Artificial intelligence opens us new application spaces, previously inaccessible on an algorithmic basis: activities of perception, classification and prediction.
But every time we digitize an activity, the speed of computers is such that the scale transcends traditional human possibilities.
Scenarios of mass customization, mass conditioning, mass surveillance, etc. open up.
If these activities are carried out by governments, what are the appropriate checks and balances?
Under what power would Montesquieu have placed egovernment?
What is the most suitable IT architecture for an institutional democratic architecture?
Very well done, Mrs. Merkel!
She’s very quick to react to Apple’s recent announcement, albeit today only limited to the US, clearly marking an independent position wrt the GAFAMs.
Angela über alles
The G7 agreement on the baseline global tax for corporations is a major diplomatic accomplishment and great news.
When implemented it will help contrast the states’ race to the bottom to attract multinational corporations fiscal headquarters.
Spain has already endorsed the plan (Spain is not part of the G79.
Negotiators hope to expand the deal to the G20 in 2021. G20 account fo aprox. 80% of international trade.
But let’s not forget a few facts:
- the average global corporate tax rate is 23% (it was about 40% in the 80s)
- in the US this is unlikely to pass soon as international treaties require a two-thirds majority in the Senate, and Republicans are voting against *everything* Biden’s administration is proposing (The US senate is 50-50 divided)
- this won’t fix the issue of profits shifting of IT companies anytime soon for two reasons. 1) treaties enabling Dutch sandwich still exist and will exist for many years to come; 2) to elide the previous (1), IT corporate taxes need to be fully payed where users reside, not where corporate HQs are fiscally located, and this is now marginal and will require some time as well
Still, it’s a major first step in the right direction. More need to follow.
Prof. Lessig correctly says that “code is law”, as software controls who can do what, without exceptions.
There has been growing discontent around the world towards the control of code by the multinational giants GAFAM, and many states are introducing “digital sovereignty” rules to assert their right to determine how software should be managed.
In fact, limiting oneself to the sole consideration that they don’t like GAFAM-controlled code, is short-sighted. By putting more focus on it, before making missteps, one should ask, “and so, who should control it ?”
Control of the public administration software/data is a major power as enables control on people’s lives.
In the end, if the State (executive or judiciary branches) say you are not entitled to a service/you are not trustworthy, your life can become a nightmare, and it will be increasingly so over the coming decades.
eGovernment software is another type of State’s power, alongside the executive, judiciary and legislative branches.
eGovernment software operation, in a democracy, requires checks and balances.
Under which branch would Montesquieu position the control over State’s eGovernment software ? Who should decide its features, functions, permitted or vetoed users ? Who should run it ? Who should control it ? What type of control, accountability, transparency, oversight ?
Can we trust, over the long period, the executive branch alone to build and run egovernment software, considering potential whims and prejudices of the pro-tempore minister/secretary and the opacity of vertically integrated structures in the executive power?
Or should we put its control under the judiciary branch, with their checks and balances, by envolving private entities in the provision of public services; audit them and impose them transparency obligations ?
Or should we give the legislative branch inspective and prescriptive powers over the development and provision of software ?
Democracy is a fragile construction; oversight on egovernment software’s power is fundamental, as we must remember that States can enter an authoritarian regime simply by voting, whilst it’s not that easy to exit from an authoritarian regime.
We should deeply think about these issues and put in place, during the sunny days, the structural safeguards that help prevent dark days.
Otherwise, after Shoshana Zuboff awoke us after a dozen years of “social torpor” revealing the “surveillance capitalism” that was in front of ours eyes, in a couple of years we could wake to a “surveillance egovernment”, with the major difference that states, as opposed to companies, have the monopoly of force.
We need to think and build now safeguards to mitigate the risk of waking up one day in a condition of “Error 404 – Democracy not found”.
There was an advertisement when I was a kid that spelled “What to give to a man who has everything ? More of the same”. It was a whiskey.
Now we have an issue about excessive profiling, excessive market power, excessive political influence.
How can we intervene ?
Apple would like to curtail advertising based business models, thanks to a tight control of their OS and app environment, which would stregthen their gatekeeping position on the app market.
