What do Microsoft and medical research have in common?

Let’s start with the simple fact that they have both been wildly successful. Microsoft has 79,000 employees, global annual revenue in the year 2007 exceeding $51 billion, and has made more than 12,000 of its employees millionaires simply by increases in stock valuation. Additionally, Microsoft’s market share for the operating systems on desktop computers, by one 2003 estimate, is 90%. Not bad for a garage startup.

In the United States alone spending on medical research is at or near the $100 billion mark in the year 2007. That’s about $300 for every man, woman, and child and more than doubles what was spent just a decade ago. According to Dan Fox, president of the Milbank Memorial Fund, a philanthropic group that works on health policy issues, the data in a recent JAMA review makes it plain that “we are spending huge amounts of money, more than any other country, to develop new drugs and devices and other treatments.”

Not bad for a profession that used to do double duty as barbers.

But perhaps the strongest connection between the multinational and the medical profession may be that both are, in some very key respects, failures.

Anyone who has ever seen the dreaded “Blue Screen of Death” or had their computer infiltrated by spyware, adware, or other kinds of malware knows the shortcomings of Microsoft software. Maybe Microsoft is an easy target for the technorati, too easy it would seem. But let’s face it– Microsoft has over the past few decades consistently churned out software which costs more, has more bugs, and hogs more memory than comparable software. Right up to the present, new software releases have been slow in coming and devoid of groundbreaking features. Certain releases, namely Windows ME and Vista– some versions of which retail in excess of $350– have inarguably added very little if anything to previous releases. One prominent commentator, John C. Dvorak, in his colorfully titled column, “Vista Death Watch” has argued that the latest operating system, Windows Vista should be completely scrapped.

Modern medicine, specifically medical research, may be in the same boat. Let’s concede that we all are very pleased that medicine has advanced to the point that a burst appendix can be safely removed, catheters can spring open blood vessels, and that eliminating wrinkles surgically is just a series of injections away– and with a toxin no less!

When it comes to chronic illness, at least in its traditional form, medicine has largely failed. According to mainstream medicine, what causes Parkinson’s Disease? What causes cardiovascular disease? What about fibromyalgia? Lyme Disease? Crohn’s, cancer, bipolar disease, depression? The consensus among medical researchers has been, “We don’t know.”

Now, maybe these conditions are all too complicated to unravel over the course of mere decades even for a world that spends hundreds of billions on medical research annually. For that matter, maybe it’s the case that having stable software largely impervious to malware is no less a pipe dream.

Even if these explanations are mostly true, the absence of real solutions for chronic illness is not good enough– not for the hundreds of millions of patients watching the best part of their lives pass them by. For all the sound and fury of medical discovery we catch wind of these days in breathless press releases (Eureka Alert!), what chronic diseases have been definitively cured in the last few decades? You can count them on one hand: stomach ulcers and cervical cancer.

Few would argue that medical researchers aren’t highly intelligent and capable or that Microsoft’s software engineers are the same, or that both groups contain people who are quite possibly the most intelligent and the most capable at what they do. If a failure of smartness isn’t the problem, then what is?

This essay examines the Open Source Software (OSS) movement as a model for a more evolved, more effective system of medical research. We look at four lessons from the OSS movement and try to apply them to medical research. Also, we look at how one treatment protocol for chronic disease, the Marshall Protocol, incorporates many of these lessons and may represent a model for future medical research in more ways than one.

Lesson #1: Use open review.

One of the most influential works in the history of open source development is Eric S. Raymond’s, The Cathedral and the Bazaar. In this essay Raymond argues that there are two contrasting software development models– the cathedral and the bazaar. In the cathedral model, represented by Microsoft, software is available for all to use, but the ability to contribute to the code is restricted to an exclusive group of developers– wizards Raymond calls them– who work in “splendid isolation.” In the bazaar model, the code for a project is developed over the Internet in full public view, Linux and Wikipedia being excellent examples. In theory, anyone so inclined can contribute.

The cathedral model holds that the key to success of a software project is the judgment and experience of a limited group of well-vetted and talented experts. Most proprietary software is necessarily developed using the cathedral model. In contrast, the bazaar model says, “Let’s engage as many developers as are inclined and qualified to participate.” The theory here is that the mistakes and red herrings of inexperienced or confused contributors is more than counterbalanced by an ability to rapidly improve the code given the fact that, as Raymond puts it, “thousands of eager co-developers are pounding on every single new release.”

