ISO 13485 et. al.

Medical device development is governed by Standards.  If you want to bring a product to market, then there’s only one way to do it: Compliance with the rules.  

And it’s not enough to say that you do.  You have to prove it.

The relevant Standard under the European Medical Device Directive is ISO 13485: 2012 – Quality Management Systems for Medical Devices.  It sets out best practices for building a medical product, including design process, control of manufacture, human resources, and business organization.  In each EU Member Country, the Competent Authority of experts sets the rules for each type of product in accord with these guiding principles.  Then private companies, the Notified Body, assesses businesses and grant permission to market product by affixing a CE mark.

Gaining ISO 13485 Certification and a CE mark is not easy. The rules are detailed and inflexible.  For years, in corporate life, i resisted the whole idea: writing down the detailed methodology and assessing risks assumed that you understood the solution.  It blocked truly creative thinking that led to innovative breakthroughs. 

I preferred  ‘agile’ approaches, in which projects step, learn, and iterate.  i was well known for the ‘Research Exemption’, setting my department apart as an exceptional case that didn’t have to follow the rules.

‘all the more ironic that I was required to build a fully compliant system in less than six months.

As previously noted, we passed our Stage 1 Audit by the Notified Body (NB) back in March.  We had three major non-conformances (gaps in our system that must be addressed), and two minor ones (deficiencies to be remedied).  Our NB gave us an extra two months to come into compliance, with our Stage 2 Audit scheduled for June 13-14.

It was an enormous amount of work.  There were endless debates about terminology ('What is a lot, and how does it differ from a batch?), procedures (How often do we deploy settle plates to assess our clean room?), design requirements (Does the static resistance of the coating matter more than the dynamic resistance?), and process validation (Do we define the operating limits of temperature or the best operating point?).  I suppose that it forces you to really understand the business, how to assure a safe product.

If you watch a professional cooking show like Chef’s Table, the key to a good restaurant is only 10% innovation and presentation.  90% is consistency: Can staff deliver a great experience to everyone, repeatedly, day after day.  Arguably, if you order the same dish twice, it should be the same dish.  Professionals strive obsessively towards that goal.

Similarly, every medical product unit that we produce should be as safe and effective as every other one.  Numbing consistency is the Standard.

So, for the past two days, two auditors assessed how numb we have become.  They combed through details of every aspect of the Colworth facility, following paper trails from purchase to product, design documentation from concept to production.  The pre-rumor was that NBs had walked out on unprepared facilities.  Everyone was involved in making sure that the apocryphal tale didn’t become our story.  Our advisors orchestrated the inspections and evaluations flawlessly, we corrected, printed and signed dozens of final documents.  We were able to discuss every procedure, produce (almost) every piece of paper, that was required.

And, at the closing review, we had only one minor non-conformance and were recommended for certification.

It’s really an amazing outcome.  A first-time startup, a half a dozen people and six months to build it from nothing.  The odds are heavily against passing on the first try in any case, much less with so little to apply to the task. 

Yet we did.  It stands testament to the quality and effort of the people, and their attitude and experience in meeting the challenge.

I think that folks in my corporate life would never believe that one of my groups could achieve this.

And most colleagues don’t understand it today, either.  The dominant theme today was ‘That’s nice: Why hasn’t my task gotten done yet?”

‘One thing at a time; we’ll all get there in the end.

Navigating our S1 audit

This starts out geeky, but leads to a happy ending.

Medical Devices require regulatory approval before they can be sold to doctors and patients.   In the US, this review is performed by the Food and Drug Administration, while Europe uses private companies under government oversight.  In either case, the goal is to check that the products are safe and effective for their intended use, and to ensure that manufacture produces product of consistently high quality. 

Europe operates under an EU Medical Devices Directive, which requires countries to set up panels of experts to establish rules and test procedures for every type of medical device.  These Competent Authorities, the MHRA in the UK, approve companies, the Notified Bodies, who apply the rules and approve devices for use.  The approval itself is a registered CE mark, which is applied to conforming product.

Thus, passing through the approvals process is the major gate standing between development and revenue.

 

The gate is opened by passing three audits: one for organizing and documenting manufacture appropriately, one for operating manufacture appropriately, and one for providing evidence that the product is safe and effective.  We have been configuring and documenting our manufacture during the past two months, ‘Establishing an ISO –Compliant 13485 Quality System’ in the industry jargon.

