Big Data (hmm), Live Data (Yes) in Conjunction with Algorithms & Intelligence

Well, there are two buzz words in our days in science and technology, big data is among the most prominent and most considerably used in the field towards transforming the society. The idea is fairly simple, and well rounded, machines using colossal amount of data will be able to transform "stored" information to wealth and intelligence and in theory better interpreting them than humans. The basic assumption is that data is available and then the problem is solved.

Such assumption I think it could be one of the greatest misapprehension of this decade. Data was and is around for almost three decades now, one can cite for example stock market, where historic data are available ages. What have changed between the past decade and this one? Well three simple things: (i) computing power has extraordinary progressed while its cost has dropped down to peanuts, (ii) convergence between computer science and mathematics towards a field called machine learning where statistics, artificial intelligence, optimization and programming principles are met together, (iii) easy access to this data either in a centralized or distributed manner. The above once in conjunction with training (usually annotated though) data were able to show great promises in a number of fields.

Having said the above, the question now is what will be the future from now. Were there are two theories, (i) data "fundamentalists" believe that soon there will absolutely no need of physical understanding of the problem/origin of information  and in the presence of massive amount of (annotated) data I should be able to perfectly model the complex behavior  of my observations (black boxes!!!!), (ii) statistical/generative modeling where certain assumptions are made which introduce physical and humanly interpretable meaning on the solution which is then optimized using training data. It is hard to predict to what direction the cursor will move however it is certain that depending on the problem being considered most likely the solution will be different. It should be noted that we leave in an information era, which practically means that more and more (new) data will become available which could be used to better enriched existing models rather than building new ones.  

From scientific view point is the big data "bubble" that different from the "internet" bubble in the nineties, or the "artificial intelligence" in seventies? Do we really need to reason on colossal amount of data towards making better predictions or taking more appropriate decisions in the general case? Is it that hard to predict everything from anything when we increase the degrees of freedom of the models? Does more info even if not relevant add something if combination is not done appropriately? Hard to say. Would that data need to be "big" not so sure, I will guess representative but not that "big". Would the answer to all problems being "black boxes" trained with massive amount of data with "complete" lack of physical interpretation? not that sure "either" even though if it seems to the trend now...

What is easy to say though is that live data thanks to the internet of things and connected objects is become freely and massively available. Personally, I believe that this will be the real motivation/driving force of wealth creation. Simple day-to-day living problems associated with live measurements transferred over the network and adaptively solved using models based on historic training data.

Having said the above, it seems though that little attention is paid on the "scientific" issues relevant to the  data interpretation that are developing new mathematical models and their computational solutions towards reasoning on this data! Very often, we have the impression that all problems will be solved  once we have collected this amount of data and used them "naively" using massive computing. We also have the impression that each of us can become a "data scientists/manager" once had obtained access to such data. Well this definitely will be the greatest disillusion  of our decade.

Academic Reseach in Industry: "Il buono, il brutto, il cattivo" for Academic Carreer?

Research in industry has been instrumental towards transferring academic and research achievements to society? In engineering sciences there are several fantastic labs that were built and supported from  the major conglomerates, like for example the Xerox PARC, HP or Schlumberger Lab at Pallo Alto on eighties, the Siemens, NEC, Lucent, GE, IBM labs in nineties, the Microsoft, Google labs in the last decade and most recently the Facebook/Apple labs in the ongoing decade.

The recipy is not hat different despite almost 40 years of interval. Industry is doing quite well, cash flow is available, so the idea was and is to hire exceptional academic faculty with three fairly basic arguments: (i) salaries that cannot be met by academia, and (ii) research environment free from the time consuming and tedious process of "grant-hunting", (iii) interesting problems and access to data that cannot be found easily in the academic world. 