Which model is more effective in software development? Raymond explains his thinking: “I worked hard… at trying to understand why the Linux world not only didn’t fly apart in confusion but seemed to go from strength to strength at a speed barely imaginable to cathedral-builders.” For Raymond, the advantage of transparency and openness is simple: “Given enough eyeballs,” he writes, “all bugs are shallow.”

The track record of a number of large-scale projects like the production of an operating system such as Windows XP seems to support Raymond’s opinion. The day Microsoft released Windows XP, the company posted 18 megabytes of patches on its website: bug fixes, compatibility updates, and enhancements. Two patches fixed important security problems. Or actually, one of them did; the other patch didn’t work. Microsoft suggested that users back up critical files prior to installing the patches. Buyers of the home version of Windows XP, however, discovered that the system provided no way to restore these backup files if things went awry. As Microsoft’s online Knowledge Base blandly explained, the special backup floppy disks created by Windows XP Home “do not work with Windows XP Home.”

Though it’s certainly not the case with all closed or proprietary software projects, really big systems are forever in danger of the “tar pit” as Fred Brooks called it in his book The Mythical Man-Month. A tar pit– I would argue that that phrase is as apt a description as any to describe the state of research into chronic disease.

Is medical research a bazaar or a cathedral? At least when it comes to chronic illness, I think it’s pretty clear that it’s the latter– the reason being that the Academy is just not open enough to opposing ideas. The results, or lack thereof, speak for themselves.

Not only are pulmonologists unwilling to consider infection as the cause of sarcoidosis, but many have actively suppressed publication of articles that are at odds with their points of view.One need look no further than the field of pulmonology. Not only are pulmonologists unwilling to consider infection as the cause of sarcoidosis, but many have actively suppressed publication of articles that are at odds with their points of view. Biomedical researcher Trevor Marshall, PhD, head of Autoimmunity Research Foundation explains, “Dr. Om Sharma from USC Medical Center has kept the lid on bacterial pathogenesis when dozens of researchers have implicated pathogens in the disease.” Marshall says, “Dr. Sharma has discouraged discussion about patients who received organs from people with sarcoidosis also developed the disease – evidence which strongly supports an infectious cause.”

As is the case with a lot of these “experts,” Dr. Sharma has authored or co-authored over 500 journal articles, some 70 reviews and sits on the editorial board of all the key pulmonology journals. It’s this last line on Dr. Sharma’s resume that has allowed him to effectively act as a gatekeeper for new ideas about sarcoidosis. Yet, even now, there is growing evidence that Dr. Sharma is mistaken in the worst way. What provision does medical research have for reining in the myopia of the Dr. Sharmas of the world? Is it really acceptable for medicine to advance but one funeral at a time?

Or, take the work of Dr. Alan Cantwell who has photographed L-form bacteria in the cells of people with cancer for decades now. Dr. Cantwell’s research into the so-called “cancer microbe” is strong. He makes a number of claims worth further investigation or, at the very least, outright rebuttal. Yet, the medical community has largely marginalized his work such that it hasn’t even attracted a fraction of the attention it deserves.

According to Dr. Cantwell, most pathologists who saw his specimens would practically refuse to admit that what they were seeing was bacteria. Few, if any, were willing to co-author a paper to that effect. “For the most part,” Cantwell said, “they didn’t want to get involved.”

Why the cold shoulder from all corners of medicine? Is it because we have figured out the “true cause” of cancer? Well, that’s not the case. Though we certainly know how diet and heredity can effectively change the odds, we still have only the faintest understanding of what actually causes cancer. Or, is it true that it’s simply impossible that bacteria can exist in the size and shape described by Cantwell? No, there’s no reason to think that it’s not bacteria, specifically L-form bacteria, to blame. Entire textbooks, namely Lida Mattman’s Cell Wall Deficient Forms– Stealth Pathogens, now in its third edition, have been devoted to the culture and life cycle of L-form bacteria.

