Principles of Good Scientific Practice in Geoinformatics, October 15, 2026, TUD

<!-- First slide: Left blank; title and author rendered from _config.yml --> <!-- .slide: data-header="clean" --> Note: - Welcome the committee, colleagues, and student representatives. - provide some calm: good scientific practice not about memorizing commands, but about building habits that protect research from falling apart - slides as a preparation for independent Master research. - Transition: Defining our formal learning objectives.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel: Bündig und unbeweglich neben dem Logo --> <div style="text-align: left; margin-left: 235px; margin-top: 0px; margin-bottom: 22px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.35em; line-height: 1.15; letter-spacing: -0.5px;"> Learning Outcomes </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.8em;"> From Fragile Workflows to Accountable Synthesis </div> </div> <!-- 4 Karten: Stabiles Layout, kein Versatz, ruhiges Fade-in --> <div style="display: flex; flex-direction: column; gap: 9px; max-width: 1060px; width: 92%; margin: 0 auto;"> <!-- Pillar 1 --> <div class="fragment" data-fragment-index="1" style="display: grid; grid-template-columns: 240px 1fr; align-items: center; gap: 30px; padding: 10px 22px; background: var(--tud-bg-light, #F8F9FA); border-left: 4px solid var(--tud-blue, #00008C); border-radius: 4px; box-shadow: 0 1px 4px rgba(0,0,0,0.03); text-align: left;"> <div style="line-height: 1.25;"> <div style="font-size: 0.68em; font-weight: 700; color: var(--tud-text-muted, #566371); letter-spacing: 0.8px;">PART I</div> <div style="font-size: 0.95em; font-weight: 800; color: var(--tud-blue, #00008C); letter-spacing: 0.3px;">DIAGNOSE</div> </div> <div style="font-size: 0.83em; color: var(--tud-text, #323F4B); line-height: 1.35;"> <strong>The Reality Check:</strong> Administrative friction, GUI traps & spatial dependencies </div> </div> <!-- Pillar 2 --> <div class="fragment" data-fragment-index="2" style="display: grid; grid-template-columns: 240px 1fr; align-items: center; gap: 30px; padding: 10px 22px; background: var(--tud-bg-light, #F8F9FA); border-left: 4px solid var(--tud-blue, #00008C); border-radius: 4px; box-shadow: 0 1px 4px rgba(0,0,0,0.03); text-align: left;"> <div style="line-height: 1.25;"> <div style="font-size: 0.68em; font-weight: 700; color: var(--tud-text-muted, #566371); letter-spacing: 0.8px;">PART II</div> <div style="font-size: 0.95em; font-weight: 800; color: var(--tud-blue, #00008C); letter-spacing: 0.3px;">STRUCTURE</div> </div> <div style="font-size: 0.83em; color: var(--tud-text, #323F4B); line-height: 1.35;"> <strong>The Resilient Archive:</strong> Open formats, Single Source of Truth & Git provenance </div> </div> <!-- Pillar 3 --> <div class="fragment" data-fragment-index="3" style="display: grid; grid-template-columns: 240px 1fr; align-items: center; gap: 30px; padding: 10px 22px; background: var(--tud-bg-light, #F8F9FA); border-left: 4px solid var(--tud-blue, #00008C); border-radius: 4px; box-shadow: 0 1px 4px rgba(0,0,0,0.03); text-align: left;"> <div style="line-height: 1.25;"> <div style="font-size: 0.68em; font-weight: 700; color: var(--tud-text-muted, #566371); letter-spacing: 0.8px;">PART III</div> <div style="font-size: 0.95em; font-weight: 800; color: var(--tud-blue, #00008C); letter-spacing: 0.3px;">CONSTRUCT</div> </div> <div style="font-size: 0.83em; color: var(--tud-text, #323F4B); line-height: 1.35;"> <strong>Freezing Computation:</strong> Managing dependency hell, containers & CI/CD automation </div> </div> <!-- Pillar 4 --> <div class="fragment" data-fragment-index="4" style="display: grid; grid-template-columns: 240px 1fr; align-items: center; gap: 30px; padding: 10px 22px; background: var(--tud-bg-light, #F8F9FA); border-left: 4px solid var(--tud-blue, #00008C); border-radius: 4px; box-shadow: 0 1px 4px rgba(0,0,0,0.03); text-align: left;"> <div style="line-height: 1.25;"> <div style="font-size: 0.68em; font-weight: 700; color: var(--tud-text-muted, #566371); letter-spacing: 0.8px;">PART IV</div> <div style="font-size: 0.95em; font-weight: 800; color: var(--tud-blue, #00008C); letter-spacing: 0.3px;">EVALUATE</div> </div> <div style="font-size: 0.83em; color: var(--tud-text, #323F4B); line-height: 1.35;"> <strong>The Generative Frontier:</strong> Governing LLM non-determinism & preserving audit trails </div> </div> </div> Note: - Four-stage competence model (Bloom). - Moving from diagnostic awareness to evaluation. - Covers data management, code, and AI. - Grounded in DFG Code of Conduct. - Transition: Why do we need this? Examining DFG Guidelines 7 and 12. --- <!