Trust bubbles: how security, trust and economic prosperity are interlinked

Here's a fascinating area of research I heard about from Virgil Gligor at the CSIT cyber security summit in Belfast last week.  Societies with a high degree of trust have better economic growth; there is a link between trust and economic prosperity.

A 2007 study by Dincer and Uslaner uses US economic data to prove a positive link between 'generalized trust' - trust in strangers - and economic growth. A 10 percentage point increase in trust yields a 0.5% increase in GDP.

This has some important consequences, especially given the current push to improve cyber security.

I can't stress how important it is to understand that 'generalized trust' is about trust in strangers, not trust bonds within community groups, social networks etc.

Typically, online security research tends to focus on the latter. How can Alice establish that Bob is a friend, and can she trust each communication she receives is actually from Bob and not some interloper?

There's a temptation to develop technology to facilitate trust in two ways. Firstly to provide certainty in point-to-point connections: Alice knows she's talking to Bob, and vice versa.

And secondly to build systems which measure and display the trustworthiness of strangers to other strangers. For example the buyer feedback mechanism on eBay and other customer review systems.

But neither of these help with generalised trust, and generalised trust is important to economic growth.

Whilst a customer review mechanism may seem to help foster trust between strangers, it is still only a specialised application to help established sellers prove their trustworthiness.  It's not a system that builds on generalised trust but a system which predicts specific trust.

A customer review mechanism can also actively exclude innovative new sellers simply because they don't yet have a formal reputation.

With zero generalised trust, we believe that any deal with a stranger is guaranteed to fail.  Customer review mechanisms can help by allowing established sellers to advertise their trustworthiness.

But in such a society only established players can ever compete. New entrants are effectively barred by virtue of having no trustworthiness and no way to establish this and the market risks becoming lazy or cartel-like.

There will be reduced drivers for improvement between existing sellers.

Now, introduce even even a moderate degree of generalised trust. Not all deals with strangers are guaranteed to fail and some buyers will be prepared to take a risk to get a bargain.

There is now a mechanism to establish trustworthiness from scratch and innovative new sellers can enter the market and disrupt lazy incumbents.

The link between generalised trust and economic prosperity poses an interesting conundrum for cyber securocrats and digital policy-makers alike.

It seems obvious. If I knew the email was from my bank - if I could trust the communications path - I would know it wasn't a phishing attack.

But establish an intricate system to broker trust in a digital community and we risk building the online equivalent of a closed community group fearful of strangers.

A group reliant on formalised trust mechanisms may become afraid to innovate or take risks outside of the trust bubble.

Or indeed take risks which buck the formal guidelines inside the bubble, leaving the equivalent of trust outcasts stuck in the bubble who will never again be trusted.

Rehabilitation is impossible inside the trust bubble.  Social mobility is also likely to be hampered.

There is also the possibility of systemic flaws being exploited within the trust bubble.  No system is perfect, and there's a risk those who learn to game the trust mechanism will prosper over other more-trusted competitors.

I suggest that a community which becomes highly reliant on formal trust mechanisms will be less prosperous than other more 'chaotic' communities.

There is an alternative. Technology has made certain cyber crimes and other exploits possible, but technology can also evolve to restore trust and reduce risk without relying on entirely new schemes to broker trust.

E.g. we all already have a built-in trust sensor honed over generations of evolution.  With faster broadband and camera-enabled devices soon it will be possible to have a face-to-face conversations with digital retailers e.g. for higher value transactions with previously unknown sellers.

Once spoken and facial cues are reintroduced back into the system it is possible to rely once again on human instinct to play a role in deciding who to trust.

Risks of exploit will fall-back to real life levels and the internet won't be seen in itself as a threat, just a communications channel.

On one hand this won't help the bottom line of retailers wanting to automate to save money, but economic prosperity on the whole will benefit through the link between generalised trust and economic growth.

With face-to-face contact becoming once again the norm, scams like phising won't scale and we will be able to have a degree of trust in online strangers.

Turing's confidence trickster? A question hangs over whether or not computers could ever fool humans, not just into thinking they're interacting with another human (ie pass the Turing test) but also to illicit a high degree of generalised trust in human subjects.

If so, the balance would swing back in favour of the phishers as online con-tricks would once again scale. One person could use technology to attempt to trick tens of millions of 'marks' at a time - until someone invents software for detecting non-humans!

Belfast 2012: privacy initiatives as an enabler for cyber security

I was lucky to get an invite to the CSIT 2012 cyber security summit.

Credit to the CSIT team in Belfast for creating a unique atmosphere which got the balance absolutely right - not too academic for industry to feel excluded, government participation without the grandstanding or justification sometimes seen, and plenty of opportunity for detailed open discussion amongst global delegates from industry, academia and government.

I attended for three reasons. Firstly, I spent the first 8 years of my software career designing secure communications (a patent from back when I worked at Motorola).

Secondly, I want to advocate against locking-down aspects of the internet in the name of security because I passionately believe open societies are stronger societies, and this applies equally to cyberspace.

And finally I wanted to discuss our work on privacy as an enabler to cyber security.

When each of us act with a high degree of autonomy, taking responsibility for our own personal data and protecting our own systems due to our inherent desire for privacy, I believe our networks as a whole will be more secure than if we defer to governments as our primary cyber defender.

This was a recurring theme throughout discussions.

There is what was referred to as an asymmetry in human resources faced by those tasked with securing our networks: relatively small numbers of professionals with varying skill levels facing off against legions of online activists and cyber criminals with an impressive skill set and access to most of the same tools (or 'weapons') as governments.

It went unsaid but one can only assume some nation states have also their own legions.

In a democracy we should be able to rely on citizens to augment the work of government in securing our networks, but for this to happen we need both mutual trust between citizens and government security agencies, and for citizens to feel motivated and able to help.

Specifically on privacy there are additional reasons why privacy can be an enabler to cyber security.

All sections of society are vulnerable to data loss through their ordinary everyday use of the internet.  Lack of data privacy can become a national security risk when personal data of those in sensitive positions becomes accessible to those with hostile intent, who may use private information to extort or to blackmail.

Whilst traditionally privacy has been seen in some quarters as at odds with security and policing requirements - e.g. the use of network monitoring tools to spot threats and investigate crime - there's also an argument against this.

In some cases a focus on intrusive policing and intelligence-gathering techniques can come at the expense of developing more sustainable community-based policing models for cyberspace.

Information technology is still developing apace, therefore capability-based policing - a reliance on a power available only to police and security forces such as network monitoring, data seizure etc - will for a while at least remain a costly arms race.

Soon after each capability is installed in our networks either technology evolves, requiring further upgrades, or the bad guys up their game leaving the capability obsolete, or both.

A worse scenario exists: the capability might be exploited by the bad guys.

Take the data retention directive, the EU law (which as far as I can establish the UK lobbied for) compels ISPs to store information which might be useful to law enforcement for a period of 6 months to 2 years.

All this potentially exploitable data sitting around in huge silos at ISPs. A capability that is also clearly vulnerable to exploit, especially given the security of such data is in the hands of private companies.

Employees have in the past looked up private data from company databases and sold this information to private detectives working for divorce lawyers, journalists and criminals.

It's not a straight choice between privacy and security. It's a balance between privacy as an enabler to creating a more secure information culture and privacy-invasive policing as a tool to detect and prevent cyber crime.

UPDATE: also, one must not overlook the importance of public trust and the benefits of consensual policing. Invasive monitoring can cause mistrust between the public and security services.

Public trust and confidence in the work of the security forces is a bigger asset, even, than the ability to monitor all electronic communications. A consensual approach makes society stronger and inherently stable.