Dr. Cantwell was asked to explain his colleagues’ refusal to entertain the idea that bacteria may be involved in cancer. He responded, “I think it’s because finding bacteria in illnesses that are not [thought to be] attributed to infection is highly controversial, and most doctors shy away from controversy.” But if that’s what it takes to open up the dialogue, isn’t controversy exactly what sufferers of chronic disease need?

If you paid attention to your bulk mail announcing fundraisers, you’d think that the only reason diabetes or sarcoidosis or any number of other diseases have not been cured is a lack of funds for research. This, in spite of the fact that the National Institute of Health alone spends $28 billion annually on medical research.

You could spend ten times as much money on medical research and without changes to the process of how ideas get considered, that money would be better used as wallpaper.You could spend ten times as much money on medical research and without changes to the process of how ideas get considered, that money would be better used as wallpaper. One of the enduring lessons of open source software development is that the best way to tame complexity is to open up the discussion to as many people and ideas as possible.

Andrew S. Grove is founder of the computer chip maker Intel. Now Grove is suffering from Parkinson’s Disease and a medical system that continues to test drugs on rats in the laboratory but offers little, if any, medical options for human beings who suffer from Parkinson’s and other chronic diseases. Recently, Grove has been publicly critical of the biomedical establishment, particularly academic researchers, who while successful in getting NIH grants and publishing research papers, seem to have little regard for whether their work leads to ideas or treatments that can actually cure disease.

“The peer review system in grant making and in academic advancement has the major disadvantage of creating conformity of thoughts and values,” Grove told a correspondent from Newsweek. “It’s a modern equivalent of a Middle Ages guild, where you have to sing a particular way to get grants, promotions and tenure. The pressure to conform [to prevailing ideas of what causes diseases and how best to find treatments for them] means you lose the people who want to get up and go in a different direction. There is no place for the wild ducks. The result is more sameness and less innovation. What we need is a cultural revolution in the research community, academic and non-academic.”

It should be a no-brainer: medical research, particularly when it comes to chronic disease, ought to be more like a bazaar. You could start with peer review. Peer reviewers are often well-intentioned, but the fact that only two or three anonymous “experts” can act as gatekeepers for what gets published in prestigious journals severely limits the dialogue. Let’s allow increasing numbers of self-selected people– amateurs, volunteers, marginalized scientists– to critique work too. Let’s then publish all reviews, positive and negative, online. To the extent that we can make medical research less like a guild, let’s do that too. Too often grants are awarded on the basis of past success in winning a grant. Do we really want to continue to reward old ideas? Not when those old ideas are failures! Is that any way to foster new ideas and new approaches? After all, until a researcher succeeds in curing a chronic disease, he is neither an authority nor an expert.

Lesson #2: Write software for yourself.

In 2005, BMG Music Entertainment started selling CDs that came with copy protection software. Unbeknownst to users, this software contained an evil bit of code called a rootkit. Rootkits are a set of programs or instructions that effectively subvert control of an operating system from its legitimate users. They might allow a third party to install keyloggers monitoring bank account information or turn your computer into a zombie and have it send out thousands of spam email messages. According to security researcher Dan Kaminsky more than half a million machines were infected by the rootkit. Since then, Sony BMG has been the subject of numerous lawsuits and eventually had to settle in a national class action suit.

What would drive Sony BMG to include rootkits with their CDs to say nothing of notoriously restrictive copy protection software? It all has to do with the bottom line. The rootkit was Sony’s heavy-handed way of preventing users from copying songs they had purchased. For the multinational conglomerate, the bottom line was protecting the profitability of their music artists even if it meant trampling consumers’ rights.

But, if you’re an open source software developer, your bottom line is straightforward as can be: solving your own technical problems. Eric S. Raymond writes in The Cathedral and the Bazaar, “Every good work of software starts by scratching a developer’s personal itch…. too often software developers spend their days grinding away for pay at programs they neither need nor love.”

What if everyone researching Crohn’s Disease had Crohn’s Disease? What if all those investigating bipolar had that illness? How would research change?The open source model has been successful time and again.AWStats, the excellent log analysis program, was written by developers who needed to know the makeup and characteristics of visitors to their websites. VLC Media Player was written by programmers who wanted to play back a range of audio and video codecs unrestricted in the way that Windows Media Player or even iTunes is. It’s laughable to think that any open source developer would include anything on par with Sony BMG’s 2005 fiasco.