Google would like to get rid of cookies, promoting a technology that would make them more than today, the platform controlling the online advertising business.
Facebook would like to change content moderation laws in a way that would reinforce their size and power.
Fact is that times for a digital regulation are approaching, both because of likely upcoming antitrust sanctions and because of new ex ante regulation.
They lost their last veil of appearance of trustworthiness. They should be ashamed of their proposals.
I am eager to read Microsoft and Amazon’s.
Un’ulteriore criticità della proposta di Google riguarda il possibile assetto competitivo del mercato della pubblicità online. Google possiede decine di brevetti che riguardano il “Federated Learning”, che è il meccanismo alla base della proposta FLOC, ed il suo uso.
Se non indirizzato, il tema della proprietà intellettuale su un ambiente che dovrebbe essere destinato ad essere la base del nuovo meccanismo di pubblicità online, potrebbe avere effetti anticompetitivi notevoli.
Google’s initiative to get rid of traditional user-identifying cookies in its Chrome browser has sparked some concerns in the United States Department of Justice regarding how it will affect competitors in the advertising space. Investigators are looking into whether the move will give Google a bigger advantage in targeted advertising, a market it already leads in.
I hear people saying that a vaccination passport will be paramount to enable resuming international travel.
AFAIK, it is not at all clear that a vaccination will make people not contagious. (nature)
Covid19 vaccines helps not getting the covid19 disease, it doesn’t imply that people is not going to be contagious.
If you are a vaccinated person travelling to the UK, the vaccination passport only tells the UK administration that you are not going to cause burden onto the NHS (National Health Service).
As we will be more and more vaccinated, to resume travelling, we will need to strenghten the capacity of PCR testing (swabs).
With swabs certifications we can show that, up to 2 days before, we were not contagious, thereby minimizing the probability of contagion.
AT&T ends zero rating.
This is more than a win for net neutrality, it’s the confirmation that – contrary to old telcos’ narratives – zero rating made little business sense. (as those readers who follow me since several years already knew very well)
If we look at the huge data packages that can be bought for few euros/month, prices brought down by competition, Zero rating makes little sense for consumers.
It would make sense if lack of competition would cause rival mobile operators artificially limiting the size of data packages provided to users. But when you have competition, you will always have an operator winning in the market by offering competitive prices with huge data packages not restricted to specific online services (those that pay the operators to be included in these zero rating packages).
So, with the present trend of reducing data costs, determined by the evolution of electronics, Zero rating could make sense only in an environment of limited or absent competition. It’s an anticompetitive tool, as FCC recognized in the past wrt. AT&T’s
Lastly, the argument that “the internet does not recognize state borders” is an obvious nonsense. The California net neutrality law does not apply to AT&T customers in other states.
Source : AT&T
Impact of California ‘Net Neutrality’ Law on Free Data Services
March 17, 2021 at 8:00am
California has enacted a “net neutrality” law banning “sponsored data” services that allowed companies to pay for, or “sponsor,” the data usage of their customers who are also AT&T wireless customers. Unfortunately, under the California law we are now prohibited from providing certain data features to consumers free of charge.
Prior to California’s law, sponsored data customers were able to browse, stream and enjoy applications from sponsors without using their monthly data allowance. AT&T video providers utilized sponsored data to offer Data Free TV, allowing customers to stream their favorite movies and shows over their AT&T wireless service without it counting against their wireless data plan. AT&T Mobility has for years openly invited any entity to become a wireless data sponsor on the same terms and conditions. Since it began, our sponsored data service, and competing offers from other wireless providers, have delivered significant benefits and saved consumers money. Consumers also have enjoyed an explosion of video streaming services.
We regret the inconvenience to customers caused by California’s new “net neutrality” law. Given that the Internet does not recognize state borders, the new law not only ends our ability to offer California customers such free data services but also similarly impacts our customers in states beyond California.
