Yes, you guessed it right.. we are talking about the Andy Grove who left an indelible legacy through his role at the Intel corporation (together with the other two founders Robert Noyce and Gordon Moore) including as CEO for a decade, fueling the growth of Silicon Valley and the world at large. Now we got that cleared, we need to introduce some background on what is called the ‘Andy Grove Fallacy’ in a commentary in 2007.
In his old age Andy Grove suffered from prostrate cancer and Parkinson’s disease. In 2007 he gave an interview (‘A Research Revolution‘ published in Newsweek – the link to the interview is broken, so points to a reproduction by someone in Quora™ ) comparing pharma and semiconductor industry , with his opinions on what the pharma can possibly learn and put to practice from the semiconductor industry. Many columnists and bloggers from the pharma and life sciences industry slammed Grove for the contents of the interview and his views at that time. A well-known pharma columnist Derek Lowe, who is (was?) a medicinal chemist and had worked in pharma, picked on that interview and slammed Grove mercilessly in the commentary ‘Andy Grove: Rich, Famous, Smart and Wrong‘. Lowe lamented how clueless Grove is and how difficult life sciences and developing drugs are and characterized that whole interview and contents as ‘Andy Grove Fallacy’. During that process he went deep into how we do not know the blue print of humans and, in contrast, knowing that makes semiconductor industry so much easier.
Some others claimed that Grove’s grouse is personal having suffered from prostrate cancer and Parkinson’s disease. That may be true to some extent, but there are at least three other reasons for near uniform condemnation of the contents of that interview. One of them is the content of this perspective, the other is examples used by Grove (referred to later in this perspective), and finally his provocation of management practices of pharma. Many also took offense on Grove’s statement on peer-review and grant system in academia – here is a perspective piece where we discussed that topic with solid examples in the past: ‘Why I am indifferent to the effect of sequestration on research funding’.
One line of reasoning that I found appalling with Lowe’s commentary is the statement “That’s partly because we didn’t build them. Making the things from the ground up is a real advantage when it comes to understanding them, but we started studying life after it had a few billion years head start.”. Life started few billion years ago, which naturally makes one infer that life as we study exists in a setting that is far older than that, governed by the principles that form engineering systems and how the world operates at the macro level.
To set the right perspective, I am a mechanical engineer turned biotechnologist now combining those and more to improve life sciences and healthcare as a starting point (modest goals, eh..). I have seen and worked with people from different disciplines in their native environment doing great things as well as stumbling through with unnecessary difficulties as they operate, as well as follow, learn and predict the trajectory of where advances from these sectors are headed and implications.
We (human beings from different disciplines) figured out principles (e.g., laws of physics) and built more engineering systems including the semiconductor chips of Intel, starting mostly with unknowns. We didn’t always know the principles, properties, environment effects, or even environments etc. in aircraft, bridge design or vehicles and gadgets for space odysseys to be able to test them at every combination of all variables. There is one key difference that the others who consider Grove’s premise a fallacy forget or fail to agree: a fundamental difference in these disciplines is that engineering evolved as very quantitative field with principles being incorporated as they are figured out, and that got ingrained into the training and practice. Whereas the fields of life sciences and healthcare though did collect lot of data (who can forget Gregor Mendel’s systematic collection and deductions that originated the field of genetics that paved way for most drug development enterprises and many other aspects of healthcare) and do so in enormously higher magnitude now, the discipline evolved with a different mindset elegantly and elaborately exposed in the articles by Lowe (and nearly all the 86 comments). Two unrelated information may be of interest to note at this juncture: (1) even Mendel’s work that paved way to genetics as we know, was parodied by R. A. Fisher, one of the most famous statisticians as ‘the Mendelian Paradox‘ that many others of his time agreed with, and (2) oh.. by the way, it is true that the examples Andy Grove used were not correct which is partially a reason that most people involved in developing for or practicing healthcare would brush off his core tenets. Below, I lay out other reasons why Lowe and others are so defensive.
If the best answer one can come up with is ‘start your own company and prove me wrong’ – there are a few major flaws you overlook: (1) if a discipline has evolved in a particular way that is not conducive to a more ‘engineering like’ approach, it does take a lot of leg work to even up the differences**, (2) people are trained to think like how Lowe elegantly wrote ignoring the previous fact, and (3) further (inadvertently) comments in that article demonstrate uniformity of thought process reflecting an inherent built-in push-back in the system.