So, what if everyone who studied a chronic disease had that chronic disease? What if everyone researching Crohn’s Disease had Crohn’s Disease? What if all those investigating bipolar had that illness? How would research change?

To paraphrase Samuel Johnson, nothing focuses the mind so much as a trip to the gallows. In sum, I think you would see a greater sense of urgency, a better openness to new ideas, and an increased willingness to do away with ideas that have gone long past their expiration date (more on this last one later).

The remarkable thing about being a researcher of chronic disease is that you can spend entire careers without actually contributing anything useful to the understanding of it.

It’s not the fault of researchers, necessarily. It’s the way these professionals are rewarded for their contributions. As it stands, developing and investigating bonafide cures for chronically sick patients can all too easily take a back seat to getting published, getting funded, and getting promoted– “time, tender, and tenure,” Thomas Goetz calls it. As it stands, you are more likely to get funded if you have gotten funded. As it stands, you are encouraged to publish research which has positive results, even with the most tepid, toothless conclusions, because positive results get published.

If there was the least bit of urgency, you might see the serious reconsideration of a variety of treatments, starting with the use of corticosteroids for sarcoidosis. The 2003 ACCESS study, which was funded by the NIH, shows that two-thirds of sarcoidosis patients do not get better on current treatments and those that do see improvement are in no way correlated to whether or not they received treatment.

Given all this, you would think researchers would be interested in any broad reconsideration of what causes sarcoidosis. One such idea that sarcoidosis is actually an infection has been posed by Trevor Marshall among others. In fact, since 2002 when the Marshall Protocol treatment was first published– open source style, online– many patients have claimed recovery on the antibiotic-based treatment. Strangely, the response from sarcoidosis doctors, primarily pulmonologists, has been lukewarm at best. “We’re not even on their radar,” Marshall has said of the pulmo doctors.

You would think that a new idea such as this one, which was backed by molecular data, would receive less of a chilly reception. You would think that this is exactly the kind of big idea sarcoidosis researchers should be looking for. If it was their own skin at stake maybe they would.

Lesson #3: Openly and honestly size up failure.

When it comes to chronic disease, there is no failure more categorical, more disastrous than the autoimmune model of illness. Autoimmunity is the idea that– contrary to the forces of natural selection or the dictates of basic common sense– the immune system *somehow* goes haywire and attacks the body.

One way to measure the worth of a theory in medicine is to gauge how that theory can functionally improve what should be everyone’s ultimate bottom line: patients’ health. In short, all the anti-inflammatory, immunosuppressive, and palliative medications given to those suffering from “autoimmune illnesses” are failures. If I were being paid by the word, I could list hundreds of first line anti-inflammatory, immunosuppressive drugs that don’t cure hundreds of chronic diseases. Corticosteroids do not cure sarcoidosis. Neither hydroxychloroquine nor cyclosporine cures lupus. Prednisone does not cure multiple sclerosis.

Of this class of drugs meant to treat autoimmune illness, prednisone is typical. Patients taking prednisone complain of weight gain along with any number of other symptoms including itching, increased sweating, irregular or absent menstrual periods, inappropriate happiness, and irregular heartbeat– and that’s just symptoms beginning with the letter I!

First proposed in more general terms by Paul Ehrlich more than 100 years ago (horror autotoxicus he called it), this theory of autoimmunity has had a lot of time to be tested and refined. Instead, further investigations have only rendered the concept of autoimmunity more flawed and more complicated. Wikipedia lists five theories about what causes autoimmunity: the clonal deletion theory, the clonal anergy theory, the idiotype network theory, the clonal ignorance theory, and the suppressor population theory. What do all these mean?… Does it matter? The point is, as an idea, autoimmunity is a failure.

A healthy sense of proportion– some would call it succumbing to realism– is much more common in the realm of programming especially of the open source variety. In fact, one of the colorful but enduring proverbs of the programming world, seen in a variety of forms is, “All software sucks.”

Ron Avitzur has said during the development of his groundbreaking piece of software called Graphing Calculator that if your software was exceptional, fellow engineers would simply say, “This sucks less.”

