Rohan Samarajiva


We’ve been thinking about the implications of differential access to data for a while. Here‘s a detailed discussion: Some scholars said Facebook’s recent privacy changes may have gone too far by also cutting off academics who behaved responsibly. “Academics would argue that we need access to primary data,” said Dr. Nielsen of Oxford. He said the changes might lead to an asymmetry, with internal Facebook researchers accumulating mounds of data while outside academics would not.
It is reported that momentum is building for rules for e commerce under the WTO. India is both a hotbed of e commerce developments (Walmart has just gained majority control of Flipkart) and a heavyweight in international trade negotiations. Its role in scuttling the Doha Round is still remembered. But India is said to lack adequate knowledge to formulate positions on e commerce, except for one blatantly protectionist issue and one that poses significant challenges to implement: Consider, for instance, one key demand by developed countries to make permanent the current ban on customs duties on ‘global electronic transactions’ that were suspended in 1998. On the face of it, this is a reasonable ask: if the ban is overturned, it would give countries the right to impose tariffs on downloads of mobile applications, streamed music from Spotify or videos from Netflix.
It is one thing to talk about the second circle of associates in terms of lawful interception, as we did in our bulk surveillance paper that will be out of review shortly. It is quite something else to see how many records are actually yielded by surveillance of a relatively small number of targets. In 2017, 543 million records were collected from surveillance of just 40 people. In 2016, the first full year for which that replacement system was in operation, the government obtained orders to target 42 people and collected just over 151 million call detail records. In 2017, the government obtained orders for 40 targets.
Much of the discussion on privacy is premised on the implicit imposition of a private-property model on data or information that is subject to control/consent. This could have worked when all we were dealing with were relatively simple data like a social security number or an address. But the really interesting data are transaction-generated data (TGD). These necessarily involve more than one person. How can I give or not give consent to the use of my TGD, when multiple entities have been involved in its production?
It appears that Myanmar’s universal service strategy has been finalized. Our work was represented in the draft that was put out for comments, but we made additional comments on that draft, which are here. Current mobile networks cover over 90% of Myanmar’s population, but the government believes the USF will be necessary to fund the development of network towers in unserved areas. Through the project the government is targeting 94% population coverage by the first quarter of next year and 99% coverage in the future. Once basic infrastructure is deployed to the rural areas, more advanced telecommunications services can be introduced in the future, the report states.
It was back in 2009 that I was criticized for saying that regulation based on an understanding of the operative Budget Telecom Network Model did not permit strict regulation of quality of service. It was also around then that we started focusing on taxes as a factor. That was repeated many times in Bangladesh by us, for example, when I outlined what needed to be done to achieve Digital Bangladesh. That these things have to be said again and again, indicates we have not been very effective. The national exchequer gets Tk 51 of every Tk 100 spent by a mobile user, leaving less money for the operators to develop network and run business smoothly, the GSMA said yesterday.
Gyanendra’s Law states that network shutdowns after a certain level of penetration has been achieved results in the political authority responsible for the shut down losing power. The underlying reason may be the economic harm that is caused. Appears this effect is seen is shut downs of Internet-based services too: Telegram tried to thwart the blockage by shifting its service to two giant American web hosts, Google Cloud and Amazon Web Services, while at the same time repeatedly changing its IP address to skip ahead of Roskomnadzor. In response, rather than chasing individual IP addresses — a unique set of numbers that identifies a computer, smartphone or other device connected to the internet — the watchdog agency elected to shut down enormous blocks of addresses, called subnets. The collateral damage hit a variety of other sites, like Viber, another messaging app, as well as small businesses including a language school and a courier service, all of which suffered financial losses.
I do not appear to have blogged about it, but the first time I brought up this issue was at the 2013 (Bali) or 2014 (Istanbul) IGF. If consent-based privacy rules are imposed on the existing concentrations of behavioral transaction-based data, there will be considerable to negative implications for SMEs and start ups on one side and for those engaged in data analytics for the public good. It is interesting that the Economist is highlighting this issue: Along with other tech firms, it should create an industry ombudsman whose jobs would include making access to platforms easier for independent researchers. Instead of opening up, however, the risk is that Facebook will throw up walls: its decision to kick third-party data-brokers off the platform has the convenient effect of both protecting users’ data and entrenching its power as a source of those data.
LIRNEasia and CPRsouth were created to contribute to better laws, policies, regulation and implementation in the emerging Asia Pacific. There are many ways to do this, including the actual training of the relevant people within government. For example, from 2013 onward we have conducted multiple training and awareness programs for legislators and regulatory staff in Myanmar. To say that Myanmar legislators need help with understanding new technologies and new business models is one thing. To say that members of the US Congress do is quite something else.
I probably learned more useful things from working as a lowly assistant for an expert witness in US v AT&T, than from my formal education. So I was all agog when the next big anti-trust case came up, US v Microsoft. But that was also when I began to realize the need rethink of the core concepts. This article in Medium (I hope it will not be paywalled) makes an insightful comparison between the tying arguments that were central the Microsoft case and the loose claims of monopoly being bandied about in relation to Facebook now. While some of the questions and concerns echoed those the senators had two decades ago, Microsoft and Facebook’s situation could not be more different.
Here is what Mark Zuckerberg said in his testimony before a Senate Committee: He also said that while Facebook is beefing up its use of AI to spot and remove offensive content, it will be five to 10 years before the company will have tools to do this automatically. So I was correct in my phrasing in the New York Times op ed: Media companies of all kinds must accept responsibility and deploy artificial intelligence and plain old elbow grease to the task. And people of good will must play their part by calling out falsehoods and reporting those responsible. So let’s apply the elbow grease while working on the AI.
Firms have always had an interest maintaining the loyalty of their customers. This has also involved knowing more about the customers. In a discussion of subscription models, the Economist, refers to what may happen because of restriction on data that may emerge because of the Cambridge Analytica imbroglio. Subscription models are becoming more popular, in part because technology has made it easier to rent rather than own assets. Instead of buying software, for example, users can get access to it as a cloud-based service.
Iran is considering a permanent ban on Telegram. To make it effective, they will probably have to go after VPNs too. Interestingly, a lot of the opposition seems to have an economic basis. Is this the basis of Gyanendra’s Law and its various exceptions? For the many small-business owners who use Telegram to market their services and communicate with clients, the ban adds to their financial woes.
Our paper on bulk surveillance is under review and will be public soon. We did not go deep into predictive policing because most of the extant material was US-centric. It is interesting that the Economist, which keeps putting out city-level murder data, has chosen to publish a piece on the use of data in controlling violence in Latin America’s cities: Rodrigo Guerrero, the city’s mayor and a surgeon by training, launched a plan inspired by the epidemiological approach some North American cities were taking at the time. He set up “violence observatories” where police, public-health officials, academics and concerned citizens could study crime data. This revealed that most of the city’s murders took place in drunken brawls, not in conflict between gangs, and that they were late at night a day or so after payday.
Occasionally, I accept invitations to speak on subjects I am still exploring in my mind. The talk I am going to give today at an event called Festival of our Future falls into that category. Should I apologize for not knowing the full and complete answer on how to fix politics? Not really. This is from a book review written by one of the world’s leading political philosophers: In recent decades, American public discourse has become hollow and shrill.
It was four years ago, in March 2014, that LIRNEasia organized an expert forum on broadband initiatives in New Delhi as part of our work supported by the Ford Foundation. The presentations included one on the design of Mexico’s unique experiment of building a single network using 700 MHz frequencies which was to be used by all operators. So it cannot be said that we were behind in spotting new developments in telecom. Four years later, the shared network is live. Here is the Economist’s account.