August 2020 - nexa - server-nexa.polito.it

Re: [nexa] L’Europa ritrovi la sovranità tecnologica

by Giovanni Biscuolo

Giacomo Tesio <giacomo(a)tesio.it> writes: [...] > Il primo problema di "computing agency", "computing literacy", > "computational thinking" e similari, è che per l'informatica i computer > sono assolutamente secondari. Beh il primo problema con le etichette è che il linguaggio naturale ha semantica variabile in funzione del contesto (epoca, cultura, preconcetti) Ho usato computing perché, come sai, gli statunitensi purtroppo non hanno assorbito il termine Informatik [1] preferendo il più imperfetto "computer science" ma… scopro ora (che ignorante) che gli inglesi utilizzano anche il termine "informatics": OK, affare fatto "Informatics Agency" è *perfetto*. [...] > Credo che la traduzione più appropriata di "agency" sarebbe "abilità". No, rimando alla risposta di questa mattina di Marco Fioretti [...] Grazie! Giovanni. [1] https://it.wikipedia.org/wiki/Informatica#cite_note-1 -- Giovanni Biscuolo

Aug. 28, 2020

SW Opex ? - Japan’s Creaking Computer Systems Are Hampering Its Economic Recovery

by gabriele elia

Interessante che in Giappone le spese per SW sono … “opex” non “investimenti” e questo è tra i fattori che rallentano il processo di digitalizzazione … tralasciamo dove sta l'italia nell'indice dell'IMD (hint: dietro il Kazakistan ...) https://www.datacenterknowledge.com/business/japan-s-creaking-computer-syst… Japan’s Creaking Computer Systems Are Hampering Its Economic Recovery “Japan is behind the world by at least 20 years” when it comes to administrative technology. Bloomberg<https://www.datacenterknowledge.com/author/--14> | Aug 24, 2020 Yuko Takeo (Bloomberg) -- The coronavirus crisis may have finally forced Japan’s government to get serious about fixing its computer systems. Before the pandemic, the problems might have been merely annoying, but now they’re impeding an economic recovery. Japan’s aging administrative pipeline has delayed aid money getting to where it’s needed during the pandemic. To fix that, the government last month included a “digital new deal” in its annual policy goals, calling for a year of concentrated investment to boost data-sharing between ministries and update websites, although no price tag has been given yet. conomists say the shoddy state of the government’s digital services has damped the benefits from fiscal stimulus and is hurting the country’s broader competitiveness by holding back technological progress in the private sector. “Japan is behind the world by at least 20 years” when it comes to administrative technology, said economist Yukio Noguchi, author of several books about tech and Japan’s economy. “The country needs to escape its system of using paper and physical stamps as soon as it can.” Japan is the world’s third-largest economy, but when it comes to digital competitiveness it ranks just 23rd out of 63 nations surveyed by the Institute for Management Development. While the government wants to make electronic signatures easier to use on its systems, physical stamps are still required for many official documents, making many transactions hard to do online. For some people, it’s also made working from home difficult during the pandemic. The bureaucracy’s failure to spend more on computer software is partly a result of the way it finances investments, according to economist Takuya Hoshino at Dai-ichi Life Research Institute. Software is classed as a “non-material” investment in government accounting. That means it’s funded by debt-covering bonds, which require a more arduous approval process than do the construction bonds used to build roads or bridges, Hoshino said. It’s easy to imagine how that technicality might discourage officials from making expensive proposals, given that Japan already has the developed world’s heaviest public debt burden. Still, the coronavirus has made the cost of having old computer systems much clearer. “It’s entirely possible we’ll have crises like this every decade going forward,” Hoshino said. “But without supporting infrastructure, the government can’t make the most of its own policies.” #end