Maybe if the medical research community could be a bit more philosophical about failure, certain ideas that have gone long past their expiration, namely autoimmunity, could have been unceremoniously discarded long ago. Some would argue that any kind of operating system Microsoft produces is doomed to be underwhelming given its tens of millions of lines of “legacy code,” and maybe that’s not such a bad analogy for the concept of autoimmunity as they both share any number of fundamental flaws.

Maybe if the medical research community could be a bit more philosophical about failure, certain ideas that have gone long past their expiration, namely autoimmunity, could have been unceremoniously discarded long ago.The problem for medical research is that it has made little to no provision for failure. With certain notable exceptions (i.e. Journal of Negative Results in Biomedicine) journals tend not to publish negative results, a phenomenon known as publication bias. The media tend not to cover studies showing the absence of a connection.

If a negative result is based on good science, why confine it to obscurity? Why would we want to, in essence, encourage researchers to produce positive results even if those results make no significant contribution? You get what you ask for, and our system is, in essence, asking for such results.

One thing medicine might quickly discover if, as a whole, it better embraced failure is just how ineffectual old approaches to chronic disease have been.

Lesson #4: Seek to rapidly deploy multiple versions of software by iterating incrementally and openly.

While Andrew S. Grove was at Intel, the number of transistors on a chip has gone from about 1,000 to almost 10 billion. Over that same period, the standard treatment for any number of chronic diseases has remained unchanged: L-dopa is still the standard treatment for Parkinson’s just as it was nearly half a century ago.

“I picked the semiconductor [transistor] industry because it’s the one I know; I spent 40 years in it, during which it became the foundation for all of electronics. It has done a bunch of unbelievable things, powering computers of increasing power and speed,” says Grove.

So, twelve years ago, when Grove was diagnosed with prostate cancer he immediately jumped into the advocacy movement, with the expectation that quick, deliberate action could push scientists in the field of cancer research in the same way that a drive for innovation and knowledge had inspired his team at Intel. But nothing happened. “I got disappointed with the lack of real output,” says Grove. “Not much has changed 12 years later.”

In the last 20 years, there are only two common chronic illnesses for which the cause has been established. I’m thinking here about cervical cancer, caused by the human papillomavirus, and ulcers also found to be caused by bacteria. Is this pace acceptable?

In January of 2007, Scott Rosenberg wrote a piece about the debacle that is Microsoft Vista. First began in 2002, the development of Vista was beset by a series of problems. Microsoft developers were ultimately unable to execute the original grand vision, which included a complete file system overhaul. As a result, the new operating system came in much later than scheduled and lacked many of the originally promised features. Rosenberg’s analysis: without discipline “too often, software teams get lost in what are known in the field as ‘boil-the-ocean’ projects — vast schemes to improve everything at once.”

Where did Microsoft go wrong? Some of the failures of process are touched upon elsewhere in this essay, but I would suggest here one of those problems was an inability to iteratively build upon their project by harnessing the power of their natural community, their end users.

Those who follow software development are familiar with the idea of “perpetual beta,” or that certain software is always being added to and improved upon. Popular examples of software that have had the label of perpetual beta include Gmail, Google Maps, and del.icio.us. For many developers, “perpetual beta” is an attitude as much as anything. It starts with the assumption that software should be better. Changes are rapid, but incremental. Testing and assessment are done frequently.

Even those of us eager for cures to chronic disease recognize that magic bullets do not magically arise as if willed to exist. They need to be worked on, worked over, the subject of productive consideration. Dialogue is the order of the day.

Jean-Claude Bradley practices something he calls open notebook science, which he describes as making his online laboratory notebook freely available and indexed by common search engines. The underlying philosophy, Bradley says, is as pithy as it is bold: “No inside information.” On his UsefulChem blog he and his collaborators share and discuss chemistry problems and on his UsefulChem wikiobservations from experiments in excess of 150 are described and logged.

What do third parties do with this data? Maybe nothing. But then again, perhaps a failed experiment or an unanswered question– the programming equivalent of a feature request made before the open source community– could spark progress. It could be the key to new advances and new understanding.

Bill Hooker on the excellent 3QuarksDaily blog writes that “From our current perspective we cannot predict more than a fraction of the ways in which openness will transform the culture and practice of science.” The advent of collaborative science and the openness that drives it has the potential to help us refine our understanding of medicine at increasing speed.

