NOW AVAILABLE The draft of my book on Organizational Intelligence is now available on LeanPub Please support this development by subscribing and commenting. Thanks.

Monday, October 31, 2016

The Transparency of Algorithms

Algorithms have been getting a bad press lately, what with Cathy O'Neil's book and Zeynap Tufekci's TED talk. Now the German Chancellor, Angela Merkel, has weighed into the debate, calling for major Internet firms (Facebook, Google and others) to make their algorithms more transparent.

There are two main areas of political concern. The first (raised by Mrs Merkel) is the control of the news agenda. Politicians often worry about the role of the media in the political system when people only pick up the news that fits their own point of view, but this is hardly a new phenomenon. Even in the days before the Internet, few people used to read more than one newspaper, and most people would prefer to read the newspapers that confirm their own prejudices. Furthermore, there have been recent studies that show that even when you give different people exactly the same information, they will interpret it differently, in ways that reinforce their previous beliefs. So you can't blame the whole Filter Bubble thing on Facebook and Google.

But they undoubtedly contribute further to the distortion. People get a huge amount of information via Facebook, and Facebook systematically edits out the uncomfortable stuff. It aroused particular controversy recently when its algorithms decided to censor a classic news photograph from the Vietnam war.

Update: Further criticism from Tufekci and others immediately following the 2016 US Election

The second area of concern has to do with the use of algorithms to make critical decisions about people's lives. The EU regards this as (among other things) a data protection issue, and privacy activists are hoping for provisions within the new General Data Protection Regulation (GDPR) that will confer a "right to an explanation" upon data subjects. In other words, when people are sent to prison based on an algorithm, or denied a job or health insurance, it seems reasonable to allow them to know what criteria these algorithmic decisions were based on.

Reasonable but not necessarily easy. Many of these algorithms are not coded in the old-fashioned way, but developed using machine learning. So the data scientists and programmers responsible for creating the algorithm may not themselves know exactly what the criteria are. Machine learning is basically a form of inductive reasoning, using data about the past to predict the future. As Hume put it, this assumes that “instances of which we have had no experience resemble those of which we have had experience”.

In a Vanity Fair panel discussion entitled “What Are They Thinking? Man Meets Machine,” a young black woman tried unsuccessfully to explain the problem of induction and biased reasoning to Sebastian Thrun, formerly head of Google X.
At the end of the panel on artificial intelligence, a young black woman asked Thrun whether bias in machine learning “could perpetuate structural inequality at a velocity much greater than perhaps humans can.” She offered the example of criminal justice, where “you have a machine learning tool that can identify criminals, and criminals may disproportionately be black because of other issues that have nothing to do with the intrinsic nature of these people, so the machine learns that black people are criminals, and that’s not necessarily the outcome that I think we want.”
In his reply, Thrun made it sound like her concern was one about political correctness, not unconscious bias. “Statistically what the machines do pick up are patterns and sometimes we don’t like these patterns. Sometimes they’re not politically correct,” Thrun said. “When we apply machine learning methods sometimes the truth we learn really surprises us, to be honest, and I think it’s good to have a dialogue about this.”

In other words, Thrun assumed that whatever the machine spoke was Truth, and he wasn't willing to acknowledge the possibility that the machine might latch onto false patterns. Even if the algorithm is correct, it doesn't take away the need for transparency; but if there is the slightest possibility that the algorithm might be wrong, the need for transparency is all the greater. And evidence is that some of the algorithms are grossly wrong.

In this post, I've talked about two of the main concerns about algorithms - firstly the news agenda filter bubble, and secondly the critical decisions affecting individuals. In both cases, people are easily misled by the apparent objectivity of the algorithm, and are often willing to act as if the algorithm is somehow above human error and human criticism. Of course algorithms and machine learning are useful tools, but an illusion of infallibility is dangerous and ethically problematic.