Aug. 28, 2020

Re: [nexa] L’Europa ritrovi la sovranità tecnologica

by M. Fioretti

On Thu, Aug 27, 2020 20:13:19 PM +0200, Giacomo Tesio wrote: > Credo che la traduzione più appropriata di "agency" sarebbe "abilità". Mi pare proprio di no. Agency in questo contesto e': "the capacity, condition, or state of acting or of exerting power" https://www.merriam-webster.com/dictionary/agency e tutta questa discussione andrebbe letta/rifatta, se non mi sfugge qualcosa a monte, con QUESTO significato in mente. Ovvero: qui si parla soprattutto della **possibilita'** di esercitare le proprie competenze informatiche, non di averle o acquisirle. Uno Stallman in un mondo in cui per legge e' utilizzabile solo windows o iOS avrebbe la stessa abilita' informatica, ma zero "computing agency". Mentre uno che non capisce la differenza fra protocolli e formati, ma puo' legiferare in materia perche' e' magari senatore, ne ha molta di piu'. Marco -- M. Fioretti http://mfioretti.com http://stop.zona-m.net Your own civil rights and the quality of your life heavily depend on how software is used *around* you

Aug. 28, 2020

Tech giants push back on forced Uyghur labour claims

by don Luca Peyron

A proposito di ingiustizia. Buona lettura dl https://www.zdnet.com/article/tech-giants-push-back-on-forced-uyghur-labour… _________________________ don Luca Peyron Direttore Ufficio per la Pastorale Universitaria Arcidiocesi di Torino www.universitari.to.it via XX settembre 83, Torino tel. 011 5156239

Aug. 27, 2020

Re: [nexa] L’Europa ritrovi la sovranità tecnologica

by Giacomo Tesio

On Thu, 27 Aug 2020 12:42:16 +0200 Giovanni Biscuolo <giovanni(a)biscuolo.net> wrote: > > Propongo "autodeterminazione". > > Cibernetica, non solo tecnologica. > > "Autodeterminazione informatica"? > > In realtà l'etichetta che più mi piace è quella di "Computing Agency" > di un ente (individuo, gruppo, stato...) ma non sono capace di > tradurla efficacemente in italiano Il primo problema di "computing agency", "computing literacy", "computational thinking" e similari, è che per l'informatica i computer sono assolutamente secondari. L'informatica concerne le informazioni, mentre i computer (anche quelli quantistici :-D) al più possono gestire certi tipi di rappresentazioni delle informazioni, che noi chiamiamo "dati". Dunque ridurre l'informatica alla capacità di programmare computer, ridurre l'informazione al dato, è come ridurre l'astronomia alle tecniche che permettono l'uso ottimale dei telescopi. Credo che la traduzione più appropriata di "agency" sarebbe "abilità". "Abilità informatica" potrebbe essere abbastanza comprensibile, anzitutto perché chiarisce bene come non disporne determini una vera e propria disabilità. Una disabilità non fisica, ma cibernetica. Eppura una disabilità estremamente invalidante. Fortunatamente, si tratta di una disabilità completamente curabile. Come spesso accade però, la prima (e spesso maggiore) difficoltà sta nell'accettare la diagnosi. La terza, è seguire la terapia. ;-) Giacomo

