February 2022 - nexa - server-nexa.polito.it

Re: [nexa] Carbon footprint of ML

by Giuseppe Attardi

> On 24 Feb 2022, at 17:41, Stefano Quintarelli <stefano(a)quintarelli.it> wrote: > > quanti di questi studi sono stati finanziati da G ? Lo studio pubblicato su IEEE è *di* Google, in collaborazione con David Patterson, di Berkeley, un notissimo informatico, inventore del RISC. L’articolo ribatte a delle affermazioni simili a quelle che sono state discusse anche in questa lista: The answers are grim: Training such a model would cost US$100 billion and would produce as much carbon emissions as New York City does in a month. e tenta di Explain why published faulty estimates and extrapolations are 100x–100,000x higher than the real carbon footprints. Il blog cita un lavoro dall’Università del Massachusetts, che include Andrew McCallum, un professore che conosco personalmente, ed è finanziato da fondi pubblici oltre che dalla fondazione Chan Zuckerberg. Sappiamo bene che i modelli di Deep Learning continuano a crescere notevolmente di dimensioni: nel 2018 segnalai il problema a Cedric Villani, che stava.lavorando al rapporto Ai for Humanity per Macron, che rimase stupito delle mie previsioni. L’articolo sostiene che a fronte di questa crescita notevole, si riesce a contenere la crescita dei consumi con i miglioramenti introdotti nella tecnologia (algoritmi, TPU, efficienza dei datacenter) Tuttavia queste tecnologie rimangono appannaggio di pochi e rendono più difficile ad altri ricercatori di restare al passo. Sottolineo a questo proposito quanto dicono proprio in questo secondo articolo, dove chiedono la costituzione di risorse cloud pubbliche per la ricerca: Academic researchers need equitable access to computation resources. It is more cost effective for academic researchers, who often work for non-profit educational institutions and whose research is funded by government entities, to pool resources to build shared compute centers at the level of funding agencies, such as the U.S. National Science Foundation. Unlike money spent on cloud compute, however, that invested in centralized resources would continue to pay off as resources are shared across many projects. A government-funded academic compute cloud would provide equitable access to all researchers. https://arxiv.org/pdf/1906.02243.pdf Le tecnologia dell’AI avranno effetti dirompenti, e quindi dovremmo porci l’obiettivo di democratizzarla, ossia di renderla disponibile a tutti, investendo nel suo sviluppo e padroneggiandola, anziché stare a guardare e lamentarsi di quanto fanno gli altri. — Beppe > il problema con i finanziamenti non pienamente disclosed e' che ti erodono il trust nella ricerca citata da chi finanzia... > > On 24/02/22 13:03, Giuseppe Attardi wrote: >> Trasmetto per conoscenza (ambasciator non porta pene) la risposta di Google alle critiche sui costi energetici del ML: >> https://ai.googleblog.com/2022/02/good-news-about-carbon-footprint-of.html <https://ai.googleblog.com/2022/02/good-news-about-carbon-footprint-of.html> >> — Beppe >> _______________________________________________ >> nexa mailing list >> nexa(a)server-nexa.polito.it >> https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa

Feb. 25, 2022

I’ve been waiting 15 years for Facebook to die. I’m more hopeful than ever | Cory Doctorow | The Guardian