Arguments are quite convincing (at least at the very beginning) conditions are amazing and things do work for certain time. However, industry has sort of cycles, in particular in the digital era things can change so fast!!! I recall when joining Siemens Corporate Research most of the funding used to come without conditions from the parent company (75%), while being there this percentage went to 50%, and the date I left the parent company funding dropped to 25% and discussions were ongoing on how to even further reduce to 10% and get the remaining 15% from government grants. Things might have changed now but that was the case then. Places like Xerox, HP, NEC disappeared and fantastic environments like Microsoft Research have considerably diminished their investment targeting few  areas of computer science & engineering! (which was not the case at the time of their creation)

This is not seems to the case for places like Apple, Google or Facebook in now days, the cash flow is enormous, but it wasn't the case for Microsoft back in nineties? or for HP/Xerox back in eighties? Working in industry as an academic researcher is a great experience but has definitely an expiration day that one can hardly predict especially in engineering sciences where economic growth and wealth creation can happen and disappear so fast...

The conclusion, join as a freshman, a place that does well, keep in mind that this will not last, and therefore spend (max) 5 years working on interesting problems (it is such a great experience and  definitely will have a tremendous impact in  your academic career) while make all possible efforts to maintain a top level research portfolio and move towards a junior faculty position, or think in the long run, assume managerial responsibilities, build a sizable group and seek an academic position at the level of associate/full professor with tenure after 10-15 years.  

Keep in mind that the situation will not last, salaries can be interesting but at some after some years you will be spending more and more time an management than real science. Of course, you can have a brilliant industrial career having a different impact on the society but this is not the view point/scope of this article targeting research perspectives in industry.

Science, Social Media & Marketing

Proliferation of means for exchanging ideas has lead to the "popularization" of sciences. It is becoming trivial for any of us to act/pretend like a real expert by composing a set of buzz and trendy words in particular when it is feasible to brand himself/herself as an expert by heavily investing social and scientific networks with "meaningless" but still "popular" content. Consequently, real-science become irrelevant and people who are usually talked about sciences have not much to do with it, or in the best case scenario fairly have a vague quite superficial idea on what they are talking about. 

Well, fortunately mastering a subject requires top level academic studies, persistence, motivation and hard work. We observe that these values are losing ground when facing modern communicators, people with limited technical background who are able to talk about everything and nothing, and were able to build an image of someone mastering the topic. I am not a great fun of social networks, however given the importance that these have taken in our society, one cannot neglect them. There are several scientific problems to which I have dedicated enormous amount of my academic career. It is frightening when seeking for content relevant to these topics through social networks. The list of top contributors have often almost no footprint or contribution in these areas but somehow manage to maintain their "virtual expert" status through what we call the theory of buzz. Keep producing as much content - even superficial - as you can things that could demonstrate that you are an expert.

The risk of such situation is dramatic both for science as well as for society. First, scientists become these "lab" people disconnected with the "society" problems who in the most general case don't know how to communicate and therefore give the impression that problems that they are working on are already solved. Such an impression could have devastating impact on them, as it concerns future research directions, funding, motivation, etc. Furthermore, convincing you highly motivated scholars who are usually the ones bringing in the greatest innovation to pursue graduate studies are becoming more difficult. In terms of society the situation is dramatic. It is fantastic to come up with ideas, concepts that could be of great interest, while having actually absolutely no means of scientifically addressing them. 

There is nothing wrong - and actually is great - about popularizing sciences and being able to disseminate complex ideas to a broader audience (beyond the scientific community).  Actually this could be a great source of growth and innovation if done properly where real needs coming from the society meet technical innovations coming from the scientific community.

However, talking about everything and nothing and substituting scientific work with a number of buzz topics/trends being alimented from non-(appropriately qualified) experts is becoming a major concern both for the society and the scientific community. Such a overdose of "useless" content could be harmful for the society, could drive the academic community on directions that are the ones which should be followed, force schools and universities to artificially address critical scientific disciplines and consequently reduce the intellectual background of the future generations of scientists and consequently their capacity to address the real problems/needs of the society.

