AI as an excuse for obfuscating responsibility for commercial decisions
Much of the current hype over Big Data and Artificial intelligence results from the large scale use of use of old techniques to hoover up personal and other data by those who wish to avoid scrutiny of the way they collate and analyse it to influence/exploit their audiences. Sometimes this is to conceal sonking (the scientification of non-knowledge, charging extra for stating the obvious or giving false credibility to guesstimates). Sometimes they have covert as well as overt objectives. Meanwhile most of the examples of public benefit (from traffic analysis to epidemiology ) in the recent NIC report on Data for the Public Good do not require new research or technology. Many are already operational, albeit in small scale across UK local authorities, from Plymouth through Bournemouth and Milton Keynes to Newcastle, not just the test beds of Bristol and Manchester or other parts of the world. The “research” most needed is into processes for the effective the governance of data quality and of secure sharing in a world of increasingly complicated intellectual property rights and of the “correct” use of analytical tools for interpretation and exploitation in the interests of all who contribute or are affected.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
The buck stops where?
The subtitle of “Move Fast and Break things” is “How Facebook, Google and Amazon cornered culture and undermined democracy”. The comments attached to the Guardian review indicate a progressive establishment caught in the headlights of the Internet cartel’s dominance over the inter-linked worlds of advertising, information, innovation, lobbying and funding. Meanwhile the US election of Trump and the UK Brexit vote indicate that small town, working class America and high street, lower middle-class England have lost faith in the “establishment” (Washington, Whitehall or Brussels). So too have the students, saddled with debt in the UK or unemployed across most of the EU, as the on-line world threatens to render their expensively (time as well as money) acquired knowledge redundant.
Amazon, Apple, Facebook, Google and Twitter have begun to appreciate that the rest of the world no longer believes they represent a brave new world, outside the traditional norms of society. The mass of voters are discovering that the Internet is not only a critical utility for their “off line” world. It is just as violent , corrupt and abusive . The tech giants are, in consequence, under mounting pressure to help make the Internet safe for the vulnerable, from children to grandparents. It is no longer sufficient to “protect” a playground for misogynist West Coast and other Metrosexual libertines (or did I mean libertarians?). Some players, like Uber, appeared, until recently, immune from the pressures coming via the advertisers, fund managers and venture capitalists who finance most current Internet businesses and the spin-offs and start-ups they support. But the backlash following appearance of Jihadist videos alongside ads for luxury SUVs finally spurred Google, Facebook and Twitter into action to help protect their programmatic advertising revenues.
The dominant players are now under pressure to help clean up the Internet
Part of their response has been the promotion of automated tools to detect those promoting fake news and generating automated click bait to harvest pay-per-view advertising revenues. The progress to date, however, has not impressed the Chairman of the UK Culture Media and Sport Select Committee, himself a former advertising executive . He has called for the kind of detail that his former colleagues require, if they are to continue to support an advertising funded Internet in the face of the damage being done to major brands by, for example, the parental backlash against the exposure of their children to all forms of undesirable content, from jihadist grooming, through child abuse to unhealthy eating.
Meanwhile many of the elderly, who control most of the nation disposable wealth, are resisting attempts to get them to use on-line banking, or even to upgrade their existing systems, because they fear that any contact from a “computer expert” is fraudulent. It is said that the information is readily available to electronically impersonate of over half the over 65s in the UK. One consequence is the promotion by banks of services like Trusteer to help protect themselves from rogue customer transactions, at the risk of inconvenience to those who try to keep their browsers up to data with the latest releases . But the war of algorithms goes much wider than battles between browsers and security software.
How will the Internet Community respond?
On one side we have the algorithms mixing data from multiple sources to create, sustain and exploit armies of false identities running over botnets of hijacked devices. Those producing such algorithms include state security and cyberwarfare operations (east and west) as well as organised crime (often better funded than law enforcement).