Rory Cellan-Jones, Was it Facebook 'wot won it'? (BBC News, 10 November 2016)

Ethan Chiel, EU citizens might get a ‘right to explanation’ about the decisions algorithms make (5 July 2016)

Kate Connolly, Angela Merkel: internet search engines are 'distorting perception' (Guardian, 27 October 2016)

Bryce Goodman, Seth Flaxman, European Union regulations on algorithmic decision-making and a "right to explanation" (presented at 2016 ICML Workshop on Human Interpretability in Machine Learning (WHI 2016), New York, NY)

Mike Masnick, Activists Cheer On EU's 'Right To An Explanation' For Algorithmic Decisions, But How Will It Work When There's Nothing To Explain? (Tech Dirt, 8 July 2016)

Fabian Reinbold, Warum Merkel an die Algorithmen will (Spiegel, 26 October 2016)

Nitasha Tiku, At Vanity Fair’s Festival, Tech Can’t Stop Talking About Trump (BuzzFeed, 24 October 2016) HT @noahmccormack

Julia Carrie Wong, Mark Zuckerberg accused of abusing power after Facebook deletes 'napalm girl' post (Guardian, 9 September 2016)

New MIT technique reveals the basis for machine-learning systems’ hidden decisions (Kutzweil News, 31 October 2016) HT @jhagel

Video: When Man Meets Machine (Vanity Fair, 19 October 2016)

See Also
The Problem of Induction (Stanford Encyclopedia of Philosophy, Wikipedia)

Related Posts
The Shelf-Life of Algorithms (October 2016)
Weapons of Math Destruction (October 2016)

Updated 10 November 2016

Monday, October 24, 2016

Learning at the Speed of Learning

According to a recent survey by McKinsey,  "the great majority of our respondents expect corporate learning to change significantly within the next three years".

It seems that whatever the topic of the survey, middle managers and management consultants always expect significant change within the next three years, because this is what justifies their existence.

In this case, the topic is corporate learning, which McKinsey recommends should be done "at the speed of business", whatever that means. (I am not a fan of the "at the speed of" cliche.)

But what kind of change is McKinsey talking about here? The article concentrates on digital delivery of learning material - disseminating existing "best practice" knowledge to a broader base. It doesn't really say anything about organizational learning, let alone a more radical transformation of the nature of learning in organizations. I have long argued that the real disruption is not in replacing classrooms with cheaper and faster equivalents, useful though that might be, but in digital organizational intelligence -- using increasing quantities of data to develop and test new hypotheses about customer behaviour, market opportunities, environmental constraints, and so on -- developing not "best practice" but "next practice".

Richard Benson-Armer, Arne Gast, and Nick van Dam, Learning at the speed of business (McKinsey Quarterly, May 2016). HT @annherrmann

Chris Argyris and Donald Schön, Organizational Learning: A Theory of Action Perspective. Reading, MA, Addison-Wesley, 1978.

Monday, August 29, 2016

The Judgment of whole Kingdoms and Nations

@Cybersal @kirstymhall @UKParliament #Brexit #VoxPopuli

A radical Whig tract was published in 1709 under the title Vox Populi, Vox Dei. The following year, an extended version was published under the title The Judgment of whole Kingdoms and Nations. I want to use these two phrases as the starting point for my submission to the UK Parliament Public Administration and Constitutional Affairs Committee, which has launched an inquiry into the lessons that can be learned for future referendums.

The first thing I want to mention is the rushed timescale. The inquiry was announced on July 14th, with a deadline for submissions of September 5th. I shall argue that this rushed timescale is symptomatic of the referendum itself, in which people were asked to make a complex decision with inadequate information and analysis.

(For the sake of comparison, an inquiry on the future of public parks was announced on July 11th, with a deadline of September 30th. So we are given more time to analyse the physical swings and roundabouts of council-run playgrounds than the metaphorical swings and roundabouts of parliamentary sovereignty and media oversight.)

To be fair, most parliamentary inquiries only give you weeks rather than months to compose a submission. This effectively limits submissions to people who have already formed an opinion, and already have the evidence to support this opinion. In other words, experts.

But then most parliamentary inquiries are about issues that people have been concerned about for a much longer time: Bus Services, Employment Opportunities for Young People, Food Waste. There is an existing body of knowledge relating to each of these topics, and it is not unreasonable to ask people to base a submission on their existing knowledge.

In contrast, nobody knew precisely how this referendum was going to be mismanaged until it actually happened. Although many people (including some Brexiteers as well as many Remainiacs) predicted that it would end in tears, and can now say "we told you so".