Render unto Caesar the things that are Caesar's

Competition (healthy and unhealthy) is among the driving forces of our society. I recall more a decade ago my promotion to program manager at Siemens Corporate Research, Princeton, NJ, USA. It came with a single person office, an important salary raise and a two days seminar where communication/management experts explain to us what is the path to success. Well it seems that there are three classes of successful managers:

• The group of highly talented and gifted individuals who assume their decisions, have a vision and take the initiatives without caring about their short term outcome. Eventually there will be cases where things will not go as expected but eventually at some point hard work, leadership and charisma will pay out and this will happen independently on politics. Usually these are the people who will make the difference and take institutions to the next level.
• The second group of people who most likely endow their career with success are the ones working hard, delivering what was asked from them and being at the right place the right time. These are lower risk cases with respect to the first category since the outcome is predictable and bounded both on the positive as well as the negative side. These are the kind of people who will assume the job and most likely deliver what is expected from them while taking almost no risks.
• Well, the third category corresponds to people playing politics and use connections to make the difference. They are talented because despite the absence of charisma, vision and loyalty they are able to navigate and survive in critical situations and they actually just care for themselves. Their survival code consists on taking other colleague's credit and putting obstacles whenever they feel threatened to to whoever threatens them.

I guess all of us are almost certain that we are highly gifted and talented. However, I recall the expert's conclusion stating that we often cross people during our professional path who belong to the third category and suggesting to avoid working as much as you can while hoping that in your institution there are people with the willingness to Render unto Caesar the things that are Caesar's.

 

ps. I was fortunate enough to work with collaborators coming from all aforementioned populations, a situation which strengthens your ability to appreciate people being loyal to their  task/job and not to themselves.

The Myth of European Research Integration

I have spent significant time of my academic research career in Europe and during the past two decades I have participated to a number of EU-funded research projects, covering almost all possible funding schemas. Until very recently (the creation of the European Research Council), EU research policy was towards European Research Integration. The underlying principles of this policy towards getting support can be summarized as follows:

• Create a reasonably spread research consortium representing well the geographic distribution of EU members (north/south, big/small, old/new members)
• Introduce artificially some industrial connection to the consortium either by involving “research” in industry, or an industrial partner as integrator for some of the calls, which will introduce to the proposal credibility with respect to the society (well industry will transfer the technology)
• Establish connection with the research office in Brussels offices, and hire a professional company to write the proposal where the “call” buzz words appears as frequently as possible and also consider a professional company as well as to assume its coordination,

While all these conditions do make perfect sense and should have led to an efficient and productive research integration, this didn't happen and tremendous amount of resources was spent on projects with questionable return to investment . There are a number of reasons for that:

• Until recently the evaluation/review mechanisms were questionable/problematic. Researchers coming from the whole Europe used to meet a week in Brussels, gaining access to - hard copies of - the proposals the day of the review, and had to made an assessment during the day without having any means to check the credibility and the quality of the teams being involved in the proposal.
• Until recently (and even now) the programs were run  from professional managers (with certainly excellent prior research experience) sitting on their positions for while and having really strong influence on the nature as well as content of proposals being funded. While it is certain that almost all of even all of them were highly capable to do so, such a system is not healthy. Same people mean creation of network which is something natural in society (both from researchers as well as from offices). Network (in particular long-standing one) is equivalent with problematic renewal both on ideas as well as on people/labs funded. Introducing at least some notion of rotation would have been beneficial.
• Until recently evaluation/follow up mechanisms were not appropriate. Reviews (either intermediate or final ones) were not done properly. It is uncertain how the selection of the reviewers is done and how they are assigned to a project but the whole process is very problematic. I recall talking to a friend who was involved with a FP7 grant involving highly visible people and end up being evaluated from one of my students (who just graduated) and another former student of one of the most visible members of the consortium. When this was pointed out to the project officer his answer was quite impressive: it is good to have fresh blood in the process...I do review grants (and spend a lot of time) for many international funding organisms (Netherlands, France, Switzerland, Germany, Israel, Canada, Belgium, Luxemburg, Honk Kong - surprisingly enough I was never involved on any EU-related proposal review or evaluation), I could state for example that the Austrian Science Foundation (among others) is doing much better job and put in place far more rigorous evaluation mechanisms.

