The power of crowds – and their limits

The first internet revolution in the 1990s saw the emergence of on-line service providers who offered the same service as off-line competitors, but cut costs.  Big winners included Amazon and LastMinute.com.

But the second, and arguably bigger, internet revolution in the 2000’s was the emergence of social networks, with , and in the vanguard.  These surprisingly simple tools – Facebook is little more than a village noticeboard with millions, rather than dozens, of potential contributors – unlocked the power of the crowd.

Crowds can do some amazing things.  Whether you need to find someone extra smart, or aggregate thousands of tiny contributions to create something huge, virtual crowds organized through social networks have made possible what was previously unthinkable.

And the next target for crowds seems to be funding businesses – and most recently even academic science.

In principle, it sounds like a great idea: a small, almost invisible, contribution from enough people can together deliver enough resources to change the world.  Platforms, like Microryza, are springing up like mushrooms to harness these ‘donations’ and channel them to deserving causes.  “Crowdfunding” is on everybody’s lips.

The key difference with crowdfunding lies in the distributed decision-making capacity, rather than just distributing the cost

Can crowdfunding really shake the foundations of current financing models for biotech companies and for academic science? Or will it remain a niche opportunity dwarfed by conventional funding mechanisms?

DrugBaron predicts that initial excitement and expansion in the sector may rapidly be followed by disappointment and contraction – unless a solution can be found to the quality control issue when the funders cannot be expected to understand the details of the projects they are being asked to fund.

After all, existing public funding of academic science is nothing but “crowdfunding”.  Can social networks really do a better job of distributing our communal investment in research than research councils or even venture capitalists? Crowds can do amazing things, but crowdfunding biotech business or academic research might be a step too far.

As the old saying goes, there is nothing new under the sun.  We have been “crowdfunding” academic science and even start-up companies through our general taxation system for over a century.  After all, every government service is funded through the aggregation of a vast number of small (compared to the scale of the costs) tax contributions.

So what’s different about the current wave of internet-enabled crowdfunding?

The key difference lies in the distributed decision-making capacity, rather than just distributing the cost.  Each individual makes a decision what they want (and equally do not want) to support – in marked contrast to the 20th Century model of crowdfunding via general taxation, where the money was gathered in and a central body (typically research councils of one kind or another for public funds, and VCs for private funds) allocated it.

Deciding whether 21st Century crowdfunding is an improvement over the old model, then, depends on whether individuals or their chosen representatives make better decisions on the allocation of capital.

At least with a fund manager you can look at their past performance – what similar metric could you apply to a Facebook “friend”?

Free market adherents (and DrugBaron is certainly one of those) will tell you that individuals will allocate their capital more efficiently than central control.  Importantly, though, the efficiency of markets depends entirely on the equality of information among the market participants.  The problem then for 21st Century crowdfunding of healthcare businesses is the sheer complexity of the subject matter.

Its important to distinguish healthcare from many tech businesses.  Crowdfunding is absolutely perfect for any consumer-focused enterprise – it combines the two most important ingredients for success into a single injection that can put the best business ideas on steroids: market research and cheap capital.  After all, if the idea for your company is to sell a new kind of mobile phone or an app for locating garden centres then the success of the business will depend on how well its unique features appeal to users.  Far to better to find out that no-one really sees the point of another note-taking app at the outset when you try and seek funding for your project than after you’ve coded it and launched it.

Internet-based crowdfunding will, DrugBaron predicts, completely change the business model for much of IT venture capital (and most other consumer businesses) over the next few years.  There is no reason at all why the majority of the capital needed for such enterprises could not, and will not, be raised that way.

But all good ideas have their limits.  And expanding 21st Century crowdfunding, with its disintermediation of the decision-makers, into academic research and drug development companies is surely a step too far?

Its not that research councils or peer-reviewers, or perhaps even healthcare VCs, are good at allocating capital.  For the most they are not (principally because its very hard – early stage biological research is a classical low validity environment, where the small amount that is typically known is poorly predictive of the behavior of the system as a whole).  Picking winners is probably impossible.  The best you can do is enrich the pool by killing losers and minimizing the capital required to advance.

It would be extremely arrogant to assume there are not smarter people out there than the current holders of the purse-strings

There surely are.  But leveraging a crowd to find them highlights the particular nature of the power of crowds: crowds are good at parallel hard tasks or serial easy ones.

