Useful case studies of problems caused by closed source hardware?

General
1

Hello GOSH community,

@briannaljohns and I were talking about open science hardware examples today when I realised that I don’t have good negative case studies where closed source hardware is a big problem.

What I mean is, are there stories like “Sally is a scientist who relied on a piece of hardware to do their work, but because it’s closed source, it made her life very difficult because [insert specific example here]”.

Any suggestions? Perhaps stories you like to use when giving talks/presentations about open science hardware?

3 Likes
2

One of the problems with “closed source” hardware is that you will void the manufacturer’s warranty if you have to repair something without going through their repair service and service contract.

2 Likes
3

Yes.

Non-repairable microscope auto-focus module

The lab where I started my PhD had a fully automated epi-fluorescence microscope (IX81).

We relied heavily on a hardware auto-focus module (ZDC) for quantitative time course imaging of single yeast cells.

Three years ago it stopped working, and we have been unable to get someone to repair it. The product is no longer supported, and Olympus no longer provides support of any kind in our country.

This means that the quality of the time series data has a much larger dispersion, which is bad for the kind of questions we asked. For example, we can no longer reliably quantify fluorecence localization over time.

For the record, yeasts are tiny, and consistent focus is very important. Hundreds of images are acquired during each experiment, so automated focusing is important.

Ice on the camera sensor

Same story, different device.

The seal in the camera broke, and no one knows how to repair it. The camera had been performing excellently for decades until that moment, when atmospheric air entered the chamber and ice crept onto the CCD sensor (preventing any decent imaging).

In theory it has a service port for resetting the conditions it requires, but the manuals mention nothing, and the company does not offer service in our country. The more knowledgeable folk around did not want to risk doing surgery on it without offering guarantees.

This set us back a year, until we were forced to spend a large part of the lab’s funds on a new camera. The old camera is now sitting on a drawer, fully functional, waiting for someone that knows how to do some maintenance.

Overall

As usual around here, the larger unavoidable cause for all of this is a general scarcity of resources, that makes everything work like #$%&/.

Then again, the companies are really secretive, and will not let out any useful information for repair (even when they will not offer to do it themselves).

So, because those problems will not be solved soon enough, from a practical perspective open source hardware is the best way to go. Had our equipment been OScH, we would have been spared of a lot of trouble.

The microscope is our main observation tool, so we are forced to lower the bar of what we can achieve, and settle for less impact (and then less funding :face_with_spiral_eyes:).

3 Likes
4

:pray: Thank you @jbrennerFIT and @naikymen, great points and examples! Sounds like they are related to forced obsolescence, lack of right to repair, and proprietary vendor lock-in in terms of distribution and support services. All important for sure.

Any other case studies of the downsides of closed source hardware in scientific research?

5

The biggest and most obvious limitation of closed source hardware is that it costs so much that you have to write a proposal to a government agency or foundation to fund something that you might be able to build and fund yourself. I want to reduce government dependency.

Moreover, by buying the equipment rather than building it, the student loses the learning opportunity. In my nanotech minor, before we started our own making course (https://fit.instructure.com/courses/604044/assignments), our students were not getting jobs at the companies that make the expensive equipment because they didn’t know how to make the equipment. That is no longer an issue.

Best regards,
Jim Brenner

1 Like
6

If you have t o write a proposal to fund closed source hardware, the time frame for receiving it, even if funded, is so long that the student who wanted it to do it has already taken a job somewhere else!

1 Like
7

Maybe this is more of a story of overcoming the problems of an old commercial system, but it is the first I thought of in this regard… When we started planning for our Corona Detective project (which uses isothermal amplification and fluor detection to diagnose infection status for Covid-19, even simply from saliva), we needed to get real-time amplification data to chose our best conditions (esp magnesium) and primer sets of the reaction. It was 3 years ago, and Hackuarium had already received an ancient ‘real time’ machine (an ABI Prism 7000 that actually makes tifs to build its fluorescence traces!) from the University of Lausanne, but how to get it to work for us was a challenge. 1st we needed to be able to run it - there was a cd with the program, but we needed to make a windows2000 virtual machine to be able to load it. Then, we also managed to get a newer program that could look at the obtained data in my laptop. The worst however was the calibration problem. The offer from ThermoFischer was outrageously high and I sent many begging emails to try and find out what should be in the 96 well calibration plate to do it properly. Luckily, in the end, I just decided to try fluorescein in all the wells (all at the same concentration), and it did the trick! We could run our reactions and compare everything… Happily, our validations with clinical samples were really excellent, but our attempts to make a pro Proof of Principal batch slowed everything down too much… What we learned is mainly for the next pandemic?? =P

1 Like
8

Dear @hpy , I know this isn’t so pertinent here (for two reasons, a) that it’s also a software issue and b) it’s practice rather than research) but I thought i’d mention some of my experience with a Medical Physics department at an Oncology Center, the sheer amount of money they have to spend and general fuss that comes when their LINACS came to the end of their lifetime. It’s also a hardware problem as you also have to change your hardware in order for everything to be compatible, you have to be trained to use the new equipment, you may even have to change some of your other equipment because of lack of interoperability, etc. (I hope, once more that this is even slightly pertinent).

2 Likes
9

Coming from precision metrology, there were always certain cases where you had to study instruments empirically to understand their limitations as you didn’t have access to the design. However it wasn’t proprietary hardware that moved me towards the need for Open Science Hardware…

If you look at those in academia who are pushing the boundaries of instrumentation (for example when I was working on measuring the Universal Constant of Gravitation or Big G as we called it), the limit is that scientific groups kept their designs secret. They published enough that the concept could be understood. But not enough that it could be replicated, not enough that it its uncertainties can be assessed. This is bad just science.

For the case of Big G, measurements don’t agree, there are deep complex theories about why, that talk about dark energy candidates, chameleon particles. However, I am pretty sure that the actual reason lies in uncertainties that are fundamental to the instruments, this has been the case for other experiments in the field, but getting the details to show this is like getting blood from a stone.

We can’t have good science if we don’t have Open Science Hardware.

2 Likes
10

Thank you @Rachel, @haris, and @julianstirling for the additional input! Really loving hearing your stories. :heart:

We can’t have good science if we don’t have Open Science Hardware.

This really jibes with what I often say when talking about open science, i.e. “Good science is open science”. I’m planning to use this line in an upcoming talk about OScH at the Edinburgh Open Research Conference in a couple weeks, and will try to learn from your stories!

P.S. I’ve also linked to this thread in my collective knowledge thread, because this stuff is worth archiving!

1 Like