This is a chapter from a book Practical Precision Livestock Farming published by Wageneingen in 2022 in which I wrote about my experience of trying to commercialise my inventions both as a government scientist and commercial entrepreneur.
I describe four models for commercialisation and my experience with each
1: License the technology to an established company
2: Open source intellectual property
3 Institutions can create a spin-out
4: Start-up your own development company
The world economy and much of our daily lives is dominated by tech developed by companies (Microsoft, Facebook, Apple, Google, ARM, Amazon) that were founded in garages by enthusiastic amateurs within the career span of this author. These are companies that scaled to prominence quickly through internet distribution of software and electronic devices. The teleological explanation is that their growth was inevitable because of the brilliance of the idea and how they monetised that idea in their offering to the customer. But for every company that is a success on a large scale there will have been many small scale failures often with equally brilliant ideas. Ninety percent of all start-ups do not last a year or even produce a minimum viable product (MVP). In addition, there are many successes that scale up rapidly to huge valuations but wither and die, as either they bet on a particular technology platform and are overtaken by events. Some very large tech companies have been prominent for a few years and then withered away (AltaVista, MySpace, MapQuest, Netscape, Palm Pilot, Blackberry). The world of agritech startups is unlikely to nurture the huge scaling that seems inevitable from looking at the history of Microsoft, Google, Amazon, Facebook, Zoom and their like. But that scaling has created an expectation that tech companies are a guarantee of future wealth for the investora and founders. This is sadly not the case.
In the 1970s I was a naïve hippy and self sufficient farmer wannabe. There was a serious back to the land movement of young people in those days who had been inspired by Rachel Carson’s Silent Spring and the Club of Rome report about how pollution was killing the natural resilience of ecosystems and that we were running out of reseources. Self sufficient farming in the old style was all the rage with an emphasis or organic as the answer to all problems. I took myself off to Galloway in SouthWest Scotland where the farming was less mechanised to learn the skills of old style farming that I thought I needed to be a self sufficient peasant. All those old skills that change with the seasons on a mixed farm growing forage crops for animals and selling milk and livestock are brilliantly described by James Rebanks in English Pastoral (2020). So I learnt to drive a tractor and plough (badly), sowing, reaping and binding. I preferred the livestock skills, lambing hill sheep, shearing sheep, calving cows and milking dairy cows in a warm byre. Years later when I entered the world of research and development the muscle memory of those skills stood me in good stead. Just knowing how much physically embedded knowledge is built into the experienced stockperson was very useful when it came to assessing the utility of yet another clever idea dreamt up in an academic environment.
The other thing I learnt was the sheer arduousness of physical labour in farming even in the semi-mechanised agriculture of the 1970s and eighties. I could see the steady disappearance of labour as people who had grown up as “farm servants” aged and retired. Throughout Europe people have largely rejected the hard life of a self-sufficient peasant and prefer to buy food grown and packed elsewhere on farms of increasing size and technical sophistication. Supermarket shelves are always full with packets of tempting looking meals which are so much quicker and easier to cook than garden vegetables and local produce. That trend seems to be spreading throughout the world with an accompanying demographic shift reducing the number of humans available to work on farms. Every so often keen idealistic people go back to the land but they rarely sustain the enthusiasm for more that a couple of seasons of hobby food production and the economic viability of small scale organic farming is limited unless sustained by a retail operation.
At the end of the seventies I moved with my young family back to my home area in South West England and took to contract milking cows, as that gave me a good income from which I could save towards buying and renting out property to participate in the house price inflation that enriched property owners through the 1980s. It also gave me flexible time to study. I realised that the hard physical labour in farming had to be replaced by automation and overcame my early neglect of education by studying for a degree with the Open University which was largely a set of engineering courses with a couple of modules in agriculture. After graduating I briefly lectured in agriculture in Cornwall before moving to Silsoe College to undertake a Masters Degree in Agricultural Machinery Engineering. I had got a taste for the research life and went on to do a staff registered PhD in the automation of milking.
In 1989 my wife and I were able to buy our own place – a ruined pig farm with hardly any land but planning permission for a temporary mobile home. Over a number of years we built up a successful goat farm selling milk to a creamery making specialist cheese. Every weekend I was milking goats, trimming feet and building more space for the growing herd. Eventually we showed it was a viable business and we were able to get permission and build our dream house. At that stage with our children leaving home we were confronted with the challenge of the empty nest and a lack of successors wanting to take on the legacy of farming. I meet a lot of reluctant farmers who have taken over the family farm out of a sense of family duty. Some never really get to run the farm as Dad is keeping an eye on things and controlling the operations. We decided to sell the farm and I took up a position as a Principal Scientific Officer at Silsoe in 1999.