The above issues have create a strong disconnection between the European Union funding and the European researchers. The system is far from being predictable, proposals are now prepared from professional grant writers/companies and there is a huge concentration on funding on institutions/groups per call which is neither healthy or normal.This doesn't mean that these institutions do not deserve funding, they certainly do but I am almost sure that their funding envelope is far from being proportional with respect to their contribution in their domains at the European level.  

Unfortunately EU doesn't put up statistics on institutional funding. It would have been for example interesting to compare/establish correlation of that with research productivity (for example as it concerns research projects) in Europe.

The complexity, inertia and the lack of transparency of the system has led many (and some of the excellent) labs to give up on getting support for their research from the Europe which in the long run could be devastating for the European research. The European Research Council seems to be a mechanism to remediate that but these are mostly individual grants which do not offer the ability to support highly ambitious interdisciplinary efforts. It is certain that European research integration through flagships grants is necessary and important in order to be competitive at the international scale. Unfortunately, current standards and practices do not  provide the best possible conditions and the inertia of the system is unbelievable. 

Lets hope that the new framework (Horizon 2020) will change things and make the process more efficient, more productive and more transparent.

* I am familiar with the computer science/applied mathematics

* EU funding schemas were (and still are) quite generous as it concern my research activities

Research in France: the philosophy of mass

Well, it is really difficult to define the best possible conditions towards excelling in research. It is almost certain that situation varies significantly from one field to another and is also certain that it is not just a question of support.

French research has been primarily structured around public research institutions like for example Cnrs (sciences/humanities), Inria (computer science/applied mathematics), Cea (nuclear, physics, etc...), Inserm (biology) and secondary around higher education institutions like Universities or Grandes Ecoles. The common denominator for almost all these institutions is that fact that research scientists or professors are stated-funded and their research activities was primarily supported through direct funding schemas (eighties, nineties) towards research institutions. Such a policy has created a rather unbalanced fragmentation of the higher educational system where Research Institutions were there to carry our research, Universities to form students and Grandes Ecoles to form the elites. Direct research funding was possible due to the relatively small population of research scientists who had to go through a highly competitive hiring process. Despite a huge number of great success stories, such a system was not sustainable. The disconnection of research from the higher education produced a huge gap between the elites and the research world which led to a deficiency of industrial research and funding as well as technological transfer. Furthermore, the research funding deficiency for the academic faculty has led to their marginalization in terms of research performance expectations reducing their role to pure teaching.

The situation has changed over the past two decades. It became evident that disconnecting research from the academic world was not such a great idea, while at the same time the direct funding model has reached its limits due to the substantial increase of research personnel as well as the declining support from the state. Internationalization has shown that indeed a professor can be an excellent researcher as well and a gradual philosophy shift was observed where particular emphasis was paid towards strengthening the research ambitions of universities/grandes ecoles through a progressive decrease of direct funding and its transformation to "competitive" one. Such a transformation seems like a great idea, which though couldn't be implemented successfully due to insufficient support of the state (I have hard time to figure out the ratio between direct and indirect funding) resulting to state-employees salaries that are far from being competitive at international level as well as lack of solid "environment" research support. Once the deficiency of the state funding is coupled with the substantial direct funding schema and the constant request towards increasing the number of state-paid researchers we have the perfect storm to kill scientific excellence.

• Due to state policies research personnel being accommodated with the direct funding model "unwilling"/"inefficient" to seek for competitive funding (France's contribution to EU funding is higher to the amount that French institutions do recover from it)
• Research personnel (actually for the past five years new openings correspond to people who retired) is not renewed since there are no pathways (in both directions) between the research institutions and the higher education ,
• Research conditions that are constantly degrading where both direct and indirect funding become insufficient to guarantee descent personal and professional conditions while being non competitive at international level,
• Constant increase of research population (which is not necessarily a bad thing) where research activities are replicated among public-funded institutions without even making a minimal consolidation effort even when there is a geographic proximity.
• Research personnel being obstructed from obscure bureaucratic rules regarding research portfolio, promotions and research funding
• Research personnel being disconnected from industry and industrial needs with limited interest on technology transfer (working on industrial problems is not the solution, but there so many problems in industry that require fantastic science for solving them).