On the other side are the “solutions” promoted by “big tech” companies to reassure their advertisers and regulators. These are usually variations on automated “censorship”. Commercial confidentiality and security arguments are used to defend lack of public accountability and governance for the techniques used or people involved – because these are often carefully tailored to avoid impacting current revenue streams or admitting the liabilities that would arise from loss of “innocent carrier” status.
The Internet Society and the Internet Engineering Task Force develop, test and implement standards on which a critical mass of the industry can agree – but we need to appreciate why that has not been, and will not be, sufficient.
ISOC was founded in 1992 to provide a parent for the standards body, the IETF. Together they form the collective of engineers who keep the Internet running. I was persuaded to join in 1995, when IBM, CISCO, EDS, AT&T et al had agreed a $4 billion war chest to “re-engineer” the Internet for mass rollout if it passed its first “commercial” challenge, the Atlanta Olympics, without serious problems. I was told that the three biggest challenges it would then face were – security, security and security. I was also told that it would have to reach out beyond the education and scientific community for its governance to be credible as the Internet achieved its potential.
It has tried.
It has not yet succeeded. The “rest of the world has”, until very recently, been content to leave it to the “experts”.
I am among those to blame. As a sometime committee member of ISOC UK, I did not try hard enough to make it reach out beyond its core community – the 3,500 or so individuals who keep the Internet running in the UK. I then gave up when I got no response to my efforts to tell others why and how they could and should get involved. It is now over five years since I last blogged on the need for business users to join Nominet, ICANN, IETF and ISOC to help protect their interests “against” those with very different priorities. I said then that the effort involved in active participation was such that those who were serious would need to share the load – via user groups, their own trade association or professional body and/or groups like the Digital Policy Alliance. I still stand by that view.
The next opportunity , the UK Network Operators Forum is a get together of the network architects who are creating the inter-operable broadband architectures we take for granted. If you look at the agenda you will see why most business users think this is not for them. Several DPA supplier members will be well represented. But unless you are helping the Communications Management Association (now part of the BCS ) represent your interests, no-one will be speaking for the user community.
Your opportunity to make a difference?
March 2018 sees the biggest opportunity to input UK user views since the IETF last met in London . Neither Government nor any of the major UK professional bodies or trade associations bothered to greet, let alone meet, them in 2014 – although with the help of Digital Policy Alliance members we were able to organise a policy briefing in the House of Commons and invitations for MPs and officials to attend a reception afterwards.
From March 17th – 21st the IETF has its 101st Meeting in London, hosted by Google and ICANN. The agenda has not yet been fixed. IETF is an engineering democracy and they are calling for proposals. The IETF has been long aware of the way in which technology dissidents (for example girls who dare to play with boys technology toys) can be hounded out of debate by idealogues, trolls and vested interests. Their policy for handling harassment dates back to 2013 and the processes were revised in March 2016, over a year before the problems with regard to sexual harassment in Silicon Valley finally hit the US press. An IETF event therefore provides an opportunity for well-informed technical discussion about the practical means of addressing the issues of organising balanced, democratic, debate on- and off-line, on difficult issues. But some-one has to put the issues on the agenda and engage the relevant communities.
Engineers versus regulators, politicians, marketers, lawyers and accountants
Politicians are discovering the need to regulate the algorithms, (including those labelled artificial intelligence), used to censor fake news and provide effective data protection. But the latter are neither new nor separate issues. Both are symptomatic of the failure of ISOC to provide the standards of governance and leadership to which its founders aspired. ISOC failed because it is composed, like the IETF, of human beings. Engineers can be as honest, corrupt, candid, hypocritical, clear sighted and myopic as the rest of mankind. in consequence, the billionaire high priests of the Internet have been able to use a mix of technical and legal obfuscation to hide the secrets of their magic (from magnetic control over IPR to enabling taxable revenues to vanish) from the rest of us. But they may now be as vulnerable as the Knights Templar after they were driven out of their “special place”, the Holy Land. The Internet is now a critical infrastructure – underpinning the whole of society. And with that comes a whole new tier of responsibility and accountability.