It is a very detailed and tedious process: we have 22 central Standard Procedures that have to be written and reviewed, nearly 100 Forms, and lots of supplemental pieces like meeting minutes, IT support, job descriptions, problem reports, training records, and company policies. 

 

Four of us have worked all day, every day for the past two months to get this all set up in anticipation of our first audit by our Notified Body today.

Establishing a Quality System normal takes 6 months to a year, but we are trying to do it much more quickly: funds, licenses, and rule changes would be deadly for the startup if things lingered on that long. 

A scientist / entrepreneur / free-thinker, I have long struggled with documentation, compliance, and audit conformance in corporate environments: it is too many limits and too much pedantic detail.  Thus, it’s both ironic and frightening that I have to set up my own system and be tested on it.

 

So, the audit was today and….We Passed!  (‘actual photo of audit in process, above) 

We had a couple of minor non-conformances (goals need to be measureable; auditors need to be impartial) and three majors (two for meetings that need to be held and one for selecting a sterilization vendor), all easily fixed.

Our Clean room will be installed on Monday, and we will begin producing product samples for testing in the following weeks.  But this is a big deal and a huge step forward towards creating a revenue-generating and profitable company.

Understanding medical stories

When a patient is brought to the A&E (Emergency room) by ambulance, the patient is quickly transferred to a waiting bed where a team of doctors and nurses begin treatment, stabilizing the patient, ordering tests, and starting treatments.  It’s a familiar scene from television dramas (and, unfortunately, to any parent).

Less visible is the rapid transfer of patient information from the ambulance to the hospital: the history, vital signs, and procedures during the initial encounter  on-scene with the medics.  We’ve gotten much better at transmitting demographic and device data ahead of the patient’s arrival.  But data alone is not enough.

Watch closely, and you’ll see the medic and physician have a brief conversation about the patient during the handoff.  They are sharing a structured story, explaining the incident’s circumstances, the treatment rationale, the observations that give meaning to the charted measurements.

Long ago, I led a research project to capture and interpret these narratives: we combined a voice recorder, transcription software, and medical language parser to try to extract the “facts of the case” from the narrative.  The technology was too crude for more than a proof of concept demo and a patent, but I’ve always had a soft spot for the idea.

Historically, machine transcription and translation based on linguistic rule-based approaches has always been fragile and unreliable.  I remember reading to Dragon for hours, training it with prepared texts and hoping it would get smarter.  We took a license to MedLEE, an expert system that extracted clinical concepts from medical texts.

Today, transcription and translation software uses statistical approaches that sidestep the whole question of structure and meaning.  Given a snippet of spoken or written text, computers simply search vast libraries of paired ‘voice-text’ and ‘native-foreign’ documents to find a match. 

As the libraries grow, so does the accuracy of the process.  Like any ‘big data’ project, the process needs well-curated content to get started.  But as it grows it can ‘learn’ and improve itself through simple removal of inconsistencies (cleaning our errors) and recognition of similarities (colloquialisms and metaphors).

The growth of medical information systems means that vast libraries of  ‘medical narrative-patient chart’ pairs are available.  Although held behind patient privacy walls, de-identified data should be useful in developing better systems that mimic the human process for extracting the ‘facts of the case’ from the stories that medics tell physicians in those first moments at the doorway to the hospital. 

It may add an important dimension to understanding the simple ‘flow sheet’, the chart of vital signs data and medications over time.  Medical stories can explaining why things happened alongside the simple ‘what’ .

Pot Odds and Portfolio Entrepreneurs

What does Texas Hold ‘em teach about entrepreneurship?

Hang in: this will be three definitions, one simple unifying rule, and a big epiphany.

1)  Texas Hold ‘em is a variant of Poker, a card game, in which two cards are dealt face-down to each player and five shared cards are dealt face-up in the middle of the table.  The winner is the one who makes the best 7-card poker hand using their two hidden cards.  It’s a beloved pastime among Wall Street and business strategy folks.

2)  Players can compute the probability of winning in a specific situation by calculating Hand Odds.  Suppose, for example, that you are dealt the 6 and 7 of spades and that the four shared cards dealt so far have two more spades showing.  What is the chance of you getting a flush when the remaining card is dealt?

There are 52 cards in the deck, 13 of each suit.  So, there are 52-6=46 cards unseen, and 13-4=9 spades remaining.  9/46 gives just under a 20% chance that you will complete your flush: you will get your card one time in five.

Do you bet on that?

