opportunities to insert ads, leading to increased revenues for both the content creators and the platform). At some point TikTok might want to claw back some of those revenues – the same happens when a user uploads a link to a YouTube clip on Twitter. In fact, Elon Musk has explained to Twitter employees that when he takes over, he aims to retain some of those revenues that are being generated by YouTube. The battle of the content services platforms is still evolving…