The Internet, like the medieval Knights Templar, has simple principles for enabling communications across cultures and continents. But it still lacks the equivalent of the trusted identity, authentication and authorisation rituals of the Templars – which they were supposed to keep secret under torture. Those rituals (updated for carriage over telegraph, telephone, fax and data transfer) still underpin correspondence banking today. Meanwhile the rest of the Internet has overlays of complexity which enable pseudonymisation, impersonation and avoidance of intermediary responsibility for error or failure: the “best efforts” approach commonly enables denial of responsibility for probity, reliability, resilience and security unless contracted for specific services.
I would trust the combination of ISOC and IETF more than lawyers, regulators or politicians, to address those failings and to produce, test and implement processes that meet the needs of most of mankind to know with whom they are dealing. I would also trust them to come up with robust processes for inter-operability between commonly used identity and trust architectures. I am less certain, however, whether we can trust the dominant Internet players, let alone Governments, to adopt such processes – unless and until forced to by market pressures. The reasons for that doubt got back many years . But engineers are not good at designing governance processes and the heart of the problem is governance processes that will work across jurisdictional boundaries. For than one has to look to the City of London and bodies like CEDR or Lloyds .
The collapse of faith in democracy, off-line and well as on-line
The world is now awash with fake identities while the information needed to electronically impersonate almost all those worth impersonating can be acquired more easily than most of us can open a new UK bank account. One consequence is that any attempt at on-line democracy risk being swamped by armies of imaginary voters – not just a few catfish creating false identities for personal reasons but million strong groups created to support click fraud with automatically created personalities, mashed up from genuine personal data and evolving over time to bypass bot detection algorithms as the latter also evolve.
Meanwhile the combination of on-line voter registration and postal voting has spread the problem of stolen elections to the off-line world. In the UK, 38% of voter registrations made during the referendum campaign in 2016 are now known to have been duplicates. The Electoral Commission has yet to collate data on the 2017 General Election but preliminary estimates for individual districts range from 30% – 70% with 4.9 million registrations (nearly a million on the last day) giving a net 1.4 million duplicate registrations. The Commission is looking at ways of reducing accidental duplication but is not attempting to estimate how many not only registered from more than one address but voted more than once in the same election. An exercise to do this would be relatively easy, using 1970s technology to compare collated electoral registers with the files of Experian, Call Credit, Lexis Nexis to identify multiple registrations and then check for multiple voting. If the problem is indeed significant (e.g. 20 – 30 seats “stolen” in the 2017 General Election) the means of addressing 90% of the problem may be relatively simple: instituting cut-off dates a fortnight before polling day for both electoral registration and postal voting. This would enable a national exercise to identify duplicates in advance of polling day, with postal votes invalidated and automated court proceedings to each address. This process would not be fool proof but would help restore faith that elections can only be “stolen” by politicians lying to the voters and/or Russian, Macedonian or other fake news stories. The main obstacle appears to be the fear of admitting that the UK still has an election system that would disgrace a banana republic .
One of the joys of social media over the last couple of days has been to watch as Remoaners accuse Brexiteers of “stealing” the referendum, only for evidence to emerge that their own behaviour was even more egregious. But the wipe-out of sources of local news and views as national and international on-line news and advertising media wiped out local journalism, was a heavy price to pay for such joys. Before Christmas I spent three months trying to make sense of the policies of our local authority, how they were actually being implemented and what local residents thought about them. That which used to be covered in the local South London press is now covered nowhere. I also spent time trying to find a local jeweller to repair a piece that had no value, beyond sentimental. None of the main search engines led me to the one (more an old-fashioned, independent junk repair shop) that I found via an enquiry on Facebook local.
Current Internet business models, not just the war of the algorithms, appear to be in the process of wiping out well-informed democracy in parallel with the local high street and out-of-town shopping mall.