Kaggle is a perfect example of leveraging parallel hard tasks to find the best solution to a complex problem.  A typical big data problem is presented to the universe of ‘solvers’ who each try their own methodology, based on experience, intuition or luck, and there is an objective assessment of the best analytical solution to the presented problem.  Kaggle matched the single “smartest” solver (in the context of the particular problem) with the owner of the problem.

By contrast, SETI@Home (the Search for Extraterrestrial Intelligence), one of the earliest crowdsourcing projects over the internet, is a good example of leveraging serial easy tasks.  A vast dataset is broken into chunks and each is scanned in turn by a different person.  Spotting patterns in the data is easy, its simply a matter of having enough pairs of eyes.  No two people looked at the same piece of data (hence it’s a serial task and not a parallel one).  But they (and their computers) achieved a scale of data-analysis that could never have been achieved with central super-computing resources of the day.

In the first example, in the limit only one person in the universe of players needed to be smart – and the large number of average quality players in the field had no impediment to the ability of the winner (everyone’s play was completely independent, at least in principle).

In the second example, no-one had to be more than averagely smart.  Numbers rather than quality was the key to success.

Both these examples fit our real world experience: very smart people are very rare; on average most people are “averagely” smart.

Crowdfunding a biotech business is very different.  Now everyone who contributes their capital to the funding has to understand the risks and rewards of the business.  The quality of the decision-making isn’t driven by the ability of the smartest player (as in the Kaggle example) nor is it benefited by the large number of people who have to make a contribution (like the SETI example).  The quality of the decision will be based on the average smartness of the players – it’s a ‘serial hard task’.

“Raw” crowdfunding, then, will make a poor job of allocating capital to scientific research or biotech start-ups.  Lots of votes (investment) means something for a consumer-focused business – even if the people voting aren’t super-smart they are at least representative of the future customers. But for a technically complex product that will be sold to doctors or governments, the average opinions of mostly average people is unlikely to be an improvement on the 20th Century crowdfunding model of general taxation and representative decision-making.

Crowds are good at parallel hard tasks or serial easy ones

But social networks mean that modern crowdfunding is anything but “raw” – the decisions made by the participants are not independent, but heavily driven by the opinions of a few mavens.  I don’t need to understand a business, I just need to be connected to someone who does (and whose opinion I trust and value).  Professional investors often decide what to invest in based on surrogate markers – who had the idea, the prestige of the institution that they come from, the other people who like the idea and so on.  So crowdfunders should be able to do the same – and potentially do it better if the networks are good.

So why doesn’t DrugBaron see such a model challenging the current status quo, with capital allocation in the hands of professional representatives (whether research councils or VCs)?  Because social networks are good on breadth and less good on depth.  You need to really know someone well to decide to trust their judgement (or rational analytical assessment, for that matter).  Social networks garner wide opinion, but not deep, trustworthy opinion.  At least with a fund manager you can look at their past performance – what similar metric could you apply to a Facebook “friend”?

And it isnt just a case of not being better – modern crowdfunding may actually be damaging.  There is already a tendency for research councils, governments and VCs to follow fashion – damaging returns as DrugBaron noted previously.  Crowdfunding will only exacerbate that tendency.  “Hot” areas (like cancer treatments and developmental biology) may be easy to attract funding – ugly-duckling fields like sepsis are unlikely to find it any easier to convince the masses than it is to convince the current generation of fund-holders.

Worse still, the key attribute for attracting funding may be charisma – in the same way Tony Blair dominated UK politics with plenty of flair and very little substance, so scientist-salesman may come to dominate crowdfunding platforms, heavy on the expansive vision and light on the details.

The efficiency of markets depends entirely on the equality of information among the market participants

Do we want a world where the subjects for scientific research are chosen for their ease of understanding, appeal to the masses, or the sales pitch of their originator?

Crowds can do amazing things – as Kaggle has shown us.  Crowdfunding will revolutionize business funding in the same way, for consumer-focused businesses.  But even crowds have their limits, and funding academic science or biotech start-ups (or any other ‘serial hard’ tasks) looks like a bridge too far.  Its not time for research councils or healthcare VCs to hang up their spurs just yet.