I can read you like a magazine ... Don't say I didn't say I didn't warn you (Taylor Swift)

No doubt the Select Committee can expect to receive a number of submissions that fall under the heading of what the Dictionary of Business Bullshit calls "Pathologist's Interest".

But told-you-so is not a good starting point for a proper analysis, because it concentrates on confirming one's previous expectations, rather than discovering new patterns. So the Select Committee might not get much well-grounded analysis. Partly because there isn't time to do it properly, and partly because many of the potential "experts" are affiliated to UK universities, which are currently on summer vacation. Looks like the Select Committee is falling in line with Michael Gove's idea that "the people in this country have had enough of experts".

("The Voice of Gove is the Voice of Government". I wonder what that would look like in Latin?)

The official announcement sidesteps from "the lessons that can be learned" to "lessons learned", which is not the same thing at all. The former suggests an open exploration, while the latter suggests merely rattling through a project postmortem for form's sake. The timescale does not seem particularly conducive to the former. So is this apparent haste triggered by thoughts of a second referendum, or it is just intended to curtail criticism of Parliament for its earlier folly?

As I have argued elsewhere (including my book on Organizational Intelligence) complex sense-making and decision-making cannot go straight from the Instant of Seeing to the Moment of Concluding, but require what Lacan calls Time for Understanding. In this respect, the inquiry repeats one of the errors of the referendum itself.

The timescale and debating rules for the Brexit referendum were modelled on a General Election campaign. But a General Election has three important characteristics that were absent from Brexit. Firstly the electorate is generally familiar with the main parties: Labour and Conservative were around before any of us were born, and the Lib Dems also have long-established roots. Secondly, there is some rough notion of symmetry between the two main parties. Thirdly the parties make promises to which they will be held accountable in the event of victory. In other words, the General Election campaign can be compressed into a matter of weeks precisely because the rules of engagement are broadly understood, and there is very little new material for the electorate to process.

In comparison, as Kirsty Hall argues, the referendum for Scottish Independence was given a lengthy period of debate and analysis, because of the perceived complexity of the issues that needed to be considered. This would have been a much better model for the Brexit referendum.

Finally, let me return to the phrase "the judgement of whole kingdoms and nations", which of course raises the prickly subject of sovereignty. Although we supposedly have a system of parliamentary sovereignty in this country, parliament occasionally permits the voice of the people to be heard. As the Latin phrase has it, The Voice of the People is the Voice of God; and as the Establishment has discovered, the People can be a vengeful God. Parliament is still learning to listen to this vengeful voice. But who will teach what these lessons mean, and in what timescale? Or will the Establishment just adopt a Brechtian solution?

Update: Since I wrote this post, the Electoral Reform Society has published a critical report on the Brexit referendum, which makes the same unfavourable comparison with the Scottish Independence referendum that Kirsty made back in June. The Society has confirmed that it will be making a submission to the Parliamentary Inquiry.

Will Brett, Doing Referendums Differently (Electoral Reform Society, 1 September 2016)

Kirsty Hall, Brexit was a Con (28 June 2016) HT @cybersal @MerrickBadger @Koann

UK Parliament: Future of Public Parks, Lessons Learned from the EU Referendum

Wikipedia: Vox Populi, Vox Dei 

Friday, July 15, 2016

Boughing to the Inevitable

What is the best time to plant a tree?

A popular answer to this question is that the best time to plant a tree is twenty years ago, and the second-best time is now.

This is often claimed to be an ancient Chinese proverb. Or an African proverb. It is unlikely to be either of these.

And obviously we are not supposed to take this proverb literally. Because if the best time was twenty years ago, the second-best time would be nineteen years ago.

But instead of interpreting this logically, we are presumably supposed to interpret it as a motivational statement. Don't waste time regretting that you didn't plant a tree twenty years ago, act now to make sure you don't have similar regrets in twenty years' time. (Do real Chinese proverbs do motivational statements? I suspect not.)

In his new book, The Inevitable, Kevin Kelly talks about the opportunities for internet entrepreneurs thirty years ago. "Can you imagine how awesome it would have been to be an ambitious entrepreneur back in 1985 at the dawn of the internet?"