These conditions (that have been around for almost three decades) define the philosophy of mass: hire as much research personnel as you can, do not really care regarding their conditions, do not really care on what happens to the ones lost in "traffic", do not really care on the replication of efforts and do distribute direct and indirect funding as uniformly as possible. The counter argument for these conditions is the fact that you are a public-servant and you have a life-time position where you can have complete freedom...However, people are deciding to follow this path are highly motivated and therefore some of them will excel because of their personal motivation, dedication and hard work without much of help form the system being in place from the state.

Having said the above it is important that policy makers take action towards excellence and not towards mass preservation. Increase of competitive funding, decrease of direct funding, consolidation of efforts/organisms/funding schemas/simplification of procedures/...

* Note 1: I am  mostly familiar with the applied mathematics/computer science field and therefore these ideas might not be that valuable in other fields. 

Three simple conditions towards a highly impactful PhD Thesis

• Carefully select the area of your thesis, avoid areas which are overpopulated or prior literature is huge since probability of penetrating is fairly small even though citation-wise it might worth the effort. Consider a topic that personally motivates from scientific, societal view point (when applicable) and be sure that the choice is consistent with your academic curriculum / scientific background / skills. In general there are three simple cases to conclude your endeavor with an impactful thesis:
• Introduce a new problem to the community that no-one has considered before which will attract attention. Since you are the first working in this area, definitely your work once published will be the state of the art and will be highly cited in the future (everyone will be able to do better than yourself in the future).
• Consider a highly challenging problem with a lot of people working on it and come up with "the" solution. This is in general hard to achieve both because of the difficulty of the problem as well because of the pressure of publications which will force to fragment your work and decrease the possibility of long term planning of your work.
• Well, look on other areas, especially in sciences we keep re-inventing the wheel. If you are able to find some theoretical work in them and being able to apply it successfully in one of the core problems of your domain then you are there!!! I will use a quote from a highly respectful colleague and friend: "be the last to re-discover the idea".  
• Consider a visible (publications/citations/standing-wise) research lab where you are sure that your work will definitely get the necessary exposure. Avoid when possible overl-populated labs, most likely you will not get the supervision/help you are expecting even if your work will be quite visible. The choice of PhD thesis advisor is critical, either go with a rising star (junior faculty who shows extreme potentials) or a highly established professor. Practically you are trading precious guidance versus visibility.  
• Be sure that you can work with your supervisor and humanly there is a reasonable connection. You will be spending 4-5 years working with him/her there should be some minimal understanding between both parties. Check/contact a reasonable number of former students and ask them whether things went smooth  and average the input opinions you are getting. Be sure that you can have a balanced life during this period and the location of your PhD doesn't have a negative impact on your performance.

and have always in mind that the success depends on you, and research quality is the result of hard-working/highly-motivated researchers with appropriate background. Also recall that 90% off the research work is incremental and there is nothing wrong with that, it is a necessary step towards ultimately solving original problems and your work is of great value even if you didn't finally solve the problem that you were assigned to.

Five Questions/Answers for your High-end High-tech Scientific Job Interview

High-end high-tech scientific jobs hunting has become really competitive task over the last two decades mostly due to the international context of the huge as well as medium conglomerates. In the most general case 3-4 equally exceptional applicants are considered for a given position and what could really make a difference is the job interview. Being prepared is a key element to success, and being able to put your application in the context of the technical job description as well as sell appropriately your personal skills. There are five questions that for sure you will be getting and you should be prepared to answer: 