Aug. 27, 2020

The Age of Cyber Sovereignty? - War on the Rocks

by Alberto Cammozzo

<https://warontherocks.com/2020/08/the-age-of-cyber-sovereignty/> Elliott Zaagman The White House’s approach to managing the potential security threats posed by TikTok, WeChat, and other Chinese-owned apps is hardly a model of procedural justice. Without a clear legal or regulatory framework, the Trump administration has issued executive orders banning transactions with the apps’ respective parent companies (Bytedance and Tencent). China’s leaders have not minced words in expressing their displeasure with U.S. actions, referring to those against TikTok as a “smash and grab” in a China Daily editorial, vowing retaliation, and even going so far as to suggest that Beijing could block a future sale. Yet despite its displeasure, Beijing seems to be witnessing the global adoption of a norm that it has long advocated. Ironically, it may do China more harm than good. For years, China has promoted the concept of “cyber sovereignty.” Although somewhat nebulous in definition, the term has been used to legitimize censorship, surveillance, and localized control of data that make up what is often referred to as China’s “Great Firewall.” Broadly put, it is the notion that the government of a sovereign nation should have the right to exercise control over the internet within its own borders, including political, economic, cultural, and technological activities. Beijing’s advocacy of the norm of cyber sovereignty has grown only more pronounced over time. Its annual World Internet Conference, which has drawn attendees such as Apple’s Tim Cook and Google’s Sundar Pichai as recently as 2017, has consistently offered a venue through which Chinese officials could lobby the movers and shakers of the tech world to see the value in this approach. China has also sought to advance this notion through international bodies such as the United Nations. For the Chinese Communist Party, cyber sovereignty has proved beneficial in a number of ways. While enabling the party to both moderate and gauge public discourse and sentiment, the technological barriers also provided space for Chinese companies to develop into some of the world’s largest, while impeding the entry and success of foreign internet platforms in China. China’s model has worked out so well that other countries have begun adopting elements of it. And it is now Chinese companies that seem to be suffering most. Spurred by a recent escalation of border tensions, India has been purging Chinese apps from the country’s internet. In addition to its pressure on TikTok, the Trump administration has recently announced the Clean Network program, a comprehensive tech stack that, if fully implemented, would almost entirely exclude Chinese technology firms. The adoption of cyber sovereignty as a global norm could now prove to be one of the greatest impediments to China’s peace, prosperity, and development. As China has few allies and worsening relations with the majority of the world’s most prosperous nations, its firms are finding a shrinking list of attractive overseas markets in which they are welcome. What’s more, China’s robust application of the cyber sovereignty principle over the past decade and a half has left it with few options through which to take reciprocal action when its firms are excluded from other nations’ digital spheres. After all, Beijing cannot go about banning apps that it already banned years ago. This reality is already being acknowledged by leaders in China’s technology industry. In an article on Sina.com that was later deleted, James Liang, co-founder of the online travel agency Ctrip, advocated that his country adopt a more open approach to its domestic internet in order to counter what he views as U.S. diplomatic aggression. “If we simply adopt a tit-for-tat strategy and implement the same xenophobic barriers, we will leave matters in the hands of the U.S.,” Liang wrote, adding, “our countermeasures should be to open up further … The United States wants to block WeChat and TikTok, so we can do the opposite.” Whether or not Liang’s prescription is correct, such an opening up would be a dramatic reversal of the direction that China and its digital sphere have been heading in, and starkly clashes with what appears to be President Xi Jinping’s vision for the Chinese internet. What does seem clear is that after years of lobbying the world to accept the notion of cyber sovereignty, China has gotten its wish. To which the axiom applies: Be careful what you wish for — you just might get it.

Aug. 27, 2020

Re: [nexa] What the A-level debacle teaches us about algorithms and government

by Antonio Vetro'