-- .slide: data-header="clean" --> ### DFG Code of Conduct (Guidelines 7 & 12) <div style="display: flex; gap: 40px; margin-top: 50px;"> <div style="flex: 1; border-left: 5px solid #004d40; padding-left: 20px; text-align: left;"> <h4 style="margin: 0; color: #004d40;">Guideline 7</h4> <p style="font-size: 1.2em; font-weight: bold; margin: 10px 0;">Persistent Code</p> <p style="color: #555; font-size: 0.9em; margin: 0;">Source code must be <em>citable, persistent,</em> and <em>documented</em>.</p> </div> <div style="flex: 1; border-left: 5px solid #004d40; padding-left: 20px; text-align: left;"> <h4 style="margin: 0; color: #004d40;">Guideline 12</h4> <p style="font-size: 1.2em; font-weight: bold; margin: 10px 0;">Verifiable Steps</p> <p style="color: #555; font-size: 0.9em; margin: 0;">Every production step must be clear enough to be <em>reviewed and assessed</em>.</p> </div> </div> <p style="margin-top: 60px; font-size: 0.85em; color: #888;"> How do we maintain this standard across complex GIS workflows, spatial dependencies, and administrative deliverables? </p> Note: - DFG Guideline 7 mandates persistent software. - Guideline 12 requires verifiable production steps. - Legal relevance for planning and assessments. - Core question: How to operationalize standards? - Transition: In theory, everyone agrees.. but daily student life looks like this.

<!-- .slide: data-header="clean" --> <img src="images/python-dependency-management.png" style="height: 480px; width: auto; object-fit: contain;" alt="Python Environment Superfund Site"> ### "It worked on my machine six months ago..." <div style="font-size: 0.5em; color: #777; margin-top: 6px;"> Illustration: Randall Munroe (xkcd #1987: "Python Environment"), CC BY-NC 2.5 ^{<a href="https://xkcd.com/1987/" class="smallsuper"></a>} </div> Note: - Classic failure mode of unversioned environments. - Python packages collide across operating systems. - Results shift silently upon library updates. - Technical debt threatens research reproducibility. - Transition: De-bunking common misconceptions about reproducibility.

<!-- .slide: data-header="clean" --> ## What Reproducibility is Not Note: - Reproducibility is widely misunderstood across academia. - Often dismissed as secondary bureaucratic overhead. - We must clarify foundational scientific definitions. - Transition: Examining Myth 1: The Afterthought. --- * **Myth 1: An Afterthought.** *"I will conduct my analysis first, and clean up the files next week."* → Post-hoc documentation almost never happens. Note: - Post-hoc cleanup rarely succeeds under deadlines. - Scientific provenance requires recording from inception. - Undocumented analytical steps are permanently lost. - Transition: Examining Myth 2: Binary reproducibility. --- * **Myth 2: A Binary State.** It is not black-and-white, but a spectrum from unorganized scripts to executable compendia. Note: - Science operates on a continuuum of contexts. - Extends from code-sharing to full replication. - Roger Peng (2011) foundational Science paper. - Transition: Examining Myth 3: Coders only. --- * **Myth 3: Exclusive to Software Engineers.** It governs the lineage of ideas, planning reports, environmental metrics, and text synthesis. <div style="font-size: 0.55em; color: #777; margin-top: 25px;"> Peng, R. D. (2011). <em>Reproducible Research in Computational Science</em>. <br>Science, 334(6060), 1226–1227. <a href="https://doi.org/10.1126/science.1213847">[↗]</a> </div> Note: - Provenance applies equally to narrative synthesis. - Environmental impact assessments require identical rigor. - Tracking model and metric origin systematically. - Transition: Why Geoinformatics faces specific risks.