3)  Before you decide, look at the pot.  What is the size of the pot (say, $60 on the table) vs. the amount that it costs to to have a chance of winning it (a $10 bet)?  The ratio is the Pot Odds: in this case 60/10=6.  So, if you won only one time in six you would make money.

There are, of course, tables for this:

Now, would you bet?  If you have a one in five chance of getting your card, and only need to win once in every six hands, yes.

The (over)simple, unifying rule:

When hand odds beat pot odds, you bet.

The epiphany:  Portfolio Entrepreneurship

80% of medical device ideas go bust.  So the odds of any single venture succeeding, the hand odds for the startup if you will, are one in five.

But the most you can lose is the money you invest, and the upside return can be tremendous.   So as long as you can stay in the game, and the returns (pot odds) are greater than five times your investment, you will make money in the long run.

I’ve thought a lot about this in the last couple of weeks, since hearing about the general idea on Radiolab.  It explains why 10x or 30x returns are mandatory or investors, and how they can tolerate the majority of businesses failing.   It also addresses a question that I’ve had for some time: is it better to be a serial entrepreneur, with only one business at a time, or a portfolio entrepreneur, running  more than one?

There is no advantage either way (the odds are the same in either case) if there is independence among the businesses.

But if a prior failure diminishes the chance of finding funding, or confidence in running the second business, then the interaction makes the second more difficult for serial entrepreneurs.

It almost makes more sense to take on several independent businesses at the same time, running them as a portfolio, to get maximum chance of success in the shortest time.

“Non-interacting” is still a key word: there are only so many hours in the day and limited total energy that any one entrepreneur can put in.  Many investors argue that anything less than total focus and commitment greatly diminishes the chances of success.

But with proper partitioning and prioritizing, good partners and delegation, and adequate funding, I think that there are advantages to portfolio entrepreneurship.  The pot odds increase the overall chances for success, independent to hand odds.  And the accelerated learning, synergistic ideas, common networks, and shared services can increase the opportunity and efficiency of the whole.

And this isn’t abstract.  A recent study concludes that portfolio entrepreneurs have more diverse experiences, and more resources than serial or novice entrepreneurs. On average, portfolio entrepreneurs appear to offer more attractive growth prospects than other entrepreneurs.

Investors routinely spread themselves to decrease risk.

Entrepreneurs can spread themselves to improve the odds.

The value of clinical trials

What is supposed to be an objective scientific process designed to assess whether product work, whether they are better than products already on the market and whether their side effects are a price worth paying for any benefits they might convey s routinely being corrupted.  Companies bury clinical trials which show bad results and publish only those that show a benefit.  Trials are often run on small numbers of unrepresentative patients, and the statistical analyses are massaged to give as rosy a picture as possible. Entire clinical trials are run not as trials at all, but as under-the-counter advertising campaigns. That is a scandal.

Economist, reviewing Ben Goldacre’s new book.

This was my jumping-point for a talk on Clinical Trials Strategy that I gave to an industry audience at Medtec Ireland this week.

My first point was a straightforward solution: Clinical trials must inform and conform.  To inform, trials must be designed to answer a significant question honestly, with the possibility of a negative result.  To conform, trials must be executed using hypothesis-driven scientific methods, following rules to assure patient safety and privacy.

My second point was strategic: Recognize that clinical trial results serve many other audiences.  Publication in peer reviewed literature is obvious.  But regulators, payers, professional clinical associations, industry risk / usability standards,  and investors also have particular needs for information and evidence.  If the planning process doesn’t take account of them, then trials may need to be repeated.  As our Business Development director reminds me, 10% more investment broadening a trial's data collection strategy returns 30x savings later when the right information is available as needed.

I’m in Washington DC now for the Anesthesiology Society of America meeting. We’re presenting five abstracts (here’s one) describing the results of clinical studies that demonstrate safety and baseline characteristics of a new surgical monitoring device in normal subjects. 

The first two day’s experiences has really driven home both points.

These are High Science presentations: hypothesis-driven, rigorous methodology, direct comparison to the  commercial gold standard, formal registration, peer-reviewed publication.  These studies establish facts and promote open discussion, first through formal presentation and then through hours of informal walk-by discussions with clinicians. Our next set of results, accepted for December, will move to clinical experiences describing efficacy and usability.