He then looks forward to the middle of the century. "If we could climb into a time machine, journey 30 years into the future, and from that vantage look back to today, we’d realize that most of the greatest products running the lives of citizens in 2050 were not invented until after 2016."

In other words, for an internet start-up the second-best time is now.

By the way, I'm not the first person to use the pun about 'boughing' to the inevitable. For example, @rcolvile used it in the context of ash dieback. "Half the trees in the country were going to be torn down. He’d already had to veto a particularly insensitive press release describing him as 'ashen-faced' about the situation, but 'boughing to the inevitable'. Meanwhile, Google is asking me if I meant 'coughing to the inevitable'. Thanks Google, it's always useful to spot something you haven't yet mastered., The Inevitable
Kevin Kelly, The Internet Is Still at the Beginning of Its Beginning (Huffington Post, 6 June 2016)

On The Best Time to Plant a Tree (Reddit)

Robert Colvile, Friends: The One with the Guy in a Yellow Tie (Telegraph, 3 November 2012)

Saturday, November 28, 2015

The Quantum Organization

In road traffic, says my friend @antlerboy, the one with the most momentum has the most responsibility. Perhaps that's true in other fields too?

Meanwhile, in a hierarchical organization, the one with the highest position has the highest authority. This is known as positional power. Unfortunately, responsibility and authority are not the same thing.

According to Heisenberg's uncertainty principle, the more precisely the position is determined, the less precisely the momentum can be determined, and vice versa. This is one of the central principles of quantum mechanics.

An analogous problem in most organizations is that responsibility and authority are poorly aligned. In other words, the person who pulls the strings isn't always the one who gets the blame when something goes wrong. And similarly, the person who does the work isn't always the person who actually knows how to do it properly. Position versus momentum.

There is a useful technique for organizational analysis known as RAEW (responsibility, authority, expertise and work), which was described by Roger Crane in the 1980s and adopted in some versions of Information Systems Planning. Unlike better-known techniques for responsibility assignment such as RACI, which describe how responsibilities ought to be distributed in an ideal (linear, clockwork) organization, the RAEW technique allows us to analyse how (badly) responsibilities are distributed in a real (chaotic, quantum, snakepit) organization.

And maybe fix some of the problems?

Related Posts: Clockwork or Snakepit? (June 2010)

Wikipedia: Responsibility Assignment Matrix, Uncertainty Principle.

Open University: Handy’s four types of organisational cultures

Updated 8 April 2016

Sunday, June 14, 2015

Scotch Eggs and Feedback

A primary school in Essex has banned various "unhealthy" snacks from pupils' lunchboxes. Teachers inspect pupils' lunchboxes and confiscate banned items.

The right-wing press is particularly aggrieved about the banning and confiscation of scotch eggs, which they appear to regard as an upper class snack. The Telegraph informs us that "they were created almost 300 years ago by Fortnum+Mason as a pocket-sized snack for aristocrats travelling by horse-drawn carriage". And Quentin Letts in the Mail enthuses about the scotch eggs at his prep school, with a sidebar describing the products sold today by a range of supermarkets including M+S and Waitrose. 

However, what struck me most about the story was the response from the school.
"Our healthy lunch box policy has been in place for some time and the majority of parents are very supportive of it. The decision to take additional steps to ensure all pupils are adhering to the policy was taken following feedback from parents and as part of our continued efforts to make improvements to all areas of the school. All school meals we serve comply with the government's school food standards, as required by law."

Feedback? The correct use of the word, according to Wikipedia, is when outputs of a system are "fed back" as inputs as part of a chain of cause-and-effect that forms a circuit or loop. In this context, the word probably means something like "a few random comments to which we have decided to overreact".

In the Deal and Kennedy model of organizational culture, a process-based culture is characterized by low risk and slow feedback. See my post On Agility, Culture and Intelligence (November 2012).