In parallel I was working for Silsoe Research Institute as project leader in automation of milking systems and I led the team that built the UK’s first robotic milking system barn at Cheseridge Farm, Institute for Animal Health to house the technology developed by Mike Street and his team. This robot was later commercialised as the de Laval VMS which celebrated its 20th year on the market in 2017. When we started the project we made a list of 26 reasons why a robot might not successfully milk a cow and steadily worked through an experimental program to show that things like cow behaviour, morphology, cleanliness and temperament would not be a barrier to robotic milking and so it proved. It became very clear that cows do not volunteer to be milked they volunteer to eat and all robotic milking systems are defined by some form of traffic management so that as cows go through their day looking for snacks they have to pass through a station where they are usually fed.
In the 1990’s life in agricultural research and development in the UK was a struggle with the continuous need to find new sources of money. Following the free market ideology of Margaret Thatcher, governments withdrew from supporting experimental husbandry farms and practical farm research. The argument ran that as a rich country that made its money in manufacturing and financial services, the UK could purchase food on the world market and that supporting agriculture was for social benefit rather than food production. The practical development work of new systems of farming was deemed to be the role of industry as it was “near market”. The Agriculture and Food Research Council was renamed Biological and Biological Sciences Research Council, funding was to be aimed at “pure research” measured by papers in prestigious scientific journals rather than demonstrable improvements in farming techniques. This policy was also backed by a desire to patent our “blue-sky” inventions and licence the IP to any company willing to pay the price for it. This policy gained some credibility when in 1995 the government agency responsible (British Technology Group PLC) for our patent portfolio in the robotic milking sector announced the development deal between de Laval and Silsoe Research Institute to commercialise the robotic milking portfolio as the VMS system which has been on sale since 1997. Which leads me to my first commercialisation project.
- Model 1: License the technology to an established company
The primary model of technology development is that clever academics invent things, patents are filed and the institution then sells the patent portfolio to a company to a company to take it to market.
However the model of academic institutions licensing IP to third parties has serious flaws which were exposed as events later proved in the case of robotic milking.
The first flaw is that to defend a portfolio one needs to be able to maintain continuous surveillance of the field of invention. One needs to be able to comment on the patent filings made by the institution and more importantly identify infringements in the competitive filings made by others. The patent attorneys need the advice of the inventors to critically assess the key features to be protected from a wish list that might be put into the first filing. It is common practice to draft vague first filings that describe the problem to be solved and all the many ways in which it might be solved. After one year you add or edit the patent claims to focus on the novel and protectable claims, sometimes in response to the patent search by the assessors. As the project develops one or more of the key features gets developed as the solution, the problem comes if one finds a different way of solving the problem which was not mentioned in the first filing. Patenting is seen as similar in value to writing scientific papers but in practice they are completely different documents with completely different purposes. The patenting process takes several years during which the original filing will have acquired “claims” emphasising the key features and sometimes “divisionals” expanding on something from the original. The patent search needs to be commented on, occasionally the patent office find documents which they rate as having an impact on the invention, these all have to be dealt with for years. In addition, competitive filings will keep appearing to cover gaps which the prior art has missed. Without a team working on new developments it is hard to keep knowledge up to date.
The typical academic research programme lasts three to five years, even before the end of which, the team with their embedded expertise in the field start looking for other projects to engage their intellects and develop their careers. There is no guarantee of continuity of a team with the unwritten domain knowledge with which to respond to claims and counterclaims in a patent battle.
The two main competitors in robotic milking filed maybe 150 patents relevant in the field and spent a great deal of money and time before a single robot had been sold. Each one needed to be assessed and commented on but most of the Silsoe team had found other jobs encouraged by the continuing reduction in government support for agricultural engineering research.
BTG had gained a reputation for aggressively defending its IP portfolio and certainly invested heavily in paying patent attorneys to file our inventions. They used to come regularly and ask what the latest ideas were to patent. As a privatised company BTG had to take commercial decisions and in the late 1990s it decided not to sustain activity in agricultural engineering and abandoned any attempt to fight for the licence income that could have derived from the patent portfolio we had established.
Technology developments take decades to come to fruition but Research Institutions have not proved stable over long periods in the UK as they have had to adapt to constantly changing demands of the politics of managing decline from an Imperial past. The latest example being Brexit which is destroying many academic collaborations built up through European funding programs.