by Alberto Cammozzo

<https://www.theguardian.com/commentisfree/2022/feb/24/ive-been-waiting-15-y…> I’ve been praying for Facebook’s collapse ever since it attained liftoff. In a 2007 article, I predicted that “your creepy ex-co-workers will kill Facebook” by demanding to know why you won’t “friend” them, prompting an exodus to the next platform. That was the social network cycle back then: a new network opens, and you and the people you genuinely like enjoy a rollicking group chat until all the people you have to pretend to like show up. That’s the double-edged sword of products that rely on “network effects” – the economists’ term for a product that gets better when more people use it. Sure, you might join Facebook because your friends are all there (and more people might sign up because you’re there), but that also means that every time your friends leave Facebook, it’s a reason for you to leave, too. My prediction failed. For a decade and a half, Facebook resisted the fate of all the social networks that preceded it. In hindsight, it’s easy to see why: it cheated. The company used investor cash to buy and neutralize competitors (“Kids are leaving Facebook for Insta? Fine, we’ll buy Insta. We know you value choice!”). It allegedly spied on users through the deceptive use of apps such as Onavo and exploited the intelligence to defeat rivals. More than anything, it ratcheted up “switching costs.” “Switching costs” is another economic term: it means “the price you pay when you switch from one service to another.” Switching from Facebook to a rival means saying goodbye to the communities, friends and customers you hang out with on the platform. Normally, tech has really low switching costs: want to change from T-Mobile to Verizon? Just port your number. Your friends don’t even have to know you did; they can still call you and you them. Tech’s rock-bottom switching costs are what kept the industry so dynamic in its early days. Microsoft could deploy an army of corporate salespeople to turn Microsoft Office into an industry standard, then Apple could come along and reverse-engineer the Office formats and make the interoperable iWork office suite. That means that Windows users could switch to the Mac and open their Word docs in Pages, their Excel spreadsheets in Numbers and their PowerPoints in Keynote. It’s different for Facebook. The company’s ascendancy coincided with an overall concentration in the tech sector, and, with it, laws that protected winners of the latest round of the interoperability wars from new challengers. Apple was able to reverse-engineer its way out of the Microsoft Office trap, but woe betide a company that tries the same trick on Apple – try to make a program that lets you run iPhone apps on an Android device, or read the media files you buy in Apple’s book, movie or music stores, and you will quickly discover that the law is now on the sides of the giants, not the upstarts. That same legal shift is how Facebook has kept its switching costs high. Fifteen years ago, it was safe to make a Facebook-MySpace bridge that would let you leave MySpace but stay in touch with your friends there by scraping your MySpace inbox and moving the waiting messages to your Facebook inbox. Try to build one of those bridges today – blasting an escape tunnel through Facebook’s walled garden – and Facebook will sue you until the rubble bounces. But high switching costs have their limits. If you make your service terrible enough, a certain number of users will find the cost of switching preferable to the pain of staying. And as users leave, network effects start to work in reverse: though every user that joins makes your service more valuable, every user that leaves makes the service less valuable. If you’re only on Facebook to stay in touch with a small group of friends, each one of those friends who departs makes it easier for you to make the jump, too. And once you go, it’s even easier for the rest of the group to bail. This is very bad news for Facebook. After years of slowing US growth, Facebook just experienced its first-ever US shrinkage, which precipitated a $230bn stock crash, the largest in global corporate history. Though most of Facebook’s users are global, its US users generate far more profit than users in the rest of the world. Losing a US user is expensive. Even more important: the US is Facebook’s home base, and its US user base is its main bargaining chip in resisting US regulation, and in securing US support in its regulatory battles abroad. Speaking of regulatory battles abroad: Facebook is on the brink of having its business model declared illegal under the European Union’s General Data Protection Regulation (GDPR). Fending off that scenario will depend on vast capital expenditures and friendly European regulators, and Facebook’s running short on both. Oh, and Europeans are Facebook’s second most valuable users. Admittedly, when a company’s shares decline, it’s not like the company itself has lost any money – those losses hit shareholders, not the business itself. However, Facebook’s costs and share-price are intimately bound together, thanks to tech firms’ reliance on stock grants as a way of scoring a discount on their wage-bills. Engineers, lawyers, and other high-paid, in-demand professionals are glad to take much of their compensation in stock, betting that the company’s share price will balloon and that they can cash out their shares and keep their winnings, thanks to the tax-preferred status of capital gains – in most of the world, the wages you earn for doing useful work are taxed at a much higher rate than the winnings you get from lucky bets on stocks. Even before its stock fell off a cliff, Facebook was mired in a multi-year hiring crisis. Nobody wanted to work for Facebook because it’s a terrible company that makes terrible products that everyone hates and only use because the company has rigged the system to punish users for switching. Facebook was already paying a wage premium, offering sweeteners to in-demand workers in exchange for checking their consciences at the door. Those sweeteners mostly took the form of shares, which means that all those morally flexible “Metamates” got a hefty pay-cut when the company’s stock price fell off a cliff. Expect a lot of them to leave – and expect the company to have to pay even more to replace them. Companies with falling share prices can’t use share grants to attract workers. Facebook is now famously trying to pivot (ugh) to virtual reality to save itself. It’s an expensive gambit. It’s going to alienate a lot of its users. It’s going to alienate a lot of its in-demand workers. It’s going to freak out a lot of regulators. Meanwhile, the switching costs for people who want to jump ship keep getting lower. It’s not merely that fewer and fewer of the people you want to talk with are still on Facebook. Even if there’s someone whose virtual company you can’t bear to part with, lawmakers in the US and Europe are working on legislation that would force Facebook to allow third parties to “federate” new services with it. That would mean that you could quit Facebook and join an upstart rival – say, one by a privacy-respecting nonprofit or even a user-owned co-op – and still exchange messages with the communities, customers and family you left behind on Facebook’s sinking ship. For 15 years, I’ve been waiting for Facebook to suffer the fate of every network-effects-driven success story – to experience the precipitous decline that is triggered by people leaving the service and taking the value they brought to it with them. Facebook now has to somehow retain users who are fed up to the eyeballs with its never-ending failures and scandals, while funding a pivot to VR, while fending off overlapping salvoes of global regulatory challenges to its business model, while paying a massive wage premium to attract and retain the workers that it needs to make any of this happen. All that, amid an exodus of its most valuable users and a frontal regulatory assault on its ability to extract revenues from those users’ online activities. Stein’s Law holds that “if something cannot go on forever, it will stop.” Facebook can’t go on forever. Cory Doctorow is a science fiction author, activist and journalist. He is the author of many books, including the forthcoming book Chokepoint Capitalism, with Rebecca Giblin, about monopoly and fairness in the creative arts labor market. In 2020, he was inducted into the Canadian Science Fiction and Fantasy Hall of Fame