• Why did you apply to this position? well, the answer should be obvious: repeat your academic and professional background and explain how it is well aligned with the job description. Furthermore, put into the context the positive characteristics of the company you are applying for...things like international mobility, access to real scientific problems, societal impact, diversity of problems are usually appreciated. 
• Did you apply to other positions and if yes how these applications are rated with respect to the one you are interviewing? Well, if it happens to be your first application, just mention that you are still in the process of deciding what is the best professional path, but given the exceptional perspectives of the advertised job you have decided to apply. If it happens that you have applied, interviewing with direct competitors, there is nothing wrong providing this info while explaining though the the job description is different (pay attention to do not provide direct preference but put emphasis on the differences between them).
• What are your long-term perspectives in the company? The key here is to be concrete, show ambition while not being aggressive. Clearly state that your immediate objective is to exceed the expected duties of the job description while better understanding the needs of the company from your unit. Once this has happened then getting more responsibilities and assuming leading technical lead in this unit would be appropriate and eventually once you have learned and understood the overall structure and the needs of the company  move towards a more operational role assuming leadership in one of the business units. Put emphasis if appropriate on international and thematic mobility.
• Can you please provide me a short description of your work/curriculum? Depending on the interlocutor you should shift from abstract but relative clear answers to technical ones. Please recall that most of the senior management is not technical anymore but still have a "scientific" background. Make efforts to explain them in the context of science what you have done without bombarding them with useless details that anyhow cannot be understood. Be clear and direct, short answers with "interesting" content are much better than long ones with useless content.
• What are your salary expectations? Never/ever do not answer this question directly. The best possible answer is to refer to people you know with quite similar job and let them know their average salaries. Once this is done, just state that you are sure that a company with the qualities of the one you are interviewing for knows how to appreciate the individual paths and the added value of its employes and therefore you are sure that the offer that you will be getting reflects the quality of your application.  

Please keep in mind that it is important to maintain the discussion whenever is possible and show confidence during the individual meetings. Avoid lengthy answers and repetitions of things, and make all possible efforts to put into context your academic/employment path.  If it happens that you do not have convincing answers to the above questions, do not waste your time and the one of the company, apply elsewhere.

Boosting Research Performance in France with Ten Simple Actions/Ideas

• Convert professors/researchers existing salaries to nine months compensation and allow them to get summer pay through grants to counter effect the absence of internationally competitive salaries
• Increase the expected PhD duration from three years to four years while allowing them to enroll for a fifth year as well while lightening the requirements/expectations regarding the final PhD manuscript (it takes approx six months to be prepared currently)
• Increase the maximum number of PhD students under supervision to eight, and remove the HDR condition as well as the CNU qualifications to become a professor or replace existing elected panels from international transparent committees
• Increase competitive funding, through the Agence nationale de la recherche and remove bureaucratic and completely obscure rules. Reduce the dispersion of the funding agencies and improve the process of grants submission evaluation, as well as make the whole system more transparent and more agile
• Restructure industrial benefits when funding research through competitive calls. Replace the "Industrial Research Credit" with a more competitive funding referring to joint state-industry funding of proposals where industry supports its own funding
• Introduce mechanisms to increase the absorption of PhDs from the industry as well as encourage industry to offer them competitive salaries towards encouraging and motivating excellent students to pursue graduate studies
• Introduce half year sabbatical leaves formally with salary preservation
• Rethink the duration of the probation period prior to become a state/permanent researcher/professor
• Improve the attractiveness of France to foreign professors in particular at the early stage of their career through more flexible and better paid appointments, where promotions are decided at the school level
• Merge teaching departments with research labs while removing the hierarchical/pyramidal structure/organization of research labs and introduce regular rotation on the upper management

Grande Ecoles @ France and their Mutation @ International Educational Ecosystem

Grande Ecoles have been in the French Higher Educational Ecosystem form almost three centuries and have greatly contributing on paving the way to the French industrial revolution. Their success recipe is not that complex: get the best of the best, at a rather mature age (that often corresponds to almost a B.Sc. degree in the Anglo-Saxon system) with an exceptional background on sciences through highly competitive exams and then offer them a general culture academic curriculum towards providing means of further enrichment for their potentials. The system is pretty rounded, and the success has been there mostly due to the exceptional quality of the preparatory classes of the highly competitive admission selection process. 

