How Roche designed and built an information architecture to facilitate Knowledge Management - Robin Breckenridge, Roche
CASE STUDY: How Roche designed and built an information architecture to facilitate Knowledge Management Robin Breckenridge VP for Information Systems Pharma Research Hoffman La-Roche
By on Jan 4, 2001CASE STUDY: How Roche designed and built an information architecture to facilitate Knowledge Management
Robin Breckenridge
VP for Information Systems Pharma Research
Hoffman La-Roche
I'sve abused my privilege of the speaker and changed the title a little bit in that I'sm going to talk a little bit about the designing and building and information architecture and architecture in a very broad sense indeed because I'sm not going to give you the bits and bytes answer, but instead some of the issues involved in surrounding how to get information into the environment and how best to use it.
So, I thought I'sd start, just to summarize what you'sve heard over the last couple of days and I'sm sure what is common knowledge amongst many of you in trying to drive a business case for knowledge management or knowledge sharing and where the pharmaceutical sector or the healthcare sector is going. It's a huge market. In 1998 it was worth about $1.2 trillion. It's characterized by a number of fragmented players. The largest market sector today is about 7% and in legal terms at least in other industries is very, very small indeed. It's characterized also by some low technology penetration throughout the healthcare sector, especially in the pharmaceutical sector and poor connectivity between, or poor understanding between the pharmaceutical company and the end consumer. Our relations today are not necessarily direct with the consumer, but with physicians and intermediaries in between them. In the IT compartment of course, we have a partly globalized infrastructure. We clearly have some functional information silos or geographical information silos. And we have very dispersed security models. And I'sve chosen these 3 points amongst many others to illustrate something as we go through the talk.
Tomorrow $2 trillion in 2005 is some of the estimates that were put together in 1999. I read recently that this is probably an under estimate, for the healthcare sector. And this huge market place is of course attracting new players into the healthcare sector. In attracting new players into the healthcare sector, of course, this is going to have a very big impact on existing pharmaceutical companies and how they act and perform. Networking is going to be the order of the day. If I use a slide I had last year instead of networking consolidation, but I don'st believe consolidation is going to be the case as we move forward. There are already indications in the market place, in the pharmaceutical company sector, that instead of consolidating, we are starting to de-merge into the healthcare. Pharmacia are planning and advocating the de-merger of the Swedish operation into Polaris, completely autonomous sector. My own company Roche, recently announced the de-merger of one whole therapeutic area into a separate company. And I think this is certainly a trend of the way things are happening and how things will go. And this will impact how information is managed and shared within the company and we have to be aware of some of these issues.
We'sve heard a lot about the XX model. The e's word. It's easy to put the e's in front of it, but what does it mean? I believe e's is all about transformation. And clearly we'sre going to have to have a better understanding of who our ultimate customers are, our
connectivity to the patient, but also our connectivity to the physician has to change, because the dynamics between the physician and the consumer is changing, we have to be able to support both the consumer and the physician differently as we go forward. It's all about information, how are we going to use it? And if we'sre talking about information, then clearly this will have an impact on our informatics infrastructure and how we deploy that.
I'sm putting up this particular slide just to re-emphasize that the business model is changing. We are moving into an environment where we have an information centric network, where we have virtual collaborative environments, where we will have many external partners and the types of information that we are sharing, no longer sit in the static range but spread across the spectrum from static information to very dynamic data. This again will have lots of impact on how business is organized and operated - the changing business landscape. Our organization and management structures are changing. Companies are becoming leaner. Alliances and partnerships are being formed. The network organization is there. We can'st deny it, we have to live with it and we have to anticipate it, we have to organize ourselves better to compete in that environment. And the only other point I'sll make here is draw your attention to the last major bullet point there, and to say that our customer base is changing very rapidly indeed. In whatever form we'sve heard over the last couple of days, we'sve talked around this. E-clinical data, e-research is all about this and how we organize ourselves to embrace this environment. That is the new customer and that's how we are going to be looked at by our partners outside. How do we organize ourselves internally to use this?