A state-by-state approach to “net neutrality” is unworkable. A patchwork of state regulations, many of them overly restrictive, creates roadblocks to creative and pro-consumer solutions. We have long been committed to the principles of an open Internet. We deliver the content and services our customers want because it’s what they demand, not because it’s mandated by regulation. We also believe Internet access should be available and sustainably affordable to all Americans, and strongly advocate for Congress to adopt federal legislation to make that possible while providing clear, consistent, and permanent net neutrality rules for everyone to follow.
Google is sending the text below (in italian) to influencers who monetize youtube
by doing so, the U.S. taxes income for content generated in Italy, by Italian citizens, watched by U.S. public.
it is a change in the current taxation principle: the income is no longer taxed where the producer or where the server is but where the consumer is.
Ti contattiamo perché entro la fine dell’anno (a partire da giugno 2021) Google sarà tenuta a dedurre le imposte statunitensi dai pagamenti corrisposti ai creator al di fuori degli Stati Uniti. Nel corso delle prossime settimane, ti chiederemo di fornirci i tuoi dati fiscali in AdSense, così da determinare il corretto importo delle deduzioni fiscali eventualmente applicabili nei tuoi confronti. Nel caso in cui non dovessimo disporre dei tuoi dati fiscali entro il 31 maggio 2021, Google potrebbe avere l’obbligo di dedurre fino al 24% dal totale delle tue entrate globali.
Che cosa devo fare?
Nel corso delle prossime settimane, riceverai un’email con l’invito a inserire i tuoi dati fiscali in AdSense. Lo strumento fiscale di AdSense, disponibile online, ti guiderà attraverso una procedura in sei passaggi con cui potrai determinare se sei soggetto a imposte statunitensi rispondendo a una serie di domande. Consulta il Centro assistenza per scoprire di più sulle modifiche in programma e visionare un elenco dei dati fiscali da predisporre.
Qual è il motivo di queste modifiche?
Ai sensi del Capitolo 3 del Codice tributario statunitense (Internal Revenue Code), Google predispone la raccolta dei dati fiscali di tutti i creator con monetizzazione attiva che risiedono al di fuori degli Stati Uniti per poi procedere, in determinate circostanze, a dedurre le imposte laddove i redditi siano generati da spettatori residenti negli Stati Uniti.
Per i creator residenti al di fuori degli Stati Uniti, a breve inizieremo ad aggiornare i nostri Termini di servizio per apportare le modifiche necessarie, in base alle quali le entrate corrisposte da YouTube verranno considerate delle royalty ai fini dell’imposizione fiscale negli Stati Uniti. Ciò potrebbe avere delle ripercussioni sulle modalità di tassazione delle tue entrate. Nei casi previsti dalle leggi statunitensi, Google provvederà inoltre a dedurre le imposte.
Che impatto avranno queste modifiche sulle mie entrate?
Se fornisci i tuoi dati fiscali, possono essere applicate imposte statunitensi solo sulle entrate mensili statunitensi ricevute da AdSense, ovvero sulle entrate generate da spettatori negli Stati Uniti attraverso visualizzazioni di annunci, YouTube Premium, Superchat, Super Sticker e abbonamenti al canale. In assenza di dati fiscali, potresti essere soggetto a ritenuta d’acconto con l’aliquota predefinita più elevata per i privati, pari al 24% delle entrate totali percepite a livello mondiale. Per stimare il potenziale impatto sulle tue entrate, segui queste istruzioni.
Il team di YouTube
the negative energy that is poured into driving violence emerges in large part from the sense of protection and power we have, encapsulated in our automotive exoskeleton
a similar sense of protection and power we experience at home, unseen, behind the keyboard, and we pour out our negative tension in online hatred
then there’s some complement of a “horde” effect like the one people feel at the stadium where they are surrounded by people that agree with them; an echo chamber in social network terms.
Can a president of a nation be deleted from Twitter or Facebook ? They are private services that have no obligation of universality, so the short answer is: yes, there is not an a-priori right to participate in a social. In some cases, however, some courts have ordered the companies to reinstate the excluded user because his expulsion was not legally valid and would have generated damage.
Can being on Twitter, Facebook, Instagram, Tiktok be an a priori right? If we’d ask a teenager, the answer would obviously be yes. A very important part of their relational life takes place in that environment. Perhaps their role as a public service should be recognized.