But more broadly, the presentations also connected with multiple audiences.  We brought a prototype of the monitoring device that illustrated the likely instantiation of the idea.  Doctor who understood the need and the evidence could comment directly on the product.   They told friends; friends talked but it on the trade floor.  Industry rep’s dropped through to inquire and to  fret.  We were featured in n article in the ASA Daily News (p.3).

We hoped for a few hours of comment by our medical advisory group.  We got very full 12-hour days of intense discussion by a diverse audience of people from academics, clinical practice, and industry.  The prototype got a month of shakeout in days.

To be honest, the momentum built so fast that I got a bit worried.  Was it being oversold?  Could we meet expectations for performance and delivery time?  Was awareness building too quickly among competitors?  Were we losing control of the process?

There’s no question of the value that connecting with customers through clinical trials  has brought to the process of developing the product and the business.   Context, credibility, awareness, and access have all accelerated; optimism and enthusiasm have infused the founding team.  It’s driving late night discussions over nibbled tapas and glowing tablets about planning and funding.  It’s exciting.

Are we losing sight of the simple maxims that I advocated in Ireland just days before?  I think it’s really fortunate that I presented on these issues immediately before the conference: so far I think we’re wholly on the side of the angels and headed in directions that will both inform and conform. 

In directions that should, ultimately, benefit patients and the business.

Launch strategies

Early work on our surgical monitoring project has gone swimmingly.  We completed the prototype amplifiers in February in Maastricht, and took it to Seattle for first tests.  Ten subjects were successfully recorded and we analyzed the data and prepared five scientific abstracts which have been submitted for presentation at the Anesthesiology Society meeting in October.  That’s five weeks from proto to papers: absolutely unheard-of execution in medical device development.  (Yes, I’m proud of the engineers and clinical sites.)

The expectation of first public disclosure in October revives an old debate: how long should  we keep the product secret?

It affects a lot of things: who we talk with for advice, when to use a confidentiality, whether to do a market survey.  Clearly it’s a bad idea to go public before filing patents, or to emerge only when ramping up manufacturing.  Apple has been a master at building buzz ahead of a new product launch (the lines in New York for the new iPad, above, are typical), but their model isn’t transferrable to a company without a reputation.

I was fiddling with how various companies do their product announcements and sketched this idea.  Early-stage companies, lower left, have no anticipation in the market, yet must keep plans secret to avoid having better-resourced competitors jump in ahead of them.  Once in the market, a successful product will build interest: it’s good to let customers know that version 2.0 is in the wings and solicit feedback about how to make it the best it can be.  Once you are established in the market, like Microsoft or Apple, then a dribble of news can feed anticipation, so secrecy again becomes desirable to avoid cannibalizing sales.

For us, it seems driven by two factors.

First, we anticipate that an established company could create a competitive product in six months once they set their mind to it.  That means that we have a six month window from launch to gain as much share as we can.

We also have to raise funds and build buzz in anticipation of being ready to manufacture so that we can fill orders and gain experience as quickly as possible, building value and momentum.

So, if market entry is expected in January, then starting to build awareness (ascending the Know/ Like, Trust curve) has to happen at least three months before starting with public announcements at the scientific meetings in October.  Limited disclosures will start earlier, composing a Board and approaching through leaders in late summer.  Then everything needs to be in place to mount a campaign ahead of launch, so brochures and summaries will need to be ready in November.

All of which gets built into the staffing and budget estimates: the goal is to maximize sales and share in that crucial six-month window.

‘not an easy problem to match publicity and secrecy.  If we succeed, though, they’ll be lined up around the block.

Peer reviewing

One of the foundations of the scientific method is publication of results so that they can be assessed and reproduced by the wider community of researchers.  Peer review is the first step in that publication process, voluntary, anonymous review of work by experts in the field, with comments returned to the author so that the paper can be improved prior to publication.

I serve as a peer reviewer for several journals and probably get requests about twice a month to comment on a paper.  Generally it takes about an hour to understand the work, to assess whether it is correct, clear, open to alternative interpretation,and properly positioned within the relevant landscape of background literature. I try to write balanced comments, praising a clear hypothesis, a well-reasoned discussion, while perhaps suggesting where figures could be more clear or an alternative method might yield better results.  The revised paper comes back the next month and I generally pass it if the author has replied to my points (whether or not those result in changes).