To remain competitive, therefore, the organization must change. We must actively seek our partners. And in actively seeking our partners, we'sre actively seeking different sources of information and knowledge and how we'sre going to use that. We have to manage that information flow across the organization differently. We have to create the culture to leverage the knowledge within the existing organization and clearly we have to promote collaboration and the sharing of knowledge, as a culture, within the organization. Many of these facets are not there today. We are trained as scientists in the R&D environment to be individuals but we are expected to operate within a team environment when we join any company.
So, having tried to set the scene here, I'sm going to take you deliberately into a completely different environment. I'sm very indebted to an old friend of mine, Chris Jones from Cerne who has given me this analogy. And it's the analogy of the High Energy Physics Community. How many of you recognize this? 2 or 3 people recognize this, maybe more. This is a computer simulation of an old-fashioned bubble chamber experiment which was trapped in the old days on an x-ray plate. And these squiggles and lines where the decay pattern of the experiment, are representative of the times of the late 60's, early 70's. In this environment the High Energy Physics Community, we'sre not really dependent on computers, although they did use them. The data was recorded photographically. We rely very much on hand scanning and measurements to do this. And there were huge errors in data and the knowledge about the physics. Informatics was important because they use it as a number cruncher, but certainly not critical to the discovery processes. With the advent of the electronic detector and the next generation of experiments that are occurring in Cerne, these are some of the stats we'sre looking at. Data is clearly recorded electronically. We are planning or the High Energy Physics Community will have to handle petabytes of information, not a year or a
month or a day, but every second. A petabyte of information is about a kilometer high of CD ROM's landing on your desk every second. We have to handle this.
Clearly this whole community is very IT dependent, but they'sve learnt to change. And I think there are some very valuable lessons for the pharmaceutical sector in how they have learnt to change. They'sve learnt that they can'st possibly store and manage this petabyte of information landing on their desks. They'sve learnt how to extract the key elements of information and reduce that so that instead of having to handle a petabyte of information every second, they can condense that to a petabyte of information every year. It still creates huge challenges when you believe that Celera are talking about terabytes of information. There's a quantum difference. Petabytes of information, of course, requires some new storage capabilities, some new ideas of how to handle that. But they'sve learnt how to, as I say, how and what to manage and store. They'sve also recognized that a special infrastructure was required to support that collaboration and realized that to survive the processes that they'sve inherently used within the High Energy Physics Community have had to change and evolve. They recognize that the value of intellectual capital should not be expended and design the experiment, but needs to be expended at designing and interpreting the data. They'sve learnt to collaborate as a community. High Energy Physics Community, after all, has given us the worldwide web. Without the www we would not be speaking about the e's word today. They have learnt all these. How many of these learnings have translated into the healthcare sector? That's a challenge that we face.
In learning to deal with this community, the strategy for survival has harnessed the technology, has harnessed the business processes and the need to change the business processes and effected the organization. People now collaborate across the globe, to run the experiment and to view the data. As I said, the intellectual capital comes from the interpretation of that data. That's where the challenge is. Business processes have had to change so that you are now sharing this information. You are now modeling the information, the experiment in the computer, in silico, e-priory, to learn to understand what it is that you need to extract from the experiment and manage as data elements. These are all lessons that we need in the pharmaceutical sector, the healthcare sector to learn and manage.
