We’ve all heard the expression, ‘many hands make light work’ – but when it comes to Intelligence, are two heads really better than one?
This week we consider the benefits of Crowdsourced Intelligence, and whether it can actually work!
TL;DR; Crowdsourcing is good for information collection and problems with clearly defined answers. But, in my opinion, less effective in terms of maintaining secrecy and predictive Intelligence Assessments.
So, WTF is Crowdsourced Intelligence?
Crowdsourcing is the phenomenon of having multiple people work on the same problem.
It involves the collaboration and shared knowledge of many to solve problems and achieve resolutions quicker than one person could.
Basically, it’s sharing brainpower –
In 2009, The U.S. Defense Advanced Research Projects Agency (DARPA) issued a challenge to locate 10 red weather balloons at 10 random locations in the United States. A team from MIT won the $40,000 prize by setting up an website for community-based leads and using social networks. They located all ten balloons in just under 9 hours.
Some of you might have used WAZE, the GPS app which uses crowdsourced information to provide alerts about car accidents, traffic, and road conditions, provided by the user community.
In 2010, DARPA again spent $13 million dollars on a project called “Deep ISR Processing by Crowds, designed to “harness cognitive and creative abilities of large numbers of people to enhance the knowledge derived from ISR systems.”
So these modern applications have proved there’s a place for crowdsourcing within Intelligence –
But is this good for modern Intelligence? Or, do we run the risk of group-think?
I recently read ‘The Wisdom of Crowds‘ by author James Surowiecki who argues that groups of people are smarter than individuals (and better at solving problems). Here are some arguments for Crowdsourced Intelligence:
Advantages of Crowdsourced Intelligence
Threat actors are likely to act and behave in similar ways. Collaboration between analysts allows for the sharing of intelligence and the implementation of timely treatments. This is good.
Networking and reciprocal relationships. As an intelligence professional, one of the first things you’re taught is ‘you don’t have to know everything, but you need to know where to find information when you need it’ – Crowdsourced Intelligence provides for increased collaboration fostering relationships and knowledge sharing. This is also good.
Context and Specialties. Different people have different points of view. A crowdsourced approach can provide unique, and interesting findings from people with broad (and often varied) contexts.
- Quick case-study – An investigator with technical experience in SIGINT recently proposed a novel strategy to gather information in prison. He wasn’t part of our team, but a professional colleague who was able to offer unique insight based on his experience.
Eliminate Bias. Like I’ve said before, we’re all susceptible to bias which can lead to over-confidence, or over-optimism while making assessments. Crowdsourcing can help to illuminate ‘blind spots’.
Bellingcat is a great example of successful crowdsourced Intelligence. If you haven’t heard of Belingcat, do yourself a favour and check them out.
Using information provided from multiple OSINT sources, and online collaboration, Belingcat participants successfully identified, and solved significant incidents including identifying the people responsible for the Skripals poisoning case in the UK.
Like the Red Balloon Example, the dispursed crowd was able to gather disparate information demonstrating the strength of the crowd!
How many Jellybeans are in the Jar? Time and time again, the phenomenon of ‘group average’, is used to justify the collective conscious, and in multiple studies, the group average is usually better than the best individual guess.
I tend to agree if the outcome is defined, like counting jelly beans in a jar, or solving a puzzle with a clear answer – then sure, crowds make sense.
But what if you need to predict the future?
The Downside of Crowdsourced Intelligence
Need to know. Intelligence, by its very nature, needs to be handled in confidence demanding a level of secrecy. Arguably, if the circle of trust gets too big then the chances of information leakage increase also – It’s hard to keep information secret when everyone knows it.
Context. Participants need to be qualified in the problem they’re solving. While it doesn’t take too much knowledge to guess the number of jelly beans or report traffic incidents on the way to work – When it comes to intelligence, participants need to have some contextual knowledge of the problem.
Source evaluation is in the eye of the beholder. Crowdsourced intelligence is impacted by the interpretation of source information.
Crowdsourcing happens quickly. You know when you’re part of a group chat and your friends start talking about catching up after work for a drink. You put the phone down for 30 minutes and when you pick it back up someones already chosen the location and meet-up time. If you’re not able to get on the bandwagon when it kicks off, it’s often too late!
Conformity. You know that experiment where everyone gives the wrong answer on purpose then those people who think different just agree with the majority – Yea, like that (Asch experiment video below)
Diluted potency. Every intelligence analyst is told to be bold, to make predictions that stand out, to bravely stand in front of the Full Bird Colonel and tell him his decision is playing right into the hands of the enemy. But can this truly be achieved by the crowd? By very definition, aren’t the outliers drawn into the mean, and don’t the most potent assessments get diluted by the ‘happy medium’?
My Thoughts –
Crowdsourcing is a powerful phenomenon, but each participant must have a specific needs role within the process and there needs to be an element of control. While decentralisation is necessary for foster creativity, ‘creative direction’ has to exist to ensure participants contribute towards a common objective. Otherwise people might just go off on a frolic.
How you can get involved in crowdsourced Intelligence
Bellingcat – As discussed before, the home of Online Investigations. Check them out and learn OSINT to support their cause!
TraceLabs CTF –TraceLabs use crowdsourced OSINT to help find missing persons. Interested in finding out more, read my article here or get involved in one of their Capture The Flag (CTF) competitions where they provide real leads to help find missing people!
Metaculus – Metaculus is a crowed-voting site allowing you to pose and answer questions about the occurrence of a variety of future events – think of it like tapping into a community of people keen to support crowdsourced intelligence.
Europol has an initiative called ‘Stop Child Abuse – Trace An Object’.
A couple of questions for people interested in Crowdsourced Intelligence –
Is there an optimal number of people in a group? If businesses and Government Intelligence Agencies use secret information for advantages over their adversaries how will the participation of 100,000 people remain secret?
Diminishing returns? Can there be too many voices?
Finally, We need to differentiate between crowdsourced ‘information’ & crowdsourced ‘Intelligence’
These two things are NOT the same.
While I agree the more sources of information is likely to increase the reliability of intelligence products, I’m sceptical of the reliability of crowds to produce predictive intelligence products. Your thoughts are welcome (below).
So, when it comes to predictive intelligence, can a consensus actually be achieved? Moreover, is it really possible to think outside of the box if we rely on the box to do the thinking?
Your thoughts welcome!
p.s. Last week I posted an article on Codes and Ciphers – I’ve hidden two secret codes somewhere within the post. If you haven’t seen the article, check it out and crack the code, then LMK to claim a prize!