But who is responsible for the things that are said on social media? In the U.S., according to section 230 of the “decency in communications” law, the operator is not responsible for the content hosted. They becomes so if, in the face of a notification, they fail to remove them. In Europe, with respect to violent/defamatory content, the eCommerce Directive applies with provisions very similar to section 230. Assessing the lawfulness of content is a task for a court. Entrusting the preventive removal of content to a private party, as established for copyrighted works by the Copyight Directive in Europe, is equivalent to entrusting these companies a first level of judgement.
Are Trump’s tweets illegal? It’s practically impossible to say. A court would have to determine that. Certainly they are full of falsehoods, but lying is generally not illegal. It becomes so when harmful effects for someone arise from the lie (e.g. defamation), cases that must be evaluated by a court.
In fact, there is one case in which lying is illegal in itself: this is stock manipulation, whereby it is not necessary for the price of shares to be altered as a result of the lie. The lie itself, aimed at altering a stock market price, is in itself a crime.
Now there are questions about whether the removal of Trump’s accounts is a legitimate and appropriate act or if it calls for regulatory changes. Trump himself has intervened several times on section 230 and in Europe the Commission presented last December a new regulatory proposal for digital services.
Intervening to regulate social networks is a very delicate issue: some fundamental rights are touched. On the merits, one can detest the messages of incitement to violence that circulate in some parts of the Parler app, but it should not be Amazon that shuts down their servers and Apple that makes it impossible to install the app on iPhones by removing it from the app store. Internet mega-nationals cannot be the guardians of people’s rights. People must be free to express themselves, any wrongdoing must be ascertained only by courts.
Before social networks, people could express themselves by addressing small groups in total freedom: think of the speakers’ corner in London, of film forums or bars. Those who could address large audiences were the publishers, for whom there were (and are) different rules that provide for an responsible editor, the obligation to correct, etc.. With social media, the potential of expression of the individual is extended from the previous limited audience to an audience characteristic of large publishers, but without specific regulation.
On the basis of the considerations that “free speech is not free reach” and that with “large audiences comes great responsibility”, one could envisage a “regulatory ladder” that places increasing constraints on the basis of the audience.
For example, mantaining the present regime for users with less than 20 thousand followers; provide the obligation to a possibility of indirect identification (for example with a mechanism similar to the one in use at airoports’ wifi ) for users with a number of followers between 20 and 100 thousand; provide a prior identification for users with more than 100 thousand followers and, for people with larger audiences, measures similar to those that regulate publishers, such as the right to rectification.
As we learned with COVID, the speed of spread is a crucial element in outbreaks. There have been experiments that have shown that virality of some contets can be limited by introducing frictions to the spread. This technique was used in a radical way by Twitter, preventing the retweeting of some of Trump’s twits. A similar measure could be administered more gradually: some posts could be exposed gradually, for example reaching an incremental 1% of the user’s followers per hour, allowing for intervention of a court judgment.
This measure of “rate limiting”, which is common practice in some IT activities, could be envisioned only for large operators and triggeret for only those twits that an artificial intelligence judged as potentially illicit.
Finally, for some categories of particular relevance to society, a tort of malicious information manipulation could be envisaged, only for persons with very large audiences, and prosecutable in court by a lawsuit by those who are directly or indirectly harmed. The fines could be imposed according to the number of followers, let’s say for example a maximum of 5 cents per follower.
Social networks are here to stay and their relevance is likely to increase. For those who do not engage in content curation, providing for responsibilities similar to publishers would be counterproductive, putting their livelihood at risk and especially disadvantaging new market entrants. Providing obligations and constraints for all users is equally disproportionate and a harbinger of rights compression.
On the other hand, there will certainly continue to be increased pressure for accountability in the online publication of content that can have dramatic effects on societies. We must be very careful not to privatize justice, reduce guarantees for individuals and limit acquired rights.
Extreme regulation, such as no standards or black/white measures for all, cannot work. If we want to make progress, we need to think about regulating the behavior of users and not only of the intermediaries, graduated approaches according to audience level, and gradual sanctions.