Sono molto concorde nell’osservazione delle reazioni eccessive, in un senso o nell’altro, davanti ai casi di automazione di processi tramite software, che per molti motivi sono sempre più numerosi (bellissimi gli esempi sul calcio). Forse segno che il fenomeno in atto è ancora poco compreso/comprensibile. Aggiungo che, nel caso in questione, da quanto mi sembra di capire dalle letture fatte finora, e oltre alle “lezioni” del primo articolo segnalato, sembrerebbe che l’agenzia Ofquam abbia: - ignorato le preoccupazioni della Royal Statistical Society inviate già a inizio Aprile ->Si veda ad es. https://www.civilserviceworld.com/professions/article/ofqual-snubbed-offer-… <https://www.civilserviceworld.com/professions/article/ofqual-snubbed-offer-…> - commesso errori banali nella validazione del modello (es training set e test set corrispondenti) e nella controversa gestione di dati di qualità non uniforme -> Si veda ad es. https://thaines.com/post/alevels2020 <https://thaines.com/post/alevels2020> e la visualizzazione interattiva https://thaines.com/content/ALevels/vis/ <https://thaines.com/content/ALevels/vis/> - non abbia coinvolto nello sviluppo i principali stakeholder, pratica consigliata in situazioni complesse come questa Oltre all’imperizia dimostrata, questo caso secondo me è un buon esempio di un altro fenomeno che osservo a livello globale riguardo alla parola “algoritmo" : il “riduzionismo tecnologico” di molti decisori politici che credono che la risposta a problemi sociali o socio-tecnici possa essere puramente tecnica. Riduzionismo che se abbinato a una visione del software come “commodity” purtroppo ancora diffusa in ambiti decisionali, non solo non risolve i problemi iniziali, ma ne aggiunge altri, come l’articolo del Guardian linkato racconta benissimo. Un cordiale saluto, antonio > Il giorno 27 ago 2020, alle ore 12:45, Paolo Atzeni <paolo.atzeni(a)inf.uniroma3.it> ha scritto: > > > >> Il giorno 26 ago 2020, alle ore 15:49, Antonio Vetro' <antonio.vetro(a)polito.it <mailto:antonio.vetro@polito.it>> ha scritto: >> >> Alcune lezioni apprese dal recente pasticciaccio dell’algoritmo usato nelle scuole inglesi: >> https://www.prospectmagazine.co.uk/science-and-technology/a-level-students-… <https://www.prospectmagazine.co.uk/science-and-technology/a-level-students-…> >> >> Inoltre, per gli amanti dei numeri, qui un’ottima ricostruzione di come ha lavorato l’algoritmo: https://rpubs.com/JeniT/ofqual-algorithm <https://rpubs.com/JeniT/ofqual-algorithm> >> E qui invece la versione sintetica e in “prosa” del Guardian: >> https://www.theguardian.com/education/2020/aug/21/ofqual-exams-algorithm-wh… <https://www.theguardian.com/education/2020/aug/21/ofqual-exams-algorithm-wh…> >> > > > Le lezioni sono ben espresse e penso che la maggior parte di noi le condivida in pieno: > - trasparenza nell’uso degli algoritmi (o forse userei la parola criteri) > - analisi accurata e validazione indipendente degli algoritmi > - disponibilità di procedure di appello per via giudiziaria > - possibilità di revisione di sistema (e non solo giudiziaria) > > MI sembra però importante aggiungere qualcosa. > > L’approccio usato dall’agenzia UK è stato sicuramente sbagliato, ma ha cercato di affrontare un problema reale, > che resta non risolto. Invito a leggere l’articolo citato nel primo capoverso: > https://www.theguardian.com/education/2020/aug/17/u-turn-exams-may-create-n… <https://www.theguardian.com/education/2020/aug/17/u-turn-exams-may-create-n…> > Il punto si può riassumere notando che un sistema come quello britannico in cui ci sono tante università e > vede una grande richiesta per l’accesso alle più prestigiose richiede elementi su cui basare la scelta > (non sto qui né approvando né criticando tale sistema, semplicemente osservo) > A questo punto, dopo la nuova decisione, la scelta si baserà sui voti attribuiti dalle singole scuole. > L’approccio basato sull’ "algoritmo” ha fallito, ma cercava di limitare le conseguenze dei fenomeni > di inflazione dei voti che si possono facilmente manifestare se il voto ha un valore assoluto > e non verificabile. Ad esempio, la lettura di molti articoli sul tema degli A-level evidenziava il fatto che alcuni > studenti sono stati penalizzati dall’abbassamento dei voti e altri forse ingiustamente premiati (soprattutto > se provenienti dalle scuole di elite). E questo è probabilmente vero, ma quanti invece avevano voti alti > più o meno regalati da scuole private? — la taratura serviva probabilmente ad affrontare questo problema. > (In modo analogo, quanti di noi sono stati in commissioni di concorso e hanno dovuto attribuire punteggi > sulla base di votazioni di laurea certamente non paragonabili?) > > Perché scrivo tutto questo: > sui giornali leggiamo spesso, insieme a contributi che mitizzano la tecnologia, anche articoli che la demonizzano. > Soprattutto la parola “algoritmo” viene usata con timore, se non peggio, come fosse il male assoluto. > Noi informatici dovremmo trovare il modo di comunicare che i problemi esistono e vanno affrontati, ovviamente bene > (quindi evitando le scatole nere e gli approcci miracolistici), ma spesso sono difficili e non possono essere > semplicemente rimossi. > Concludo con un paragone leggero (se lo vedesse, un collega olandese con cui ho collaborato nel Nucleo > di Valutazione mi direbbe ancora una volta che noi italiani facciamo solo esempi sul calcio): > Durante il lockdown, si parlò della possibilità di assegnare lo scudetto e definire le retrocessioni > sulla base di un algoritmo: > https://www.lastampa.it/sport/calcio/2020/05/31/news/calcio-un-algoritmo-pe… <https://www.lastampa.it/sport/calcio/2020/05/31/news/calcio-un-algoritmo-pe…> > e i commenti al riguardo furono di ogni tipo. > Poi si arrivò alla conclusione che in caso di non completamento del campionato lo scudetto non sarebbe stato assegnato, > ma per le retrocessioni e le promozioni si decise che sarebbe stato adottato un algoritmo. > https://www.goal.com/it/notizie/nuovo-regolamento-serie-a-assegnazione-scud… <https://www.goal.com/it/notizie/nuovo-regolamento-serie-a-assegnazione-scud…> > Perché? Le uniche alternative sarebbero state: (a) utilizzare la classifica al momento dell’interruzione (ma con numero diverso di partite > giocate, in totale o in casa e fuori); (b) congelare la composizione delle categoorie (serie A, serie B, …) > In fondo la classifica si basa su un criterio semplice (punti) ed è considerata equa perché tutti hanno le stesse possibilità. > Quando ciò non è vero, si deve trovare una soluzione (semplice o complessa), ma è necessario assumersene la responsabilità. > > Saluti, > > -Paolo Atzeni > _______________________________________________ > nexa mailing list > nexa(a)server-nexa.polito.it <mailto:nexa@server-nexa.polito.it> > https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa <https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa>