We cannot predict more than a fraction of the ways in which openness will transform the culture and practice of science.A prime example of this openness is the world of genomics. Right now, anyone anywhere can search NCBI’s online gateway Entrez and dig into any number of mRNA, genomic DNA and protein sequences using some thirty interconnected and all freely available databases. It’s easy to see how these kinds of resources made freely available can really allow researchers to rapidly and iteratively build knowledge.

Even in the field of chronic disease, we have seen promising developments where clinical data is posted online and made freely available for inspection. I’m speaking here about the Marshall Protocol site. As of December of 2007, the MP site has 5,000 members and 125,000 posts, the vast majority of which come in the form of self-reported statements of progress. When it comes to the sheer scale of open clinical experiments, the MP may be in a class of its own.

Patients suffering from a variety of chronic diseases including fibromyalgia, sarcoidosis, multiple sclerosis, even autism are on the Marshall Protocol, a novel antibiotic-based treatment for chronic diseases which is in some respects the original inspiration for this essay. Using instructions that can be downloaded from the web, patients work in conjunction with their prescribing physicians to put the Protocol into practice: avoiding vitamin D and taking low-dose pulsed antibiotics as well as Benicar, a Vitamin D Receptor agonist.

While the science behind the Protocol hasn’t changed much since 2002 when it was first made available to patients, the site is designed so that each patient reports symptoms and reactions to medications in a weekly progress report entry. Not only are these progress reports open to other patients who can ask questions about data, or on occasion offer advice, but they are also open to analysis by the FDA (who actively monitor the site) or any other health care agency, think tank or organization who wish view the data. Any doctor, scientist, or medical researcher can become a member of a special forum for medical professionals, which not only allows them to view progress reports, but gives them a place to actively engage in discussions about specific cases and share ideas about treatment and research.

Another indication of this study site’s uniqueness is the chance to get feedback on one’s progress from nurses and researchers including, often enough, from the originator of the Protocol, Trevor Marshall. In fact, this is very similar to the open source software community where end users can be in direct contact with the developers from whom they downloaded the source code.

This is in marked contrast to most other treatments where a patient’s reactions usually do not go much farther than the doctor’s office. Under such circumstances, each separate case history ends up in a different place, almost always at a metaphorical dead end. Such reports go hidden from the eyes of the public, other researchers, and medical professionals who might under many circumstances offer important feedback or learn something.

With open clinical data, such as the kind on display atwww.MarshallProtocol.com, patients can discover a thing or two from researchers themselves but also from each other. How will I react to Benicar? Have patients who share my condition really experienced the telltale reaction to antibiotics? These are all questions that can be effectively and convincingly answered by fellow patients. Such insights make a tough treatment that much more manageable, and it’s the openness that allows it.

Who can use this study site? The science behind the MP has broad application, so a range of patients with different conditions can participate. Almost no one is turned away. This is quite different from other clinical trials in which participants are required to fit specific inclusion criteria with the aim of accumulating the most homogeneous group possible.

Had the Marshall Protocol followed that model, how much longer would it have taken to confirm that the Marshall Protocol works to treat other diseases besides sarcoidosis? Surely, since Marshall’s first model was based on sarcoidosis, any other clinical trial model would have carefully excluded patients with other diseases, ailments, and symptoms. But it was by allowing patients with a wide array of symptoms to join his open internet-based site that Marshall soon collected data demonstrating that symptoms other than those related to sarcoidosis respond to Benicar and antibiotic therapy – a reality that would have taken decades for a standard clinical trial to reveal.

A parting thought

If the research medicine were anything like the open source software movement, study sites like Marshall Protocol.com would be springing up left and right as other researchers jumped on the possibility that their data might also have a greater impact if freely available on the Internet. You might also see a greater number of bad ideas relegated to historical curiosity.

Medical research today has all the turning radius of an oil tanker. As a whole, it is too slow to accept and instigate positive change, and any sensible person would be right to lay a good chunk of that blame on the process itself.

Innovation: how can researchers better innovate in the field of chronic disease? I argue that when it comes to innovation, the open source software community is particularly inspired– and instructive.

For those of us who truly care to fight chronic disease, what lessons can we learn? Release early and often. Iterate. Practice openness to the point of promiscuity. Collaborate. Embrace failure. No failed or failing system has to be as it is. We can be worlds more productive. Let’s get to work.