- Model 2: Open source intellectual property
Academic institutions funded by endowment or government programmes do have a certain luxury of being able to set long term objectives. Some of these are strategic for example, in keeping a capability to conduct research in a strategic area such as climate change or vaccine development. The academic model is based on combining research and teaching. Historically a learned scholar published and taught and drew a coterie of research students around him or her. Scholars banded together into colleges and began to provide intellectual support to powerful person and institutions. During the twentieth century the university became formalised and massively expanded and an emphasis was placed on publishing papers preferably in the most prestigious journals. Prestige generally goes with the pure sciences focused on theoretical studies sometimes backed by experimental data. The constant dilemma between the ability to teach and the ability to publish papers is not one to concern us here.
When I first joined the Silsoe Research Institute in 1989 my head of division pointed out to me the criteria for advancement up the pay scale was based on the numbers of papers I would publish. At the grade I was appointed to I should produce two refereed papers a year and progress to the next grade within five years. Similar criteria exist all over the academic world. The aim was quantity rather than quality as there were few measures of quality although impact factors based on readership and prestige began to creep in. Gaming the scoring system was an inevitability especially when institutional funding became related to papers written and there was even a transfer value for staff who moved between institutions. However, papers in areas where there was a potential for commercial exploitation could be embargoed until patents and other IP had been protected. Much of the work to develop robotic milking was being comprehensively patented by the BTG group and the expense did not accrue to the Institute itself.
However, after about 1997 BTG decided that agricultural innovation was not to be a major commercial benefit and later explicitly focused on the life sciences and medical research. In future, the costs of filing and protecting patents was to be the responsibility of the Institute and in this respect was to resemble academic institutions such MIT, Stanford and Cambridge and Imperial College. After the sale of the robotic milking portfolio to de Laval I had undertaken commissions for our Ministry of Agriculture to begin to identify those aspects of dairy cow health and fertility that were essential to enable robotic and highly automated systems to match the performance of good husbandry. There was no budget for filing patents and I heard it expressed that work funded by the government was for the public good wherever that was expressed. This was a continuation of a long standing British tendency to invent things and then fail to commercialise the invention. So we embarked on a policy of publishing all our work which enabled us to advance our careers and broadcast all our secrets.
It worked well until the money ran out. For example, I published my work to develop inline methods of monitoring metabolic disease and fertility parameters (Mottram, 2000) and later saw this work developed by others with a more commercial mind set. At that stage our funding model for science was being turned upside down with the reforming of the Agricultural Food Research Council as the Biological and BioScience Research Council (BBSRC). I had personally decided to focus on monitoring anaimal health and this change did lead to some useful collaborative projects with Universities. However, the change also reduced the funding for what was seen as near market research as the governing council was dominated b academics who had made their careers in fundamental research. The dirty business of making amazing scientific breakthroughs actually work in field conditions was definitely not a priority. The mindset was that of the magic bullet, genetic modification to breed better plants and animals, amazing biochemical treatments that would prevent pests and diseases. At the same time funding from the Ministry was being withdrawn from practical agriculture and replaced by levy board funding. The levy boards were generally chaired by older farmers whose mindset did not really focus of the future. In one shocking incident millions of dairy research funding was re-allocated to funding generic milk advertising. This meant that on the one hand we had funding bodies that saw success as being scientific papers and on the other by farmers who thought we had plenty of technology and just needed to use it. By 2005 it became obvious that the Research Council saw no future for agricultural engineering research in the UK and looked for ways to dispose of the assets which led to the dispersal of the institution teams to various universities. A few of us chose the alternative which was to work in industry and even to commercialise inventions that had potential markets.
A classic problem for genuine conceptual inventiveness is that it may be difficult to patent an original as it describes something that cannot yet be implemented. A classic example is the paper published in 1945 by the sci-fi author Arthur C Clarke that proposed that geostationary satellites could provide global wireless communication. This was at a time when no rockets existed capable of putting an vehicle into a low orbit let alone a geostationary position and the invention of the transistor that has made small lightweight radio systems possible was still in the future. If Clarke had been allowed to patent the idea it would have had no financial value for another fifty years until commercial launches of satellite communications began. Patents expire after twenty years.
Even when the breakthrough is visible and patented, most development programs will eat up years of the patent filing and this is particularly the case in the biosciences where any product that can be used in animals and humans may need extensive testing and validation. Most pharmaceutical products get only a short period before going out of patent. And worse still the details of the invention need to be disclosed in the patent, testing and validation so there can be no reliance on secrecy. Deciding when to apply for a patent is a critical decision that has to take into account what the competition is likely to be doing, how easy it will be to copy and how long it will take to get it to market. Getting it wrong can have consequences years and decades after the event.
