Grant applications
But first let me tell you a bit about how the thing works. One of the ways scientific funding is distributed to researchers in most countries is through competitive proposals: a selection committee will pick the best ones and funding will be awarded to them to perform the research they proposed. In principle this is a sound way to decide what to fund and what not, but over the years many researchers have noticed how the correlation between the strength of a proposal -its scientific robustness and value, the innovation potential, the effectiveness of the proposed methods, and everything else in between- and the chances of it being funded are not very highly correlated.
Why is it possible that a committee of true experts in the field may fail to identify the strongest proposals? Well, one reason is that no committee can read 1000 proposals and rank them: there need to be many different selection committees in place, to absorb the material. For big calls like the EC one I mentioned above, a referee may be involved with ranking just a handful (3 or 5, e.g.) of proposals for one call. And the response of these different committees may vary, regardless of how carefully they have been instructed to stick to rigorous criteria. So in the end when you take the proposals that got the highest scores, they are not necessarily the best ones.
Researchers in some cases are so frustrated by the apparent randomness of the selection process of these big calls, that they have often suggested that the procedure be changed to a less, rather than a more, reason-driven one. A first skimming should reject "really bad" proposals - ones that are evidently sub-par - with the aim of selecting maybe 60% or 70% of proposals that are sufficiently good to be given a chance; and then, a random selection should determine the winners. This is perceived by some to be a more fair procedure. But it also is tantamount to throwing your hands up...
It is also a fact that nowadays all proposals are much better written than they used to be - because of ChatGPT and similar models, that can provide much more than grammar and syntax checking. They can do fact-checking, may propose text rearrangement, can spot redundancy and incoherence, and suggest ways to tick all boxes of the requirements that your text must comply with. So if you are a referee, you may find it really hard to rank proposals as they all look good. But maybe this is a good evolution, coming to think about it: as the referees can really concentrate on the scientific value of the ideas per se, rather than having to play the game of spotting shortcomings that have more to do with imperfect writing skills or other irrelevant aspects.
Innovation, innovation
The call I submitted my projects to is called pathfinder for a reason - it aims to fund projects that can start a research direction leading to innovations producing societal benefits in the end. Projects suitable for this funding line must be ones that are at a low "technological readiness level", namely one to four. The scale, developed by NASA in the 1970ies, goes from 1 to 9, where 1 is the idea, and 9 is the generic use of the final products that employ the technology developed from the starting idea (this is my own spin - for a more formal definition which is less specific to the kind of setting we are discussing here, see wikipedia). For pathfinder, projects must aim at TRL 1 to 4, i.e. reach a level of proof-of-principle, and maybe a first attempt at creating a prototype of some new technology. A demonstrator, so to speak.
But what does my research have to do with creating new products? Well, yes, I do work in fundamental science, and there, as you know, we are not really very concerned with the applications of what we study. We want to discover how the world works, and this means everything and nothing in this context. If we, for example, discover a new particle, there is absolutely no way that we can exploit that new piece of information to create a working new technology. But wait -sometimes that is the case instead. If we discover that the new particle carries a new force of nature, we might find ways to exploit that in some way. We basically have zero idea of what we are going to find.
The classic example of this is Roentgen and his x rays: he did not really know what he was doing when he experienced with vacuum tubes and electric currents, but when he saw that inside vacuum there was a propagation of some kind of rays generated by accelerated electric currents, and that those rays were highly penetrating through matter, he could use them to image how bones were fractured in injured limbs of patients - which helped to fix them. But there are a large number of other spinoffs that derive from the discoveries physicists made by studying Nature for the sake of it.
And then there is the technological developments side of things. In order to do our experiments we are brought to push technology to the limit - as measuring subnuclear particles is not such an easy feat. That technology can then be used for applications that are useful for society. The simplest example? Proton-computed tomography is a technique that allows physicians to scan the interior of the body of patients with protons from an accelerator, and then irradiate areas that need to be burned away - like tumours - with the same beam.
E-MODES:
End-to-End Machine-learning Optimization of the Design of Experiments and Systems
So it is entirely natural for a particle physicist to think at the innovations side of things, and to ponder on how the European Community can be interested in his or her research. In my case, I have submitted one project as a coordinator, and I participated in a submission to another project as a PI of one involved institution. The projects of this kind are in fact "multi-institution" ones: we leverage the different skillset of scientists in different universities, research centers, or industries, and propose collaborative research that involves the participation of all. The focus is also a lot on interdisciplinarity: the more a proposed research topic can exploit diverse competences, the better.
The project I am coordinating is called E-MODES, and it stems from activities I have started five years ago in trying to demonstrate how AI can produce a second revolution in fundamental science, by allowing us to perform co-design of the hardware and software of our experiments. Co-design means you design at the same time the instruments that collect your data, and the software that extracts information from those data. By optimizing together the parameters of both systems you can achieve an alignment toward higher performance - something that separate optimization cannot do. But co-design is extremely hard, especially in particle detectors that include thousands, if not tens of thousands of construction choices, materials, geometries. And the software part is no easier, as it must be perfectly tailored to the kind of data you elicit from the physical process of interest.E-MODES wants to demonstrate how to achieve co-design by hybrid computing strategies (digital and neuromorphic) and hybrid optimization techniques (gradient-based and evolutionary). We aim to demonstrate the technology on a few interesting applications where end-to-end optimization can bring large gains (hadron therapy, muography, and innovative calorimetry) as well as develop a general software that may be easily customizable to arbitrary tasks. It is a really ambitious project, and if we get funded we will extremely busy for the next four years - but it would be a fantastic journey. E-MODES includes 11 nodes from European universities in Italy, Spain, Sweden, Germany, Belgium, and the Netherlands, as well as two industries.
We will know if the EC referees like E-MODES in about four to five months. In the meantime, of course I am still going to work in the same direction, with no funding but, of course, a lot of willpower!
In EC funding calls like the one I participated in, for instance, the success
Comments