<!-- .slide: data-header="clean" --> ### Why Geoinformatics Faces Unique Vulnerabilities <div style="display: flex; justify-content: space-between; gap: 20px; margin-top: 50px; text-align: left;"> <div class="fragment" data-fragment-index="1" style="flex: 1; padding: 15px; background: #fafafa; border-top: 4px solid #b71c1c;"> <h4>1. Deep C/C++ Stack</h4> <p style="font-size: 0.85em; color: #555;">GDAL, GEOS, and PROJ hide behind Python code.</p> </div> <div class="fragment" data-fragment-index="2" style="flex: 1; padding: 15px; background: #fafafa; border-top: 4px solid #b71c1c;"> <h4>2. Silent Drift</h4> <p style="font-size: 0.85em; color: #555;">Datum shifts silently alter geometries without raising errors.</p> </div> <div class="fragment" data-fragment-index="3" style="flex: 1; padding: 15px; background: #fafafa; border-top: 4px solid #b71c1c;"> <h4>3. The GUI Trap</h4> <p style="font-size: 0.85em; color: #555;">Clicking tools in desktop GIS leaves no audit trail.</p> </div> </div> <div style="font-size: 0.55em; color: #777; margin-top: 25px;"> Reference: Kedron, P., Li, W., Fotheringham, S., & Goodchild, M. (2021). IJGIS, 35(3), 427–445. <a href="https://doi.org/10.1080/13658816.2020.1802032">[↗]</a> </div> Note: - Spatial data relies on compiled C-libraries (many from the 1990s). - GDAL, GEOS, and PROJ run underneath. - Datum shifts silently alter spatial geometries. - GUI workflows leave zero analytical logs. - Transition: The pitfalls of point-and-click GIS.

<!-- .slide: data-header="clean" --> ## The GUI: A Trap <img src="images/21_hot_spot_02.gif" style="height: 360px; width: auto; object-fit: contain; border: 1px solid #ddd;" alt="GUI Interaction Chaos"> <div style="font-size: 0.55em; color: #777; margin-top: 15px;"> Example: Recording complex UI interactions in GIS tools (Tag Maps Workshop 2025) ^{<a href="https://slides.ad.ioer.info/tagmapsworkshop_tud_2025/" class="smallsuper"></a>} </div> Note: - Spatial analysis in interactive desktop GIS. - (Example: Tag Maps Workshop, me trying to communicate the workflow) - Chosen buffer thresholds remain completely undocumented. - Results cannot be independently peer-reviewed. - Transition: A historic example from practice. --- <!-- .slide: data-header="clean" --> <!-- .slide: data-background-image="images/08_BurnettParkPWPLA.jpg" --> <!-- .slide: data-background-size="cover" --> <!-- Zitation, sicher verankert unten rechts --> <div style="position: relative; width: 100%; height: 65vh;"> <!-- Zitation: TUD-Dunkelblau Badge mit Frosted-Glass-Effekt --> <div style="position: absolute; bottom: 0; right: 0; z-index: 100; background: rgba(0, 20, 80, 0.72); backdrop-filter: blur(6px); -webkit-backdrop-filter: blur(6px); border-left: 3px solid var(--tud-blue-2, #97C6FF); padding: 8px 14px; border-radius: 4px; text-align: right; font-family: 'Droid Sans', sans-serif;"> <div style="font-size: 13px; color: #ffffff; line-height: 1.4;"> Image: AutoCAD Sheet Manager Guide, 2009, PWP, Berkeley </div> <div style="font-size: 12px; color: var(--tud-blue-2, #97C6FF); font-style: italic; margin-top: 2px;"> A motivated intern </div> </div> </div> Note: - 2009 PWP Berkeley CAD management guide. - Hand-drawn manuals trying to document clicks. - Demonstrates failure of manual GUI documentation. - Outdated already a year later (due to GUI-changes) - Transition: The structural downside of GUIs. --- * Desktop GIS (QGIS, ArcGIS) offers immediate visual feedback and accessibility. * **Downside:** Dialog boxes, manual layer exports, and sliders leave **no audit trail**. * Which threshold was chosen? Which intermediate layer was clipped? Unrecorded. Note: - Point-and-click creates unreproducible analytical dead-ends. - Parameter selection remains unrecorded in memory. - Automation is required for scientific integrity. - Transition: Where research project time vanishes.