Occasionally I run into situations where I struggle.  I received one where the author had copied an earlier paper, word for word, yet not cited it. I talked with my editor (the lead author was fairly senior) and we recommended that he include the citation: it was enough to alert him that we’d found the earlier work.   Another was proposing pages of dense calculation for an obscure visualization technique that, in the end,m didn’t reveal anything new in the data.  I think that, in six revisions, we did everything but rewrite the paper for him.

Last month, I received a work that was analyzing structure in short-and long-term heart rate changes.  It’s a tricky analysis, since it’s statistical rather than deterministic, and the author had calculated some crude ratios to support his hypothesis.  Worse, his underlying technique destroyed the very ordering among data points that he based his hypothesis on.  I pointed out that the results weren’t supported by the method, suggested some alternative approaches fo analysis, and worked out an example to illustrate my point.  Throughout I was constructive and respectful (I though), allowing that I might have missed his point, but didn’t follow the logic.

The authors absolutely blistered me in their reply.

…there are many comments made by the Reviewers with which we disagree, comments which we consider unhelpful and comments which are just wrong…we strongly believe that the Reviewers should be aware of the fact that, even though they are anonymous, they should put time and effort in reading and reviewing the papers. We are frequent reviewers for a number of peer-reviewed journals,and we always try to helpful for the Authors and never write from the position of someone who, by definition, knows better.

Most of my comments were subsequently dismissed with this is an opinion and there is nothing we can do about it, except state that we disagree.  Fair enough: I went back through my analysis and still cannot link their conclusions with their data using their method.  I had to just recuse myself after that: while we might have solved the disagreements over a beer and a whiteboard, their reaction was so strong and their comments so dismissive, that there just weren’t actually any explanations to discuss.

I felt badly about the comments and about the outcome: I do take the job seriously and give every paper a good read.  I know how much work it takes to do research and to write it up.  In the end, if they are right, they should be able to refute the counter argument and to say why.  If they are wrong, they need to make corrections.

And, fundamentally, I disagree with them that Reviewers are there to serve the Authors. 

Reviewers are present to serve Readers.

If, as a first reader, I think that a paper is wrong, I have a responsibility to (nicely) say so, suggest why, and present alternatives.  Maybe it’s a real mistake, maybe I just don’t understand, maybe the Methods contain omissions or ambiguities.  But honest and differing perspectives should be addressed openly.

We might argue that every author is entitled to his view and the Readers will sort things out. Eventually, yes.   But errors enter indexed literature, then pop up on searches, news reports, and blogs.

The Letters to the Editor don’t have much impact, and are presented as a debate that obscures rather than resolves issues.  It can take years before the truth is sorted by Readers, years of contradictions and misinterpretations in which true facts have been undermined and genuine debates get muddied. 

The argument presented by these Authors, that Reviewers should only check for clarity of language and labels on graphs, is wrong.  The peer review process is intended to engage experts as surrogate Readers, not as supplemental Editors.

Research management

Before I became an expat, I was a research director with Physio-Control, a Redmond WA company, the world leader in resuscitation techno0logy.  It was a great job, we had 20 talented people, a $3 million annual budget, and a dozen projects and studies ongoing that each had potential to directly save lives.  It was a great job.

At the time, though, I didn’t pay enough attention to the context of the job.  I was good at managing projects and people but missed the context of my department: someone was paying the bills and expected something in return.  I thought that performance alone mad us immune to budget cuts and office politics; I was wrong.

I met the current research director for dinner shortly after I returned to Seattle this week, and we talked about managing scientists and expectations. 

It seems to me that there are two dimensions involved:  the balance between creating insight and technologies (ideas vs. innovations) and the balance between internal and external focus (on Customers or on the Corporation).  In classic b-school partitioning, this creates four styles of research management. 

For those with an idea orientation, research produces some combination of white papers for the marketing folks, bolstering the company’s offerings through focused studies, or conducting blue-sky investigations alongside (or in place of) university labs.  

If innovation is more the style, then you create some combination of components for project teams or try to start entirely new businesses.

Intrinsically, all are valid: I leaned towards creating startups, the current director is a solid idea guy: both work.

But politically, the department can’t survive if the model doesn’t align with senior management’s conception of what research is for. 

Despite lip service about organic growth and open innovation, the company didn’t really want an intrapreneurial group challenging established projects.  In the end, I was out and the group was disbanded in favor of a more consultative model.  The current director plans the politics better than I ever did, and the group; has greater stability and is able to take on longer-term projects.

It’s not just about positioning and performance.  Awareness and alignment also matter.

What kind of business vehicle?

