How do we support continued development and maintenance of the microscope? What new projects should we pitch, and how do we balance that with supporting the crucial but less saleable activities like updating the documentation and ensuring high code quality?
We talked a lot about how to fund the project - there are perhaps two differnt perspectives, from end users (@dgrosen gave himself as an example) and the core development team. Finding funds to build microscopes is OK, but supporting ongoing development/maintenance can be hard. We’d be delighted to see time for the core team included in other people’s grants, e.g. to add new features, support reproduction, and continue development.
Reproducibility is super important here - being able to have lots of replicates is good for science. Making the microscope reproducible is a useful angle for funding: funders are increasingly aware that reproducibility of experiments is a problem, and open hardware might offer a solution to many aspects of this. That might not fund OpenFlexure, because it’s quite specific, but it might be a route to funding some of the underlying infrastructure like GitBuilding or HardOps. How do we pitch these to funders, which funders, and how big a userbase do we need to find before it’s viable.
Career strutures at Universities don’t lend themselves well to this - currently Richard has a permanent job but most developers (e.g. @j.stirling) are only employed in a short-term way. That means it’s not viable to support 10-20% of someone’s time because it doesn’t make a whole job.
Being ambitious and pitching multi-million pound(/dollar) projects is one way to get some long chunks of time for development. This probably takes a network of people, including both instrumentation people and lots of applications. EU Horizon projects are exciting but the UK’s departure is a problem. Wellcome Trust is a funder we should engage with more and we are hoping to do this. Foundations like Sloan or Gates might be a good option, but it’s tricky to figure out how to engage effectively, and to pitch the right project at the right time. Maybe engaging better with GOSH would be a good way to form those links.
Day 2 has ended at #openflexurecon22
@r.w.bowman , reproducibility hits the nail. Some thoughts which might help addressing this important requirement.
- If not yet done so, You might want to engage with other faculties, disciplines; e.g. faculty of engineering, mechanical and electronics, and supervise a thesis to e.g. consolidating OpenSCAD version and OFM/DS/BS. Assuming you want to stick with OpenSCAD. This will reduce maintenance burden and reduce sources of error, in other words funding is better spent.
- Consolidating the entire electronics creating a homogenous module (motor control, LED, illumination, structured light,…many more functions to come, and each with some needs for electronics). This will lower the threshold of acceptance, i.e. help growing the user base (medical sciences, diagnostics,…)
- You might want to protect (in a sense of design ownership) the core flexure mechanism so funding can be allocated to the core. Why not providing 3mf models instead of the outdated STL? 3mf is digested by all slicers and has big advantages over STL. See [Wikipedia] or (3D Manufacturing Format - Wikipedia) or Prusa I personally would prefer a parametric model in Fusion360, but that’s another story.
- I know you are working on this. (Automatic) calibration and subsequent test runs. Have you though about a "Quality benchmark, which is submitted to your server - the owners’ site - to build up a database of installations and respective quality - in lack of a better term an OFM benchmark? Investors like to understand where and how their money is being used.
- I like the Prusa printer print bed procedure, it scans 7x7 points on the bed and compares the height, if out of limits an error is thrown. What about a startup routine where the OFM travels left, right, up and down ending at the center stage? With a similar heuristic - quality check - users can be confident that the microscope does well and can be used, starting at/in a reproducible state.
As said, these are a few immediate thoughts to this important topic.
Cheers,
Jörg