The one other learning from that sector, coming from the collaborative environment, if you wish, is, instead of doing things in the sequential sense, that community has learnt to build on each other and to drive some value much faster. I'sm indebted to Ken for introducing Maslow, because I'sve also used Maslow and have abused Maslow because of a colleague of mine who developed the Maslow pyramid for IT. Typically IT has been organized along operational excellence where we'sve produced infrastructure and business and software. As we go up, we recognize that we need to interact with and provide a service to the user community, therefore we have some customer intimacy ideas around customer services, help desks, support desk training, etc, etc. Only very high up in the pyramid do we get anything about value coming out of informatics. And this is the paradigm shift that we have to enact because we have to turn this on it's head. We have to turn this on it's head so that we are seen as a community to be using the value of information to improve the business sector. And that's a challenge because that challenge does not just rely upon IT doing a better job or better algorithms, but it relies on the integration of those 4 circles that I articulated on. The strategy, with your technology, with your organization and your process. And the challenge in that is of
course that the technology cycle is very, very rapid. It's running at 3-6 months for a PC. How quickly does organization change in your company? How quickly do processes change in your company? And very often that these 3 cycles are running against each other and the end product is a very poor ROI. On any investment in technologies, whether it be IT, whether it be high throughput screening, or whether it be genomics. We have to learn to try and bring these 3 loops together.
We have to learn to create value and in creating value out of changing our practices into a collaborative mode and sharing that information. Having said that, I don'st want anybody to go away and assume that IT costs are going to decrease, they'sre not, let's face it. They'sre going to continue to increase. But we can manage that increase. If we didn'st do anything then we'sd have a very steep curve. But with the advances in the IT technology per se, we can control that and we can control that by starting to simplify the way we handle information. And the challenge is to try and get to this level, because if we continue to expand our expenses in IT at a very high rate, then business cannot support us. And it will make it even more difficult to drive or to demonstrate the value of information out of the organization.
Clearly we are facing an explosion of data information. The analogy with High Energy Physics goes so far and only so far in terms of the volume of data. The complexity of the data is very much higher in the pharmaceutical sector, so we shouldn'st forget that and that is something that makes us different from the High Energy Community, but nonetheless the analogy is still valid.
We have moved from a situation in the recent past where the types of data that we'sve involved and managed tend to be in the static environment, where we'sve been able to influence the control and format that data in our own environment and as we'sve moved into the external environment, we'sve lost a little bit of that control. But primarily most of the data that was used within the company was within our control. As we move forward, we'sre going to lose that control. We'sre going to lose the ability to influence how that data is formatted because we'sre going to be combining information from right across this spectrum of internal work, of public information, of information coming from our collaboratism partners. And we'sll have more or less control over the format. We have to do things differently if we'sre going to be able to integrate information in this environment.
There's also this crisis about software. It's permanent. It's not going to go away. Let's face it, let's not promise more than we can deliver. Who flew across the Atlantic recently? Lots of you, I'sm sure. We'sve all flown planes that were designed more or less by computer. We'sre happy. We'sre happy because they'sve built in certain checks and balances to make sure that the computer designs that they'sve used to generate the manufacturing of that aircraft does actually work. And more or less it's been substantiated. So, there is some credibility in software as well. But there are some lessons that we can probably take out of that environment. I strongly believe that we cannot continue to be a software house. The healthcare, the pharmaceutical company cannot be a software house. In most large pharmaceutical companies you are talking in terms of the informatics community of 1000 people, 2000 people. That's a very large software house. But that does span the range from very basic infrastructure, all the way up through knowledge management. But it's still a very large community internally. But if you'sre going to generate the value out of that community and for the business, then we can'st be a software house. We have to be able to go out and buy the appropriate
software to do it. And there are certain criteria that I believe are necessary if you'sre going down that path. I believe that we need to define and use our standards and the technologies. It's no use saying, we'sre going to use SAP or Oracle or MDL or any of the other technologies out there. We have to consistently use an architect, the data that goes into these technologies more efficiently internally. How often have we identified SAP as a technology, say, okay, we'sre going to use SAP's? Fine, how difficult is it to implement SAP? How difficult is it to manage SAP over generations? And I apologize for anybody who's represented from SAP, I just use that as an example. It is ubiquitous. It could be SAS. But the point is there. We have to do more than just declare a standard technology, we have to declare how to use that technology inside the company and how to derive benefit from it and in doing that we'sve got to be able to balance innovation and customer expectations. Innovation because the rapid change in the technology is there, it's a fact of life, we have to cope with it. On the other hand, customer expectations are some what more conservative and we have to understand where and how to draw that line and how to balance those acts.