A Board Meeting is coming up on Monday, and its making me think hard about business vehicles.

<------------

 

No, not that sort.   It would be nice to trade up from the Fiesta, but they’d never go for it…

No, the vehicle I’m thinking about looks a bit more like this

------------>

Back in 2006, my academic advisor and I were debating the principles of new business formation. 

Classically, you start with a market need, a product or service that can fill it, a business model that can profit from it, buttressed with good people and abundant funding.

As an exercise, though, what if you built the structure ahead of the product?  Put together a process for transforming undeveloped ideas into refined ones, add people with good generic skills in a few key project and financial management roles.  Qualify a good list of supporting service providers, from legal and regulatory through laboratory and fabrication.  Then find something to apply it to.

In short, build the vehicle before loading the cargo.

It’s a lot like building a house on spec: take undeveloped land, manage a project to put a house on it, sell it onward.

But there’s a danger to that analogy too: market conditions.  What if the process succeeds, but the buyers aren’t there?

I suppose, having built the house, we could live in it, rather than let it lie fallow or sell it at a loss. 

Similarly, having  created a prototype product, validated it, and run it through regulatory approval so that it’s ready for market, we could simply start selling it.

Take that a step farther: should we be making revenue shipments?

And this is the core of the thinking – maybe the whole idea of building a great product, then giving it to someone else, is flawed.

It’s all about what the business should be when it grows up:

A. A vehicle, creating and selling a portfolio of assets needed to make the product: it’s formula, test results, CE Mark, and patents.  License revenues are reinvested into building the next spin-out prototype.

B. A company, manufacturing a product to sell through distributors.  Profits are reinvested in market development and product improvement.

Either way, it’s a sustainable business, one run as a process irrespective of the product, the other focused on product without leveraging the process.

I’ve been thinking about how the business would develop under each scenario.

We’d need more people and plant for Plan B, producing higher revenues but lower percentage profits. 

We’d be more flexible and profitable under Plan A, but its risky to depend on repeated success of early stage projects. 

Plan B creates jobs and tangible assets to float on a stock exchange; Plan A is a partnership that deals in intellectual property.

I haven’t figured this out yet.  We definitely validated the model, creating a successful vehicle.  The question now is whether to drive it or trade it in.

Among the boffins

I knew that this would be a different evening on walking into the anteroom.  For once everyone was my age.  Or even older.   ‘Shaking a few hands and making introductions, I found the second anomaly: everyone in the room had actually built something.

The circuit of networking events in London and Cambridge tend to follow a type and pattern. Groups such as the MIT Forum or the Pitch ‘n Mix are a blend of well-heeled investors and young hopefuls, talented, but all with something prove or to sell.  Social evenings become speed-dating events, looking for the right person who believes and accepts you for who you say you are.

This group had a more clubby feel, people who wanted a relaxed venue to share stories and ideas.  There were people who pioneered digital television systems, mobile telephone software, early computer operating systems, and text parsing systems.  The speaker, a principal from the Technology partnership, came prepared with a dozen slides recounting the pre-history of technology companies in Cambridge.  He apologized that his remarks would only take fifteen minutes.

For the next three hours, everyone chimed in with their reminiscences of people and events, amending and embellishing one another’s stories.  There were folks who met the founders of Aero Research, had used PYE wireless products, and knew the  Marshall family story.  The Sinclair people squared off with the Acorn people.  There was a debate about whether Cambridge Consultants spun out or kicked out Domino, the folks who invented ink-jet technology. 

And the funny glyph, above, is  the original logo of the Cambridge Instrument Company, founded by Darwin’s youngest son.  Engineers will understand the Cam centered in the Bridge circuit.

It was a delightful evening.

CHASE holds meetings monthly, and a compendium of stories is being built up for a book: the Cambridge Phenomenon.

And, separately, LinkedIn did a deep dive into their membership roster to see who the Entrepreneurs were.  The results are shown below – only 5% are over 50 !?  Yike…whatever became of all the boffins?

In my element

I have always lived in the future.

When I was young, I read Tom Swift books and imagined the things I could invent, the adventures I could have.  I avidly followed ‘hard’ science fiction, especially Clarke and Wells.  In elementary and high school, I took on a special science curriculum in fourth grade (Optimum conditions for growing brine shrimp as a prelude to establishing a salt water aquarium), assisted in the computer lab in high school (a PDP-8/S that I programmed in FOCAL), and hacked the school’s timeshare system in the Guidance office.  I fell out of love with Physics and into affection with medical signals (and the campus radio station) in college, then on for advanced engineering and medical degrees.  ‘never much good with hardware, I had a knack for programming signal interpretation, and that defined my passions and career path for decades to come.