Feb. 25, 2022

Re: [nexa] Carbon footprint of ML

by Stefano Quintarelli

quanti di questi studi sono stati finanziati da G ? il problema con i finanziamenti non pienamente disclosed e' che ti erodono il trust nella ricerca citata da chi finanzia... On 24/02/22 13:03, Giuseppe Attardi wrote: > Trasmetto per conoscenza (ambasciator non porta pene) la risposta di Google alle critiche > sui costi energetici del ML: > > https://ai.googleblog.com/2022/02/good-news-about-carbon-footprint-of.html > <https://ai.googleblog.com/2022/02/good-news-about-carbon-footprint-of.html> > > — Beppe > > _______________________________________________ > nexa mailing list > nexa(a)server-nexa.polito.it > https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa

Feb. 24, 2022

Carbon footprint of ML

by Giuseppe Attardi

Trasmetto per conoscenza (ambasciator non porta pene) la risposta di Google alle critiche sui costi energetici del ML: https://ai.googleblog.com/2022/02/good-news-about-carbon-footprint-of.html <https://ai.googleblog.com/2022/02/good-news-about-carbon-footprint-of.html> — Beppe

Feb. 24, 2022

Facebook Shadow Profiles

by Alberto Cammozzo

<https://www.cesifo.org/en/publikationen/2022/working-paper/facebook-shadow-…> Data is often at the core of digital products and services, especially when related to online advertising. This has made data protection and privacy a major policy concern. When surfing the web, consumers leave digital traces that can be used to build user profiles and infer preferences. We quantify the extent to which Facebook can track web behavior outside of their own platform. The network of engagement buttons, placed on third-party websites, lets Facebook follow users as they browse the web. Tracking users outside its core platform enables Facebook to build shadow profiles. For a representative sample of US internet users, 52 percent of websites visited, accounting for 40 percent of browsing time, employ Facebook’s tracking technology. Small differences between Facebook users and non-users are largely explained by differing user activity. The extent of shadow profiling Facebook may engage in is similar on privacy-sensitive domains and across user demographics, documenting the possibility for indiscriminate tracking.