Now let me take you into knowledge sharing and come back to R&D. The challenge for R&D; I agree with Ken that we'sve got to get away from the very linear process and are very attracted by what Millennium are trying to do and I know others are trying to do something similar. But fundamentally we still have a problem because our rate of identifying a compound that goes onto the marketplace, compared to our rate of identifying targets and early discovery is very, very important. The nutrition rate is abysmal and we'sve got to be able to change that curve in some way. That's our challenge and only information and better use of information will help that. And information of course requires people and requires technologies and requires closed synergies. These are some of the challenges that we face and have to handle and deal with.
In any organization today there are a number of different barriers that we face. We have functional and geographical barriers, but within these functional geographical barriers there is at least a common language. In process barriers there are some common goals. But if you put it together, we end up with little islands of information. How many people recognize that? Is there a company that's exceptional to this? If there, is I would like to join them. This is a fact. This is the way we have to work today. And how do we meet that challenge and how do we tackle it? Clearly we have a loss of information at the time, our inability to find the right information. The people side, the culture side, the not invented here's syndrome is very rife. We have a very poor customer focus. We have internal competition rather than corporation and we have inefficient accountability.
Some of the challenges: If we are moving into an environment where we are going to be information centric, then we can start with these quadrants where on one axis I have structured information and the other axis knowledge, where I have defined knowledge as information in a context with a conclusion. And if you do that, and we'sve started in the past by trying to structure our data and we'sve moved it forward into an environment in 2 axes. We'sve started to structure it and reduce the redundancy of that information. The value is still pretty poor. We'sve improved efficiency, we'sve improved the operation costs, but not increased the value. If we go in the other dimension, we take information, a number of different pieces of information and you draw some conclusion from it. You'sre generating some value for the organization by doing that. But in the act of writing these documents, we'sre actually, again, creating redundancy in the data and information, in the
way we work today. There is no audit trail in how you got to that information. Audit trails are something very essential if you want to write an NDA. And the link to the original information and any change to that information is of course lost. We'sre moving across a diagonal today and tomorrow I believe that we have to be in an environment where the processes, our underlying processes and our underlying ways of working have changed somewhat more considerably. Documents are structured, not necessarily in the formal sense of an Oracle database. There is an audit trail. The content is fully tagged and we'sve reduced the redundancy. We'sve got to the stage of a virtual document where the original data stays in a database and it's compiled and delivered at will for the respective needs of that individual. If we get to that stage then we will have lowered organizational costs. Information should become much more ubiquitously available and knowledge can then start to be searched. That's a challenge in an organization where the workforce is becoming much more mobile and moving. We'sve got to learn to capture that information and knowledge residing in the heads of individual. The knowledge sharing program in Roche has 3 main goals. We try to balance innovation and re-use. We try to promote the pervasive sharing and leverage our knowledge. And that's a cultural issue, a behavioral issue that we need to change and we need to integrate the explicit and tacit knowledge elements within the organization. I'sve gone so far without a definition, so I thought I'sd start with one.
Sammy Johnson once said that knowledge is of 2 kinds. A subject we either know or we know where to find information about it. I think that's probably quite valid in the way we have to operate today. If that is the case then I would contend that we can draw some kind of boundary around our information space. Our information space falls into either formal or informal processes, into structured and unstructured data where you'sve got these typical segments, and these are just typical examples and you might want to query public domain data and structured but informal processes. But it's there, it's the web. There is some structure on the web that you can go and get. It doesn'st come from a formal process as an individual company might understand, but it is there. The challenge for us is divine and architecture that helps to integrate this information.
If we'sre going to enable communities of practice, then we need to combine these typical pieces of information and I would suggest that the way we might think about combining them will require a better understanding of our meta and master data. A better understanding of the taxonomy, the language of communication between different functions and values and we need to define and understand who has access to what information. It's very important. It's an issue that is perhaps overlooked in many, many companies today. Because I believe that information can be used as a strategic asset and it has to be used as a strategic asset and if you'sre going to use it then we need the structure to manage that.