Dan Hyde, Scotch eggs branded junk food and confiscated from children's packed lunches (Telegraph 9 June 2015)

Quentin Letts, Keep your hands off my rusty cannonballs: As an Essex school outlaws Scotch eggs, QUENTIN LETTS explains why he's a life-long devotee (11 June 2015)

Wikipedia: Feedback, Organizational Culture

Thursday, June 4, 2015

Political parties and organizational intelligence 2

#orgintelligence #politics @rafaelbehr contrasts the behaviour of the Conservative and Labour parties.

Before the 2015 election, the Labour party practised collective denial ("misplaced confidence", "kidded themselves"), believing that "organization could compensate for uninspiring leadership". Following the election, "a danger now is oversteering the other way".

Denial and oscillation are two of the principal symptoms I have identified of Organizational Stupidity (May 2010).

Contrast this with the Conservative willingness to invest in 'blue collar conservatism'. Behr attributes this initiative to George Osborne, one of whose political gifts "is the self-knowledge to identify gaps in his own experience and to plug them with astute appointments". Cameron, he suggests, is much less intellectually curious than Osborne. And yet it is Cameron who carries through Osborne's plan to appoint Robert Halfon in order to recalibrate the Conservative's relationship with the working classes.

What reveals itself here is a form of intelligence and leadership that is collective rather than individual, a form of collaboration and teamwork that has not been strongly evident in the Labour Party recently.

Steve Richards goes further ...
"During Cameron’s leadership the Conservatives have become more alive as a party, impressively animated by ideas and debate. Cameron appears to be an orthodox Tory but likes having daring thinkers around him, even if they do not last that long. ... In recent years Conservative party conferences have been far livelier than Labour ones, which have been deadened by fearful control freakery."
... and insists that "the next Labour leader must not be frightened by internal debate".

One of the essential duties of leadership in any organization must be to boost the collective intelligence of the organization. Not just debate, but debate linked with action.

Patrick Wintour reports that there was plenty of (apparently) healthy argument in Labour's inner circle.
"Meetings were quite discursive, because there were a large number of views in the room. ... [Miliband] enjoyed that. He used the disagreement as a means to get his own way. It is a very interesting case study in power, in that he would not be described typically as a strong leader, but very consensual. The caricature of him is as weak, but internally he had great control."
But that's not enough.
"The team that Miliband had assembled around him consisted of highly intelligent individuals, but the whole was less than the sum of its parts – it was, according to many of those advisers, like a court in which opposing voices cancelled one another out."

Furthermore, an important requirement for organizational intelligence is that it is just not enough to have an inner circle of bright and well-educated 'spads', and to appoint either the cleverest or the most photogenic of them as "leader". Perhaps the Labour inner circle deeply understood the political situation facing the party, but they neglected to communicate (forgot to mention) this insight to others. The vanguard is not the party. Any party that aspires to be a movement rather than a machine must distribute its intelligence to the grass roots, and thence to the population as a whole.

Exercise for the reader: count the ironies in the above paragraph.

Finally, intelligent organizations have a flexible approach to learning from the past. @freedland argues that Miliband was single-minded about the future, and refused to tackle the prevailing narrative about the Labour government's role in the 2008 economic crisis.
"The management gurus and political consultants may tell us always to face forward, never to look over our shoulder, to focus only on the future. But sometimes it cannot be done. In politics as in life, the past lingers."


Rafael Behr, The age of machine politics is over. But still it thrives in the Labour party (Guardian 4 June 2015)

Jonathan Freedland, ‘Moving on’: the mantra that traps Labour in the past (Guardian 5 June 2015)

Tim Glencross, Attack of the clones: how spads took over British politics (Guardian 19 April 2015)

Brian Matthews, The Labour Party and the Need for Change: values, education and emotional literacy/intelligence (Forum, Volume 54 Number 1, 2012)

Steve Richards, Labour’s next leader should look to David Cameron, not Tony Blair (Guardian 1 June 2015)

Patrick Wintour, The undoing of Ed Miliband – and how Labour lost the election (Guardian 3 June 2015)

Chris York, The Rise Of The Spad: How Many Ministers Or Shadow Ministers Have Had Proper Jobs? (Huffington Post, 13 November 2013)

Related Posts:

Symptoms of Organizational Stupidity (May 2010)
Political Parties and Organizational Intelligence (May 2012)
Dark Politics (May 2015)

Updated 6 June 2015