So today VP Vestager and Commissioner Breton introduced the drafts of the new EU digital regulation.
A very nice piece of work, I must say.
There are some very fine grained measures and more general principles. Among the latters I think that the regulatory ladder, interoperability and device neutrality are particularly worth of consideration. Also the Digital Services Coordinator is a good general principle: a nice idea to improve effectiveness of the regulation itself.
Let’s think of future amendments…
Auctioneers implement an number of strategies to push auction prices up.
Not only the auction mechanism itself like the second price auction, which is well known and understood, but also other more sophisticated strategies like some form of “semantic expansion” which automatically expands the size of the bidding base, or granting to non profits some “money equivalent” to spend in their auctions, or suggesting the bidding price, or altoghether completely managing the bidding price for the bidder (behaving like they are not the auctioneer and the party selling the product, but a mere technically “neutral” part).
Suppose you have a shop, you want to sell shoes online and the bid winning price for your ad is 10c.
The auctioneer goes to the non profit X that gives away shoes to the poor and give them a 10K grant as a “credit” which can be spent on their own adverts (which, by the way, have a zero marginal cost).
They don’t give a contribution in kind (“exposure of the non profit X advert”) which would be accounted for 0USD, given it has a zero marginal cost.
Instead, the non profit X will bid for the “shoe” keyword on the search engine using this “chinese money” (it’s not real money as it cannot be cashed out for fiat dollars) thereby pushing up the price “real bidders” (like your shoe shop) have to pay to win the auctions to place their adverts on search results. (Would that contribution to the non profit also account for a tax break ? I don’t know).
Another example of “non linear behaviour” in auction could be done with rebates. Suppose you have a B&B close to Hoover Dam that you promote via an auction. You would bid for “B&B Hoover Dam” and would compete against online travel agencies (OTA) that bids for the same keywords.
Suppose the auction winning price is 30c per clic. If you win it, then you’d have to pay the 30c. But OTAs are among the highest spenders of digital ads. We don’t know if, at the end of the year, the auctioneer gives the OTA a rebate thereby effectively lowering their true cost to win that clic for the B&B at the Hoover Dam.
It might well be that, in the end, thanks to rebates, the OTA ends up really paying 25c for their “30c bid win”; and with these 25c they win over you, when you have offered up to 29c.
We don’t really know if this kind of things happens. But we once run an experiment for accomodations close to Milan pushing the limits of bid offerings to unreasonable levels. A popular OTA won the bids at a price higher than the cost of the rooms themselves. It is indeed possible that the OTA decided they’d loose money on that transaction, but given that bids run automatically based on algorithms configurations, it is also possible that their euros had a different value than our euros.
The key point here is that we don’t know for sure, because it’s the same entity producing the good that’s being auctioned (that has a zero marginal cost) running the auction and choosing the winner, without a third party inspecting the process and certifying everything runs smoothly and fairly.
This reminds me when FTC and SEC had powers in case of abuses between banks’ retail and investment arms, one telling the other how good was their product and the price to pay. Some ex ante regulation works best…
I also think, in some cases, we will need some other provisions re. search…
As a general rule, IMHO, search functions should be separated and audited from other applications and the user should have the choice of selecting different search engines (this has already happened with browsers where you can choose a different search engine).
This is going to become the more evident with voice assistants.
When you search something on your screen, you get a list of results and, among those few, you choose which one suits you best to answer your need.
When you ask a voice assistant for something, you directly get a result, with no shortlist selection.
Tying the search function to the device, with no possible choice for the user to change it, creates a really strong lockin to the voice search engine. You would’t buy a full set of new voice assistants (several at home, car, etc.) just to replace the search engine.
When (eventually) antitrust would kick in, the durable effect on the market has already happened.
These devices would likely be considered as a specific market, separated from all other related and intertwined interests of the supplier (e.g cloud services and others) and potential remedies would have a very limited impact on the overall interest of the supplying company. The possible sanction would not be a disincentive to misbehave.
So we’ll need some ex ante measures as well.