Why meta master data? Well, I'sll define a transaction process. A very formal process that we have which supports the gathering of data and an action that comes out of it. On the other hand, information integration will extract information. Information elements from a variety of different databases and deliver it. We all recognize that a call center needs to extract information from our SAP databases because we need to know what volume we are generating from any enquiry. What is our return on investment from a particular vendor. We need to understand information about that particular caller in some detail and we extract it from various different transaction process databases. But in a call centre you can also imagine that there is a feedback loop which will in itself
activate a transaction. And if you'sre going to enable that to happen, then your information integration subsets need to use and have the same language as your transaction databases. You need to have a common definition of meta and master data. And you need to tie them together.
We'sve also come across this paradox where there are many generic tools that will search unstructured databases and give you an answer. But are you aware of the quality of that information that you get back? You get lots of information back. Are you aware of the quality and the context it's derived? On the other hand we have very few tools that will be able to search and combine our structured data sets. They tend to be point solutions today. We need to make searching in the combination in our structured world as easy as the tools that are available to search all this, to find some of the data in the unstructured world, on the internet. On the other hand we also have to find vehicles, ways of trying to extract more valuable information from all this very unstructured data. And these are 2 opposing dilemmas that we have to try and tackle.
We all recognize a process. It could be a collaboration between continents, between functions, what have you. Within each function, or in each activity, it is relatively straight-forward to be able to access the data that that activity gave rise to. But for this poor individual in the middle who is just trying to extract information from these different activities, it is exceedingly difficult. It happens because the poor user will then go to one database and extract a piece of information, cut and paste into an excel spreadsheet, go into a second database do the same, go to the third database do the same. You can do it. It's not very efficient. We have to move, we have to change that paradigm.
Technologies are certainly available to help us do that. We can make this guy very happy. We can increase our value very rapidly if we start to structure our information sets, if we start to use our information sets in a coherent fashion across the organization. And when I say across the organization I don'st just mean internal to the organization. We have to be able to integrate information that originates outside of the 4 walls of your company. And it is double, it is double the technologies that we have available but there are certain premises that we have to identify and work with. So, in Roche we have tried to take a holistic approach. The credibility of IT, of the informax community is, in general, very poor. Very senior management in most companies still perceive IT as a cost center. They don'st understand how they derive value out of it. To try and combat that philosophy we'sve taken this holistic approach and tried to demonstrate the feasibility and value and lead by example of combining the people, process and technology issues that I'sve tried to articulate over the last half hour.
We impose some structure on it once we'sve got the momentum moving forward. We try to define some Governance, rules, so that we start to tie these different initiatives across the organization together. And finally, our goal is to make it pervasive into the environment. If we do that, then we can move from our current environment which is currently document centric today into an environment which can be information centric, into an environment where it's easier for individuals, through standard portals, to go and extract and integrate information. It's easier to compile documents for whatever process and whatever purpose. It becomes easier to navigate these informations because we have a different paradigm of searching and retrieving. Knowledge sharing must enable the business strategy. It's not an end to and itself. It has to focus on the critical business knowledge because not all knowledge is equally important to the business.
Just as in the High Energy Physics case they'sve had to recognize, out of the petabytes of information that was delivered as raw data, how to generate some value very rapidly out of it. We need to do exactly the same with our information elements and understand what are the critical pieces of information that we need to manage and keep for a time and which we can throw away rapidly.
It's not easy because it requires behavioral change. We'sre all very, very conservative when we think we will be impacted. We have to change that philosophy. We have to change the reward and recognition mechanisms within all our companies to make collaboration and knowledge sharing a real entity within the organization. And let us remember the technology is only an enabler here and not the driver. And if we can do this then we have a chance of living this particular statement from Charles Darwin; because it's not the strongest in a species that survive, but the one that is most adaptable to change. Thank you very much for your attention.