**Very relatable examples are the ongoing recognition and correction for insufficient women and minority in top positions and lack of equality from many perspectives, including opportunities and pay. My view there is, despite we trying to close the gap by rapidly doing some damage control in visible ways, true change will likely come in the next generation where youngsters are trained with this current disparity in perspective.
It was a dozen years ago when Andy Grove made these statements that Lowe reacted vociferously to label that a ‘fallacy’, why rekindle that now? More recently (November, 2018) , Vijay Pande an investor at the venture capital firm Andreesen and Horowitz (a.k.a. A16z) elaborated in an article (‘How to Engineer Biology‘) and a podcast how life sciences and healthcare operates in a bespoke manner and ‘after all’ the common components are building blocks like amino acids, nucleotides etc. and at the next level cells, organs etc. and these components are like legos to build from, aiming to promote more ‘engineering like’ practice in life science (biotech/pharma) enterprises. Pande comes from academia – was a Professor focusing on computational medicinal chemistry and structural biology – at the highly reputed Stanford University. Well, Lowe being a passionate follower of the field as he now is a sector columnist and reporter revived the strong views expressed in his previous article where he categorized Grove’s interview statements as ‘Andy Grove Fallacy’, slammed Pande in a commentary entitled ‘Engineering Biology, For Real?‘ and another blogger Keith Robison a computational biologist with pharma drug discovery experience even titled his article as ‘No, the Grove’s Fallacy Can’t be Retired Yet‘. Robison states ‘Apollo knew the majority of the challenges before the program even started. It was possible to have a very stepwise program because almost all of the steps were known! ‘. However, if one evaluates objectively they were meticulously assembled by thought and trial as we do in biology. Lowe’s passion for chemistry and looking at the world through his knowledge/experiences, and defense of them can be exemplified from the recent blog he wrote ‘The Good Stuff Goes One Way..’, where for example he raises serious doubts on a recent publication with defensive statements like ‘…and you’d put the rest of us out of business’. Ashutosh Jogalekar, another well-known life sciences blogger, with professional interest in medicinal and computational chemistry, wrote a blog in a ‘bit more’ forgiving tone ‘Can we turn biology into engineering?‘, however he relegated all the tech as enabling, than going to be integral part of life sciences and healthcare. Jogalekar used examples of aircraft, bridges, spacecraft that we introduced above and somehow concluded that ‘engineering systems lack non-linearity, redundancy and emergence’ hence not comparable in complexity. However, I disagree that engineering problems (the same examples we use here) lack non-linearity, redundancy and emergence in the strict sense. If the problem being solved in engineering does not already have those properties, they are often built-into or tested under those conditions, depending on what the intended use is. Often they appear easier because the field evolved to tackle problems using approaches and practices that now make them seem simpler, intuitive and more amenable to be modeled/automated/scaled etc.
This perspective is meant for broader audience to stimulate alternative thoughts and discussion, hence we cannot delve too deep. Since Lowe, Robison, Jogalekar put forward nearly similar views for the purposes of this perspective, we will refer to them as LRJ in the rest of this piece.
While Pande’s intention and broad goals in that commentary and podcast is very commendable and a needed change, he did simplify the problem too much too early. Including due to some of the above-mentioned reasons we are far from that level of modular build a drug development pipeline from basic blocks such as amino acids, nucleotides, and at the next level as cells and organs etc. for life sciences and healthcare. Hence, LRJ were not fully wrong, though still largely conservative and too defensive. So if both these camps are neither correct, nor incorrect and Andy Grove did not leave a fallacy for this sector (you would’ve inferred that is the position that I am trying to put forth for thought by now), what does this perspective intend to do by contradicting itself?
I stated my background earlier as an engineer and a biologist. Those times I used to attend lot of technical seminars and work presentations/reports. In their own (paraphrased) statements an engineer working on problems in biological domain would consider me one of them (an engineer) in biologists’ clothes and vice-versa with biologist. That was often based on the incisiveness of questions I pose that are answers to questions that they are studying. In that sense, I am a process logician having traversed many aspects of life sciences, healthcare and engineering.