Such a system was a win-to-win situation for the students/Grande Ecoles. Students were certain to graduate (Grande Ecoles couldn’t allow dropouts - it was harming their reputation - that were extreme exceptions and in many cases the degrees are not associated with grading/distinctions) and their degree could secure a highly successful professional career (mostly nationally) either at the public or private sector. Grande Ecoles through this process managed to build gradually a powerful network/community holding key positions at the private and public sector and their graduates become policy/decision-makers and contributed on putting the grounds for the increasing influence of these Ecoles at the French society. The system at it was had almost no forgetting memory, what was really important was the success admission after the preparatory classes and the graduation from a Grande Ecole in conjunction with the individual career path. 

On the downhill (until recently where we observe a real change of direction), research was insignificant (with some notable but still small number of exceptions) at these schools, international exposure was absent, while at the same time expected students efforts during their studies were “accommodated” to be minimal. I recall discussing with the three different deans of academic studies in three different schools who did explain me that the percentage of the dropouts is so small simply because our students are so well trained and so carefully selected, so even the worst ones could determine the minimal required effort to graduate. These elements have been propagated to the French industrial ecosystem resulting on a substantial diminishment of the importance of “graduate studies” like for example M.Sc. or PhD as well professional mobility and continuous learning. Students at their most productive years, with exceptional abilities were taught that the most important thing is already achieved that was the admission at the Grande Ecole and the rest will come even though in most of the cases excellent academic programs were at their disposal. 

Then, French industrial tissue became international, degrees of Grande Ecoles world-wide became not as important as the ones of the top international schools (MIT, Oxford, ETH,…) due to the lack of visibility of these schools outside France (in particular in the absence of strong research portfolio) and lack of graduate studies which is a natural manner of attracting foreigner students and spreading the reputation of the school worldwide.  The generous support of the French state, gave the possibility to these schools to live beyond their means. International collaborations and double degrees were initiated at proportions (given the number of students) with unfortunately highly unbalanced flow of incoming/outgoing students of same international university-reputation. Furthermore, towards addressing the research gap, the schools have started pursuing more aggressive policy towards convincing their own students to pursue graduate studies which definitely doesn't contribute much on improving result visibility of these schools through imprinting. The important thing is to get excellent graduate admissions from other top-level schools and send your own students to them. These creates a flow and contributes on increasing research perspectives. However, due to the absence of a centrally organized admission system for graduate studies at these schools, their visibility was limited and in most of the cases international admissions are due the reputation of the professor and not the one of the school. The small number of students of these schools (and consequently the number of faculty as well), the wide thematic academic program spread, the lack of research investment and the declining support of the state funding have contributed constraining the impact of their footprint at the international ecosystem. 

What didn’t change though was the amazing/exceptional quality of the admitted students and their extreme potentials. The maturity of these students is often equivalent to the ones of the graduate students at the top Anglo-Saxon Universities.

It is obvious that given the inevitable decline of the state support the current system cannot sustain - unless moving towards the Anglo-Saxon system with the aberration of tuition fees which will be really unfortunate - and a modality culture change should be envisioned. 

International exchange programs with an insignificant return-to- investment are not sustainable. Such collaborations should  be preserved but combine research excellence and academic interest and should involve equitable partnerships. The general engineering culture should be preserved at the beginning of the educational curriculum, however better organization should be envisioned in the final years towards areas where these schools can make the difference. These should be well identified areas with top level academic faculty supporting the ambitions of these schools and targeting emerging / demanding needs of the society from the scientific view point. But most importantly, consensus/mutual understanding/sharing the resources between these schools is inevitable at least regarding the general engineering culture academic program. Differentiation should happen through focus on different research and department majors. Being in all possible areas of engineering is neither possible nor sustainable for these schools. It is better being the leader on a subset of sciences than no-one everywhere.

There is nothing wrong being small (Caltech is a small school both in terms of faculty and number of students), the problem arises when the lack of resources is associated with the dispersion of efforts, a combination that completely diminishes the potentials of the effort. 

To conclude, these schools should mutually mutate to a new era, become more competitive at international level both in terms of academic focus/offer as well as research portfolio.  This can only happen through a profound change of the vision of decision/policy-makers as well as a colossal progress on the mentalities of the past, current and future graduates of these schools.