Q: I thought the presentation was excellent. One issue about knowledge management is, how do you capture the knowledge and how did you guys face that issue and I know you outlined that it was imbedded or you should imbed it in your processes and allocate time, but how did you guys really do that?
A: It would be wrong for me to say that we'sve solved the problem. There are some techniques of doing this and one of the ways that we are using is to capture information in collaborative communities so that when somebody makes a comment, it can be captured within that community and made available to that community. So, these are the tacit pieces of information as we move forward. But we also assume that much of the knowledge is today captured in a documentary form. And our main emphasis just now has been to try and target that and try and structure that information set and make that information set much more searchable so that you can use it in a different sense.
Q: Robin you have mentioned at the beginning of your presentation that this industry should develop standards. Where do you see the most need of developing standards when we look at the expanding scope of where you exchange information?
A: I think the standards can be applied in many, many different small areas. And my priority areas would initially be in the regulatory environment, in our submission documents. If we can define standards, a common basis, a common language that we will all agree to use and the FDA and other regulatory authorities accept, then we can certainly start to cut out large sections of time in the development cycle. That's one example, but I would certainly target that as one of the key areas to go for. The bioinformatics community, because it's young, it hasn'st got a legacy, could well be another arena where you want to start developing some standards and push forward for. And that's beginning to happen because it's happening, as I said, because it's a young community, it's a young science, using modern technologies, and they'sve had to work together to exchange information and by exchanging information, they'sve by default started to develop a much more refined language of communication. What's happening there, we need to accelerate in other areas where we have much more legacy involved.
Q: I think the industry is making progress there, but what about the regulatory environment, and this is, I think, a step in your desired direction. But regulatory is one thing. Working in different cultures is another one. What can you tell us here?
A: I think it's something that all companies struggle with and especially companies that have gone through mergers and acquisitions because within each company there is a culture today and when you bring these cultures together, it creates trauma within the organization and if you can manage that trauma then the company will benefit. But very often mergers and acquisitions fail because they don'st manage that trauma very well. At least that's my personal opinion. There are obviously other reasons as well, but that's a contributing factor.
We communicate with English, we communicate today with English, but English is not our native language, or not everybody's native language. That's a cultural change. There's a huge cultural difference between Europe, East and West Coast of America and the Pacific Rim.
Q: Somewhere in your speech you mentioned reward measurements to foster information sharing cultures. Can you expand on that? Also maybe a following question, have you implemented some kind of top management policies around sharing policies?
A: Okay, 2 questions. The first one about reward, I think that the people that you should speak to about reward in this sense are probably the consultants amongst us because they have traditionally tried to work an environment where they'sve depended very much on the ability to share information within their culture, within that company. And many of their reward schemers are very dependent upon depositing of information and the use of information. It can'st just be about the deposit, it has to be about the use. Reward of course can take many different forms. Financial, but also recognition and there's a balance that needs to be drawn in there. I think most companies HR departments have recognized this as an issue and are trying to tackle it. Some of the more advanced companies are outside the pharmaceutical sector, in my opinion, that have tackled this. Siemens and Philips, in particular, are very good examples where they have, because of the trauma over the recent past, in their companies, have had to introduce some of these schemers and are quite successful.
The second question was more about the top management and the sharing. You'sre right, we need to get top management to practice what they preach. They would like to involve the talk about knowledge sharing but very often they are very poor at practicing it and implementing what they say. Over lunch we were having a discussion about the issue about how confidential company strategies were and if you just take that as an issue, how many companies actually publish their strategies in a relatively detailed way so that people can interpret this? Once we start to break down those types of barriers, then we might start to get to an issue where it's there. But there are some efforts, certainly within Roche there are communities of practice within the senior management to try and encourage that to happen and many of the senior managers recognize that they need to do something and that's certainly the first step moving forward. But it's very difficult because there's, especially when it comes to information, there is a need to balance the value of information where value could be counter productive if it falls into the wrong hands. So, you need to balance that and I also mentioned that you need to define your information excess policies and procedures. This is something that many companies have to go back and look at because their policies and procedures are outdated for the e-world that we live in today.
