Lasse Schlör
lasse-schloer@unibo.it

MMM General Assembly
2026-04-13

Open Source Software and Workflows

Funded by the European Union’s Horizon Europe Marie Skłodowska-Curie Actions scheme, GA no. 101120256.

Open Source

What does “Open Source” mean?

Open Source means:

The instructions for making something are public
Anyone is allowed to use, study, modify, and share them

Examples:

Furniture, bicycles, farm tools, prostheses, …
Cosmetics, soaps, food products, …
Computer hardware

Antonyms:

Proprietary
Closed source

Free and open source software (FOSS)

The Free Software Foundation defines four essential freedoms:

To run the program
To study and change the program in source code form
To redistribute exact copies
To distribute modified versions

Selling FOSS

“Free” not as in “free beer”, but as in “free speech”.

Company sells	Example
Support and consulting	Red Hat
Hosted service	WordPress.com
Advanced features	GitLab
Dual licensing	MySQL
Custom development	Collabora
Long-term maintenance	Canonical

Why Open Source?

Headlines: EU countries pushing for digital autonomy

Los Angeles Times: France dumps Zoom and Teams as Europe seeks digital autonomy from the U.S.

Heise: France's plan: Away from Windows, towards Linux

Reuters: Dutch parliament calls for end to dependence on US software companies

Bloomberg: Nordic Edition: Northern Nations Start to Pull Plug on US Tech Giants

Interoperability

Connecting closed tools is permission-based.
Connecting open tools is possibility-based.

Examples of vendor lock-in:

Poor integration of iCloud services on non-Apple devices
Microsoft bundling Teams with Office
Kindle e-books not readable outside Kindle
HP printers refusing third-party cartridges
…

EU Digital Markets Act

Embrace, extend, extinguish

Wikipedia: Embrace, extend, and extinguish

Right to repair

Closed tools' lifetime is vendor-controlled.
Open tools are fixable.

Examples of anti-repair design:

Dealer-only software locks on John Deere tractors
No disassembly of Microsoft Surface laptops
Samsung washing machines requiring whole-tub replacement
…

EU Directive on repair of goods

Headlines: Planned obsolescence

BBC: Apple pays out over claims it deliberately slowed down iPhones

Ars Technica: Trains were designed to break down after third-party repairs, hackers find

The Connexion: ‘Built-in obsolescence’ study targets women’s tights

Security, privacy & transparency

“We love our phones, but we do not trust them. And love without trust is the definition of an abusive relationship.”

— Maria Farrell

Closed source asks for trust.
Open source allows verification.

EU-FOSSA: Free and Open Source Software Auditing

Headlines: Security, privacy & transparency

Google settles $5 billion privacy lawsuit over tracking people using 'incognito mode'

TikTok fined €345m over children's data privacy

SPARC Report Urges Action to Address Concerns with ScienceDirect Data Privacy Practices

Collaboration & extensibility

“Further, science is a collaborative effort. The combined results of several people working together is often much more effective than could be that of an individual scientist working alone.”

— John Bardeen, Nobel banquet speech

Closed tools limit development participation.
Open tools are shared building blocks.

EU Open Source Programme Offices Network
EU Open Source Solutions Catalogue

Reproducibility & replicability

Reproducibility: Same data, same methods
Replicability: New data, same question

“More than 70% of researchers have tried and failed to reproduce another scientist's experiments, and more than half have failed to reproduce their own experiments.”

— Monya Baker, Nature

Closed tools are black boxes, hiding methodological details.
Open tools expose the full workflow, without license barriers.

Downsides of open-source software

Funding problems
Fragmentation
No guaranteed accountability
Licensing complexity

Workflows

What is a scientific workflow?

A defined sequence of repeatable steps
Each step transforms data or materials
Outputs from one step feed into further steps

Makes analyses repeatable
Makes results reproducible
Facilitates checking, tracing, reusing, and sharing

Workflow example: Coffee

Planning a workflow

Future-proofing:

What might the workflow look like a year from now?
Could you still understand this workflow a year from now?
Could somebody else understand it without your help?

Dataset detective work

ID	RES	VALUE	NEW_VALUE	FLAG	__ETA__
001	Y		N.A.	64	0.1
002	Y	12.4	12.8	64	0.1
003	N	13.1	13.4	96	0.1

Version control by emotional intensity

filename	used_for_analysis
results.csv	no
results_v2.csv	no
results_final.csv	FALSE
results_final2.csv	no
results_reallyfinal.csv	maybe

Eyewitness testimony

# nobody knows what this does
# Bob wrote this in 2019
# ask him if you can find him
def do_it(x):
  y = []

  # loop over input
  for i, z in enumerate(x):
    # calculate result
    y += [h(q(z), y[-1:] or y)]

  return y

Best practices for reproducible workflows

Name things clearly and consistently
Write documentation:

“How must this step be carried out exactly?”
“What does this variable measure?”
“What is this script for?”

Reduce hidden dependencies between steps
Automate repetitive work if possible
Prefer open-source tools where practical
Choose suitable, open data formats
Keep things traceable and backed up

Version control

tracks changes to files over time:

Trace text, figures, datasets, code
See what changed, when, and why
Recover earlier versions safely
Reduce confusion from many copies
Collaborate without overwriting each other's work
Keep an online project backup

Computational reproducibility at scale

1.	Dependency snapshots Freeze package/library versions conda, renv, requirements.txt, uv, …
2.	Containers Freeze the software environment Docker, Apptainer/Singularity, …
3.	System snapshots Freeze the whole system (if containers aren't an option) GNU Guix, NixOS, …
4.	Workflow managers Freeze the full workflow Nextflow, Snakemake, …

Privacy-enhancing technologies

Federated learning
Homomorphic encryption
Secure multiparty computation
Differential privacy

Emerging methods for privacy-preserving data analysis
Allow statistics without sharing raw records
Can support open science when open data are impossible

Conclusion

Open Source benefits society broadly and science in particular

Reproducibility starts with well-designed, open workflows

Give a try if you haven't already!

Thank you!

This presentation is open source. Its source code can be found at:
https://github.com/publik-void/publik-void.github.io

Go to index (relative)
Go to index (online)