I visited the London Science Museum last Tuesday – it was coming home for me. Where I go through gardens, bird refuges, and natural history exhibits, I fall out of the ‘naming games’ that the British are so good at.  But in halls of machines, I could name them all, knew their purpose, principles, and inventors.  I was among familiar friends.

The purpose of the visit was to hear Ray Kurzweil, an inventor and futurist who invented scanners, reading machines, music synthesizers.  Lately, he’s been promoting the idea that all technologies grow exponentially, and that we would soon be facing a singular moment when machines become smarter than humans.  At that point, they would become capable of advancing their own evolution, with uncertain consequences for humans left behind.  Philosophers debate the consequences, engineers speculate how machines can be harnessed, but it seems clear to me that human dominance and self-determination would be at an end.

The talk itself was a bit of a let-down: Prof Kurzweil has created a movie promoting his ideas and the MIT Forum meeting turned out to be a screening, preceded by some introductory remarks.  I had to leave half-way through to catch the late train back to Cambridge, but absorbed the self-promotional thrust before leaving.  I would have preferred to hear his ideas instead of his story, our future instead of his past.

I hurried past the rows of silent machines: future vision can always be turned towards Cyberpunk-horror or Trek-utopianism.  The debate has been going on for centuries, yet man always finds his way past the supposed end-times, yet never arriving at Ecotopia.  I’m  optimistic: machines ultimately serve human ends.

Yet, what if this time was different?

I’m reminded of another MIT scientist Edward Fredkin.  I heard him speak in a radio interview, starting with a simple question.  Throw a ball: imagine it stopped at an instant of time, mid—flight. Advance a moment, and the ball moves ahead in a straight line.  Why does it do that?

To Friedkin, the ball must somehow remember where it has been to know where to go next.  Or, perhaps, the space the ball is in remembers how it entered and has rules for where it should go next.  Perhaps, then, the whole of space and time is made up of cells with programmed rules, much like John Conway’s Game of Life.

According to Friedkin, the universe is a computer, unwinding the future step by step according to local rules of “digital physics”.  It’s a fascinating notion, yet one that comes up again and again.  Thermodynamics, for example, has intimate links to information theory via entropy.

If  space and time are quantized and the laws of physics reflect  local rules coded into that cellular automata, then information is more fundamental than matter or energy.  The physical universe and consciousness could simply be a solipsistic epiphenomena of this underling reality

So, Matrix-like, the Singularity could be the latest evolutionary refinement in how the Universe conceives Itself.

I smiled, headed toward the all-too-physical Tube with all-too-little Time.  It would have made a better talk: be courageous with the idea that computing is fundamental and see where it takes you into the Future.

Creating better circles

Or, as the British Ministry of Defense calls them, Multi-Source Intelligence Working Environment.

The problem is, they aren’t working very well, and there’s hard money on offer if you can figure out a way to make things better.

I came across this opportunity while looking for a way to get a clinical study funded.  The Ministry of Defense gets lots of data, information from satellites and intercepts, intelligence from informants and spies.  The problem is how to make sense of it all, connecting the dots.

Grant money is being offered for methods to understand how teams collaboratively interact and reason over information, make decisions and derive intelligence.

There is a need for innovative, multi-disciplinary research on team sense-making concepts and cognitive processes, together with applications and services which offer support to those processes. We seek novel concepts and solutions that make the most of emerging technology and/or optimise combinations of existing technologies. In particular, we would welcome multi-disciplinary proposals from outside the computer science domain (including but not limited to psychology, sociology, business intelligence, mathematics and statistics) where there is an opportunity for solutions, methodologies or theories from other areas to be used for intelligence collaboration for the first time.

I think this is a fascinating opportunity: if you can think of a way for people to work together more effectively, then you can get up to £50,000 to spend 100 days demonstrating the feasibility of your idea.  The details can be found here, and proposals are due by January 25.

Genius doesn’t work alone, so part of the solution may be in how groups are composed.  Chris Anderson recently wrote that innovative groups (crowds, tribes) trend-spotters, evangelists, networkers, and skeptics in addition to creative innovators.  There must be a links, ways for each member’s work to be visible to the others, and there must be motivating rewards (perhaps only recognition within the group).