Feb. 23, 2022

Call for Papers - Florence Seminar on Standard-Essential Patents

by Galli, Niccolo

Carissimi, sono lieto di annunciarvi che è aperta la Call for Papers per il Florence Seminar on Standard-Essential Patents<https://fsr.eui.eu/event/seminar-sep/>. L’evento è co-organizzato dall’EUI Robert-Schuman Center for Advanced Studies<https://www.eui.eu/en/academic-units/robert-schuman-centre-for-advanced-stu…> e dal Max-Planck Institute for Innovation and Competition<https://www.ip.mpg.de/en/> e si terrà a Firenze il 6-7 Ottobre p.v.. Mi farebbe molto piacere se voleste partecipare con vs interventi sugli aspetti giuridici e economici delle licenze FRAND e del contenzioso dei brevetti essenziali a norme industriali (i c.d. SEPs). La scadenza per l’invio degli abstract è il 31 maggio p.v. . Vi prego di diffondere l'informazione tra i vostri contatti e nel caso foste intenzionati a partecipare all'evento non esitate a contattarmi. A presto e buon lavoro Niccolò Galli Niccolò Galli Research Associate Competition & Regulation / Robert Schuman Centre Villa Schifanoia - Via Boccaccio 121, 50133 Firenze (FI), Italy Phone: +39 0554685781 European University Institute Via dei Roccettini 9, San Domenico di Fiesole (FI), Italy eui.eu<http://eui.eu> [cid:870C5D4E-AF10-4120-B2BD-238EEF07BBE2] The information transmitted is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any review, retransmission, dissemination, distribution, forwarding, or other use of, or taking of any action in reliance upon, this information by persons or entities other than the intended recipient is prohibited without the express permission of the sender. If you received this communication in error, please contact the sender and delete the material from any computer.

Feb. 23, 2022

Re: [nexa] Rettifica Link dell'incontro | 95° Nexa Lunch Seminar | 23 Febbraio 2022

by Angelo Raffaele Meo

vedo che anche Nexa ha scelto Microsoft Teams. Congratulazioni!!!! Il 23/02/22 12:57, Nexa Media ha scritto: > Gentilissimi, > > per problemi tecnici, > abbiamo modificato il link per partecipare all'incontro > che sarà il seguente: > https://teams.microsoft.com/l/meetup-join/19%3ameeting_YWEyZmUzYjktZjExYS00… > > > Grazie molte, > Anita Botta > > > -------- Messaggio Inoltrato -------- > Oggetto: Memento | 95° Nexa Lunch Seminar | 23 Febbraio 2022 > Data: Mon, 21 Feb 2022 10:07:09 +0100 > Mittente: Nexa Media <media(a)nexa.polito.it> > A: nexa <nexa(a)server-nexa.polito.it> > > > > Gentilissimi, > > > Vi ricordiamo che *mercoledì 23 febbraio, dalle ore 13.00 alle ore > 14.00*, > si svolgerà online il 95° Nexa Lunch Seminar, con un incontro dal titolo: > /*"L'istituto del Whistleblowing negli scenari europei e nazionali: la > tutela di chi segnala illeciti nelle amministrazioni e nelle imprese"*/. > > Ospiti: l'Avv. *Mauro Alovisio* (Fellow Centro Nexa), l'Avv. *Alessio > Ubaldi *(Fondatore Studio UB LEGAL) e l'Ing. *Giovanni Pellerano* > (GlobaLeaks Project Lead). > > Sarà possibile partecipare al seminario connettendosi al seguente > indirizzo: https://didattica.polito.it/VClass/NexaEvent > > Maggiori informazioni sull'incontro sono disponibili alla pagina: > https://nexa.polito.it/lunch-95 > > > Cordiali saluti, >   > -- > > Anita Botta > Communication Manager > > Nexa Center for Internet & Society > Politecnico di Torino – DAUIN > Corso Duca degli Abruzzi, 24 - 10129 Torino > > web: https://nexa.polito.it/ > mail: anita.botta(a)polito.it > tel: 011 090 7219 > > _______________________________________________ > nexa mailing list > nexa(a)server-nexa.polito.it > https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa

Feb. 23, 2022

Data Act: Commission proposes measures for a fair and innovative data economy

by Marco Ciurcina

Today, the Commission proposes new rules on who can use and access data generated in the EU across all economic sectors. The Data Act will ensure fairness in the digital environment, stimulate a competitive data market, open opportunities for data-driven innovation and make data more accessible for all. It will lead to new, innovative services and more competitive prices for aftermarket services and repairs of connected objects. This last horizontal building block of the Commission's data strategy will play a key role in the digital transformation, in line with the 2030 digital objectives. ... https://ec.europa.eu/commission/presscorner/detail/en/ip_22_1113

Feb. 23, 2022

Rettifica Link dell'incontro | 95° Nexa Lunch Seminar | 23 Febbraio 2022

by Nexa Media

Gentilissimi, per problemi tecnici, abbiamo modificato il link per partecipare all'incontro che sarà il seguente: https://teams.microsoft.com/l/meetup-join/19%3ameeting_YWEyZmUzYjktZjExYS00… Grazie molte, Anita Botta -------- Messaggio Inoltrato -------- Oggetto: Memento | 95° Nexa Lunch Seminar | 23 Febbraio 2022 Data: Mon, 21 Feb 2022 10:07:09 +0100 Mittente: Nexa Media <media(a)nexa.polito.it> A: nexa <nexa(a)server-nexa.polito.it> Gentilissimi, Vi ricordiamo che *mercoledì 23 febbraio, dalle ore 13.00 alle ore 14.00*, si svolgerà online il 95° Nexa Lunch Seminar, con un incontro dal titolo: /*"L'istituto del Whistleblowing negli scenari europei e nazionali: la tutela di chi segnala illeciti nelle amministrazioni e nelle imprese"*/. Ospiti: l'Avv. *Mauro Alovisio* (Fellow Centro Nexa), l'Avv. *Alessio Ubaldi *(Fondatore Studio UB LEGAL) e l'Ing. *Giovanni Pellerano* (GlobaLeaks Project Lead). Sarà possibile partecipare al seminario connettendosi al seguente indirizzo: https://didattica.polito.it/VClass/NexaEvent Maggiori informazioni sull'incontro sono disponibili alla pagina: https://nexa.polito.it/lunch-95 Cordiali saluti,   -- Anita Botta Communication Manager Nexa Center for Internet & Society Politecnico di Torino – DAUIN Corso Duca degli Abruzzi, 24 - 10129 Torino web: https://nexa.polito.it/ mail: anita.botta(a)polito.it tel: 011 090 7219

Feb. 23, 2022

Re: [nexa] LittleThings e Facebook (sequenza Twitter)

by Marco A. Calamari

On mer, 2022-02-23 at 09:19 +0100, J.C. DE MARTIN wrote: > Sequenza Twitter da leggere: > > https://twitter.com/jspeiser/status/1496174626434469889?s=21 Affidarsi alle piattaforme per la propria vita è pura follia, sia per le entità commerciali che per quelle fisiche. Self hosting everywhere! > jc > > (Sent from my wireless device; please excuse brevity and typos (if any)) > _______________________________________________ > nexa mailing list > nexa(a)server-nexa.polito.it > https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa

Feb. 23, 2022