Aug. 27, 2020

Re: [nexa] AI Watch - Artificial Intelligence in public services | EU Science Hub

by Antonio Vetro'

Segnalo e riporto il breve commento al documento da parte di Algorithm Watch all’interno della sua ultima newsletter del 14.8: "The European Commission published a report on “Artificial Intelligence in Public Services” in early July. The authors analyzed 230 instances of artificial intelligence being used in the public services of EU member states, the United Kingdom, Switzerland and Norway. Eight of these examples are described in section 3.2 (p.40), which unfortunately does not mention any sources, but the other 222 are not listed. As is very often the case with documents published by the Commission, AI is considered good in itself. This leads the authors to argue that, while no evidence exists that deployments of AI achieve their intended results (p.17), AI should be deployed faster in public services (p.85).” antonio > Il giorno 5 lug 2020, alle ore 18:30, J.C. DE MARTIN <demartin(a)polito.it> ha scritto: > > > https://ec.europa.eu/jrc/en/publication/eur-scientific-and-technical-resear… > > > (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

Aug. 27, 2020

Web Browser Privacy: What Do Browsers Say, When They Phone Home?

by Alberto Cammozzo

<https://www.scss.tcd.ie/Doug.Leith/pubs/browser_privacy.pdf> Abstract We measure the connections to backend servers made by six browsers: Google Chrome, Mozilla Firefox, Apple Safari, Brave Browser, Microsoft Edge and Yandex Browser, during normal web browsing. Our aim is to assess the privacy risks associated with this back-end data exchange. We find that the browsers split into three distinct groups from this privacy perspective. In the first (most private) group lies Brave, in the second Chrome, Firefox and Safari and in the third (least private) group lie Edge and Yandex. [...] VIII. CONCLUSIONS We study six browsers: Google Chrome, Mozilla Firefox, Apple Safari, Brave Browser, Microsoft Edge and Yandex Browser. For Brave with its default settings we did not find any use of identifiers allowing tracking of IP address over time, and no sharing of the details of web pages visited with backend servers. Chrome, Firefox and Safari all share details of web pages visited with backend servers. For all three this happens via the search autocomplete feature, which sends web addresses to backend servers in realtime as they are typed. In Chrome a persistent identifier is sent alongside these web addresses, allowing them to be linked together. In addition, Firefox includes identifiers in its telemetry transmissions that can potentially be used to link these over time. Telemetry can be disabled, but again is silently enabled by default. Firefox also maintains an open websocket for push notifications that is linked to a unique identifier and so potentially can also be used for tracking and which cannot be easily disabled. Safari defaults to a choice of start page that potentially leaks information to multiple third parties and allows them to preload pages containing identifiers to the browser cache. Safari otherwise made no extraneous network connections and transmitted no persistent identifiers, but allied iCloud processes did make connections containing identifiers. >From a privacy perspective Microsoft Edge and Yandex are qualitatively different from the other browsers studied. Both send persistent identifiers than can be used to link requests (and associated IP address/location) to back end servers. Edge also sends the hardware UUID of the device to Microsoft and Yandex similarly transmits a hashed hardware identifier to back end servers. As far as we can tell this behaviour cannot be disabled by users. In addition to the search autocomplete functionality that shares details of web pages visited, both transmit web page information to servers that appear unrelated to search autocomplete.