Q: Robin, just to be provocative, there was one piece in your presentation that I found very interesting and it was the observation on the lack of acceptance by senior management of the IT, IS knowledge role within some organizations and I'sve been thinking about it and I'sm just asking myself, what do you and your group have to do to be accepted at absolute top management level, number one. Number two, I would expect the function you perform is to provide decision enabling information and knowledge at senior level. If there's a lack of acceptance, then by definition some of these decisions are being made either on best guess, uninformed or an alternative method. So, coming back to the question, what do we have to do to get a buy in?
A: How about all 3? What do you have to do to get a buy in? I think that today, IT is very much thought of as the bits and bytes, the why's that help the communication but not about the value of the information that that IT helps to transport. And it is that quantum leap that needs to occur in the mindset. You have to be able to demonstrate the value of the information, not talk about, we need to invest in a PC because,'s or we need to invest in another storage mechanism because we'sve got lots of data.'s They don'st want to know that. What they want to know, what they would like to try and understand is, what is the return, how are decisions going to be improved if we make this investment? And that's the change that has to happen and that's the change that is beginning to happen in many companies, hopefully including Roche.
Q: Robin, coming onto this question about metrics, essentially, we are talking about creating value here with IT, so you'sre talking about metrics and metrics is one of the ways to actually persuade senior management about a good foundation of a certain enterprise. Now we all know for the last few years that any major pharmaceutical company needs between 2 and 3 NCE's to be put out on the market to guarantee a 10% raise year on year of our business. Now, it's pretty trivial to extract from that and given today's probabilities of success that is, I would say, standard across all organizations in the pharmaceutical world, that any form of technology that has the capability of raising by, I would say 3-5% the probability of success, overall of an R&D process, would actually be able to produce one NC in every 5 year time frame. That means about 1 billion of value in the market in terms of peak sales. Now how would you take that sort of comment of issue forward and how would you like to comment on that comment?
A: I comment from my experience and discovery and that is that one of the arguments to go with high through put screening and develop high through put screening technologies was the fact that it would generate information a lot quicker. High throughput screening has not actually improved the research of productivity profile at all. We generate a lot more data but we'sre not producing any more compounds out of it because we'sve not, I would contend, understood how to use that technology and made the necessary changes in the processes and the organization and skill sets required to draw that value out of it, that investment. Does that go someway towards answering your question?
Q: I just wanted to add, so how do you get the information technology people and how do you get the leaders of the company to change their minds? One of the things that's happened at Millennium is, the IT people are in the lab and we literally have individuals linked up, probably on a base of 1 to 3. 1 IT person for every 3 scientists that are on
their team. And so the scientists, when you come down to budget setting time, which is sort of your question of, how do you get people's attention, the scientists are demanding more IT people, as compared to the IT people are saying, we can help you.'s Because most of the time, in fact even in Millennium, even a year ago, they were in a different building, no-one could see them and they were doing their work and so they weren'st visible enough. And I think once they became visible physically, in the labs, working there in the lab with the scientists they were then sold by another group and also I think they appeared to be a much more strategic weapon for the company as compared to an expense. So, that was a major change. It's a little bit of the slight of hand but it was a major change in how they were perceived.
A: So, proximity it helps. The other mechanism that, in my experience, management have used is when, as it happened to me when I'sve gone and said, I need more head count.'s It's been put before my research executive colleagues for their support and decision and what has been asked of the research executive colleagues was, are you willing to commit a chemist or a biologist? Are you willing to give up a head count?'s And in the recent past the answer has been, yes.'s So, there is a mindset change within the organization that people are willing to give up a chemistry headcount or a biology headcount or pharmacology headcount and donate it to IT because they do start to see the value that's coming out of it. That's set at a slightly lower level than the level that I pitched the original answer at which was at the Board level. And at the Board level we still have a long way to go.