My take on this is life sciences and healthcare can adopt far more ‘engineering like’ approach than it currently does and LRJ and the likes are willing to even allow discussion on. This is based on my thoroughly grounded understanding that principles of biological processes are often conserved and design principles reused, but not conserved at the component level. Therefore it is certainly not possible the way Pande outlines, hence LRJ lament that what works in mouse (or simpler model systems) doesn’t in human. By the way, we already made a case when we cursorily discussed some example engineered systems (airplanes, bridges, spacecraft etc.) that we do not need to know every aspect of a system or what it would face to design a new one or one that works for a purpose and evolve as we go. Enough simulation of scenarios combined with experimentation with the previous statement in perspective and careful choice of path(s) forward makes us faster, efficient and innovative. That should certainly help us adopt ‘engineering like’ approaches and ‘Silicon Valley principles of business operation’ – but not to the level technology company building clichés like ‘build and fail fast’. Remember, we do not have to build or understand organisms bottom up to the complexity of humans to find drugs and treatments (as Lowe points passionately to Grove that the whole plan of Intel chips is known but not of humans and other living organisms), especially in a commercial setting. Let us ignore the part where Lowe claims that the blueprint of Intel chips and many complex engineering systems in comparison to biological systems are equivalent ingredient labels of a shampoo bottle.
Though we had remarkably complex medtech devices like MRI, CT scans etc. for a long time, the level of transformation we are discussing here is made easier by influx of lot of newer high-volume and sophisticated measurement technologies into life sciences and healthcare which in turn got more people from fields ingrained with ‘engineering practices’ of operation and thought. My observation is – currently, these practitioners from other fields often succeed when life sciences/healthcare people revere them for what they themselves do not know ‘and’ when those who come in from other fields ‘fit in’ to the often rigid hierarchy of the system purely for the access they gain. Depending on factors like the extent of power they wield and support of successful people they worked with, and lots of luck (including facilitated by the previous two factors) they move the system often incrementally disproportionate to the effort they invest. These are currently temporary (but necessary gains) and real change will happen as I stated for gender/racial equality, the change has to come bottom up.
- Do I say life sciences or healthcare are less complex than Lowe does? No.
- Is engineering fundamentally less complex than the life science and healthcare counterparts? I don’t think so either.
- Is life sciences and healthcare lot more complex than engineering disciplines that LRJ and many other practitioners claim? Yes, but my observation says that is significantly so due to the way these disciplines evolved and that can be corrected.
Yes there are fundamental aspects of all these fields and there are experts who are motivated purely to solve such complexities without thought of industrial applications, otherwise we wouldn’t have people who do fundamental research. Such fundamental contributions often become the building block principles and pave way for many advances in both engineering and life science sectors. How much of these belong to industry is a key question – with centers for eminence in fundamental research promoting more cross-pollination that may be less of a concern.
In the context of this perspective, the latter part of a Grove’s famous quote: ‘only the paranoid survive’ is not a good thing – as this paranoia is about experts inside a bubble who at-large are unanimous in the opinion that their field is too complex for bottom-up ‘engineering approach’ to be of significance, and consider that the notion itself is belittling – in part due to their training in a field that evolved differently. Paraphrasing and summarizing the words of currently successful people in life sciences and healthcare, artisanal nature of their specialty is in no way ready to be significantly changed by adopting ‘engineering practices’ as an essential part of their discipline – the context discussed in this perspective. Though the current level of knowledge is not sufficient to get for-profit sector practice the level of modularity (lego like) that Pande proposes, the state of knowledge and the number of cross-disciplinary people should allow the sector to take a huge leap in that direction. I propose a middle ground based on my experiences (acknowledging the sector specific unknowns): viz., based on processes, makes adopting ‘engineering practices’ possible in biology driven fields (e.g., pharma, healthcare). This will be with the condition that we not just rely on currently known and popular processes but also apply to how we find others, knowing that (i) the objective function measures and readouts are representations of many different cross-reactive process, and (ii) that the components of these processes are more often not identical across systems.
In summary: Andy Grove’s fundamental suggestions in that interview are not a fallacy for pharma, his examples were wrong. Let us jump-start to embrace the new transformative future where ‘engineering like practices’ are considered as integral part (not just as potential enabler and tools) of life sciences and healthcare enterprise. This is similar to what we are doing with gender and racial equality. That way we will have tangible short-term gains and also transform our enterprises and culture geared towards the next generation where significantly more principles of life sciences and healthcare appear as simple and fluent as we see in engineering systems today.