Aug. 27, 2020

Law, code and systems – Public Strategist

by Alberto Cammozzo

<https://www.publicstrategist.com/2017/11/law-code-and-systems/> Should law which results in the creation of new public administrative systems receive different parliamentary scrutiny from law which doesn’t? That’s the question raised in a recent blog post by Gordon Guthrie which caught my eye for a number of reasons, not least a typically thought provoking recent conversation with John Sheridan. As so often, apparently simple questions quickly get complicated when you dig into them a bit, and the first complication is to work out what question this special scrutiny is intended to answer. As a first approximation, there are three very different dimensions to this: How law is translated into code How law is translated into systems What role legislative scrutiny has in improving the way in which the first two are done. Law and code On the surface, law doesn’t look much like code. Despite heroic efforts by parliamentary counsel in recent years, it mostly looks like arcane wordplay. Looking just a little below the surface, though, patterns start to emerge which look intriguingly like code. Data elements are defined. Conditions are identified and tested for. The test leads execution down distinctive paths. A single text is unlikely to be self-contained but will refer to and draw on other legislative elements – calling functions if you will. But there is a deeper level still where the analogy breaks down again. The compiler or interpreter is not immediately deterministic – it can take years and many levels of the judicial system to work out the correct output from given inputs. Particularly, but not only, in a common law system, many functions are called by interpretation without it ever being explicit that they are being called at all. The idea that law can be some kind of pseudo code is attractive but insubstantial.1 That’s not to say that there is no connection at all. As one example, Richard Pope suggested a few years ago that there might be scope for using software testing techniques to provide some insight into whether regulations are being applied. More generally, there are many contexts where rules defined in law need to become rules applied in practice, and rules defined in code will almost always be a vital intermediate step. Law and systems Law may be a necessary component of an administrative system, but it is never a sufficient one. Data needs to be collected, cases need to be managed, help needs to be offered, processes and roles need to be understood, edge cases need to be identified, transitional arrangements need to be put in place – and the list goes on. In the implementation of a legal change, the capturing of the legal element itself is likely to be one of the smaller and more straightforward aspects of what needs to be done. That doesn’t mean that the project within which such a change is embedded will be small or straightforward – and government is littered with examples of projects which are anything but. It does though suggest that the extent of legislative change may not be the best indicator and certainly isn’t the only indicator of the level of risk in developing or redeveloping a system. Legislative scrutiny Law is the responsibility of legislators. They are both the primary decision makers about what the law should be and the primary means by which the executive is held to account for its implementation and application.2 Legislative debate is normally strongly focused on the merits of ideas and policies; there is rarely much attention given to the practical questions of implementation, partly because that is rarely specified within the legislation, partly because there is limited scope for preliminary work to have been considered during its passage,3 and partly because that doesn’t tend to be what legislators focus on. The result is that, as things currently work, legislative approval almost necessarily precedes any point at which it is possible to assess the likelihood of successful delivery. It’s certainly possible to see a model in which legislatures took a more active and more structured role in monitoring whether what was being developed had the characteristics of success, but however that were to work, it wouldn’t be a fundamentally legislative role. So perhaps the question is less whether legislatures should do more of this, and more whether there is a shortfall in the external challenge and oversight of programme management, and if so whether that shortfall is best addressed by legislators.4 Back to code I remain strongly of the view that thinking about law as code (or perhaps better, law as system) prompts interesting ideas and insights. At a more mundane level, it can prompt attention to language being used not just not in line with its natural meaning, but in flat contradiction to it. But that’s not because law is code; it’s because the parallels are strong enough and the differences clear enough that the cross-fertilisation of ideas is useful. The greater engagement of legislators with that debate might have some very interesting consequences.

Aug. 27, 2020