One major benefit of open publishing is that it does prevent others from patenting one’s ideas although it is sometimes important to ensure that patent assessors are aware of the prior art. That is easier in the internet age but search terms can be misleading. One tactic we have used in my companies is to draft and file patent applications which is relatively low cost (£130 in the UK) and see whether the patent assessors find any prior art. The patent can then be allowed to lapse at no further cost, allowing anyone to use the idea but no one can prevent others from developing it further. Sometimes the key patent is in a feature that makes for a simpler, cheaper device. Others then have to find another way of solving the problem.
This method might be called open source intellectual property.
- Model 3 Institutions can create a spin-out
A fashionable idea in academic institutions is to encourage staff members with potentially commercial ideas to found companies and licence the IP generated by the academic research to the start-up or spin-out company. There are some very successful examples of this (BioVex, a University College London spin-off, was sold to Amgen, for up to US$1 billion) but also a very high failure rate.
There is a whole bureaucracy of Technology Transfer Officers (TTOs) within institutions trying to encourage and manage this process. There are also many grants and fellowships available to researchers who wish to follow this route and these often have useful training packages attached.
In the closing years of Silsoe I had been asked to develop a rumen telemetry system to monitor the efficacy of new pharmaceuticals being developed by the Pfizer company. This was technically challenging at the time but led to a viable if rather clunky system. The Institute then set up a company which we called Well Cow Ltd. The senior management took control and nominated a few shares for the founding staff.
There are a number of reasons why academic spinouts seldom succeed in attaining genuine commercial status with products on sale generating revenue.
The first problem is that TTOs and senior academic staff like to stay in control using the naivety of junior staff to accept their leadership and with the bulk of shares retained by the institution. This almost inevitably leads to a lack of speed and motivation to drive the business forward. Unless the institution has been working closely with customers for the new process or technology it is unlikely that an academic team will have the contacts and experience to develop a marketing plan quickly. It is hard to do market research unless you already have marketing experience. There is a cultural divide here too with scientists often disdaining the ways of speaking and thinking that are common in commercial circles.
As a start up moves from research to development it needs more resources and the price of money and the role of existing shareholders becomes a major consideration. A typical scaling of resource would be initial research 1, Development 10, Pre-production and launch 50. So if a supervised PhD costs £100k the next phase will cost £ 1 m and raising that much money has a price. The ideal would be to borrow the money at loan interest rates (say 5-10%) but no commercial lender would consider this unless there is collateral such as a guarantee from the institution or a founder with a property. Not many post-docs have a house worth £1m! So inevitably equity investors will be sought and they will want a massive return on their investment. Suddenly the management has to accept a dilution of the shares and manage expectations of shareholder returns. This requires the company to sell a dream rather than the reality which is that the investors are buying a gambling chip on an unknown roulette wheel which is what the market for a hitherto unknown product.
When I was first developing the idea of a biosensor to measure progesterone in milk in the 1990s. I was basing the science on the voltametric blood glucose disposal sensors. These sensors used were being produced in millions using lateral flow chemistry to convert glucose to glucose oxidase (GOD in the trade) and the release of electrons as a measurand. Our sensors used a similar lateral flow principle and voltametry and when I described the process to investors their response was often. “Could this work on blood glucose?” The reason they gave for asking this was that they could calculate a return on capital as the market for blood glucose monitoring and its growth was well known. Gaining a percentage share of that was something they could evaluate. Evaluating a market share in a market that doesn’t yet exist is impossible.
The skills needed to sell a dream whilst managing a transfer of activity from research to development is requires flexibility, speed and probably youthful naivety. Senior management of academic institutions are unlikely to have these skills and their role on the board of a start up is often to slow down progress and block opportunities.
- Model 4: Start-up your own development company
In 2005 as Silsoe Research Institute was closing, I was headhunted to an initiative of the Scottish Development Agency to set up a programme to develop automated condition based monitoring (CBM) for animals using wireless technologies. The initiative was called Intermediate Technology Institute with the idea being that these intermediaries with very large budgets would contract RTOs and companies to develop new technologies in strategic areas. The aim was noble and bold, to put in sufficient cash to stimulate new company formations, creating jobs in Scotland and moving away from the silicon glen screwdriver jobs which were disappearing to China. I arrived from the financially squeezed academic research sector into a world of sky high salaries and big budgets. Most of the senior staff at my division, ITITechmedia, in its glossy open plan office in the financial district of Glasgow,Scotland had been recruited from senior levels of major companies and this did not link well to the ethos of either academics or start-ups which tend to be started by people with a big vision and ambition and virtually no experience of business management. On day one I was told I had £6.2 m to spend and it must be done in 3 years. And by the way I had to spend £500k in the next 3 months. I spent the autumn of 2006 writing a massive contracting document and filed a patent on their behalf to protect the key IP of a wireless hub collar talking to multiple sensors mounted in or on the animal. At the same time I knew that this was not the real solution to the cow