I remember a team-building exercise where we were to select items from a plane crash that would be useful in the arctic.  They scored our choices against what an arctic expert would choose.  Then they asked us to make a new list, deciding as a group.  How would the consensual answer compare to the average of individual scores?

The group did worse than the best of it’s individuals, and slightly worse than the individual average.  It bothered me that the best ideas were not the ones that got heard and adopted, or that the group couldn’t pick the best ideas from each of its members.

Every day, in many ways, we fail to make the most of people’s knowledge and insights when we try to solve a problem as a group.  If you’ve got a good idea for tools, methods, or organizations that can improve the situation, the MoD wants to hear from you.

A failure of imagination

When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is probably wrong.

When I first began going to scientific conferences to present my graduate work, I could always count on questions from grumpy older scientists.

These would be distinguished folks who remembered a similar paper from 40 years ago, a case report that disproved my thesis, an alternative hypothesis that they hatched over sherry with another grumpy elder.  It’s a tricky confrontation, because I can’t ignore them, can’t take them head-on, and often can’t understand what they’ve mumbled. Over the years, I’ve perfected a two—step deflection of a beaming embrace of the idea, followed by an invitation to discuss it over coffee.

I’d note, however, that this technique did not work when my thesis advisor jumped up during my Defense to erase a few equations and propose an alternate idea in front of my Examining Committee.

Now that I’m birch-bark blond and over-50, I’m trying not to fall prey to the disease (or at least to be constructive when I’m tempted).

Still, I genuinely do see old ideas come around again, and again, at scientific conferences.  In the coffee sessions this week, a few of us wondered why.  The easy answer is that these examples are genuinely good ideas that solve significant problems, and are thus discovered anew by each generation.  Alternatively,

Technology lagged the concept: wireless data visualization had to wait for 3G cell phones.

Technology led the science: patients could be cooled long before we were sure it did any good.

It takes time for an idea to find it’s audience.  Conventional wisdom holds that a medical device can get to idea to market within five years.  But it takes at least another five for the funding-study-publication cycle to generate a good literature base from multiple authors. Adoption into Guidelines and reimbursement codes could take another five.  During that decade–long process from idea to standard of care, lots of other people will start down the same path independently.

But the most frustrating reason is a simple one: Failure of Imagination.

I know so many instances where good ideas, working in prototype, delighting clinicians, were not pursued because senior managers couldn’t back the project.  Lack of understanding (of the innovation or the market), competition for resources, ‘making the tough decisions’ in favor of safe alternatives all contributed to the problem.  But it’s corrosive: in the worst cases, I’ve lost not just the project, but talent as the whole team faltered, then departed.

At worst, management keeps a research group of bright people only as an operating reserve, never intending productive output.  ‘Let’ them pursue personal projects in the off-periods, then throw them at problems in times of need.

I have, in retrospect, left jobs over chronic Failure of Imagination, never for anything else.  It just makes life impossible, no matter what the mission statement says.

The only way of discovering the limits of the possible is to venture a little way past them into the impossible.

quotes from Arthur Clarke, Profile of the Future.

Configuring an idea(l) conference

I was talking with friends the other night about what makes a good research conference.  There was a time when it was a big conference, filled with important people and companies all showing off their latest findings and innovations.  It was a distant city, a nice hotel, good dinners, and visits by the corporate brass.

It does seem like a tender age.

These days, I really enjoy a smaller gathering, one where there’s a singe track of presentations, a limited audience, immersion in a topic, close discussions of creative ideas, promises to follow up on collaborative projects.  It’s more of a retreat, things to see and discover together, shared meals, an excursion or two.

The big conferences have their place, but I really don’t learn as much or get time to talk about ideas with others.

I’m headed up to Ireland for a few days for such an event, ‘looking forward to the time away (even though I have to finish a grant application during the evenings).  The organizers have a good sense of how to assemble a group: they get a good mix of young and older investigators, people actually doing work, and mix the theoretical/technical sorts with the practicing/clinical ones.  There are a few repeat members, a few industry folks, a core focus for everyone to put their heads around.

It reminds me of the times running a research group, a lot of the same considerations lead to a lot of the same solutions.  You want to gather together a group of smart folks with different perspectives who can work together around a common cause.  Flatten the hierarchy, celebrate successes at every level, and keep the networks humming.

When it works, it’s really one of life’s delights.