<!-- .slide: data-header="clean" --> ### Where Does Research Time Actually Go? Note: - Perceived versus actual researcher time allocation. - Disparity between statistical modeling and wrangling. - Transition: The machine learning roadmap reality. --- <!-- .slide: data-header="clean" --> <div style="display: flex; align-items: center; gap: 40px; margin-top: 20px;"> <!-- Linke Hälfte: Große Grafik + Quelle direkt darunter --> <div style="flex: 1.1; text-align: center;"> <img class="restricted" src="data:," style="max-height: 520px; width: 100%; object-fit: contain; border-radius: 4px;" alt="Machine Learning Roadmap Reality"> <div style="font-size: 0.5em; color: #888; margin-top: 6px; text-align: center;"> Source: Machine Learning Roadmap, Christophe Bourguignat <a href="https://medium.com/@chris_bour/an-extended-version-of-the-scikit-learn-cheat-sheet-5f46efc6cbb/" style="color: #666; text-decoration: none;">[↗]</a> </div> </div> <!-- Rechte Hälfte: Didaktische Kernaussagen --> <div style="flex: 0.9; text-align: left;"> <ul style="font-size: 0.88em; line-height: 1.5; margin: 0; padding-left: 20px;"> <li class="fragment" data-fragment-index="1" style="margin-bottom: 22px;"> <strong style="color: --tud-blue;">Expectation:</strong><br> Building cutting-edge spatial models and statistical inferences. </li> <li class="fragment" data-fragment-index="2" style="margin-bottom: 22px;"> <strong style="color: --tud-blue;">Reality:</strong><br> >50% of time lost to path debugging, data conversion, and format mismatches. </li> </ul> <div class="fragment" data-fragment-index="3" style="margin-top: 25px; padding: 12px 16px; background: #fafafa; border-left: 4px solid #b71c1c; font-size: 0.78em; color: #444; line-height: 1.4;"> Without structured conventions, researchers drown in administrative friction instead of doing science. </div> </div> </div> Note: - Over 50% lost to administrative friction. - Path errors, encoding bugs, and format conversions. - Cognitive load diverted from scientific reasoning. - Conventions liberate attention for real research. - Transition: Real-world case study in Dresden. --- <!-- .slide: data-header="clean" --> ## Principles vs. The Real World ### Case Study: 150 Construction Site Profiles (Dresden) Note: - Real-world environmental impact assessment case study. - City of Dresden building site assessments. - Municipal requirement: Editable Microsoft Word format. - Transition: The quantitative scale of profiles. --- <!-- .slide: data-header="clean" --> <div style="display: flex; gap: 40px; align-items: center;"> <!-- Linke Spalte: Didaktische Kernaussage --> <div style="flex: 1.05; text-align: left;"> <ul style="font-size: 0.88em; line-height: 1.6; color: var(--tud-text, #323F4B); margin: 0; padding-left: 20px;"> <li style="margin-bottom: 12px;"> <strong>Mandated Format:</strong> Client requires editable <code>.docx</code> reports for internal revisions. </li> <li style="margin-bottom: 12px;"> <strong>Dimension:</strong> 150 separate documents × ~150 data cells = <strong>>22,000 discrete values</strong>. </li> <li style="margin-bottom: 12px;"> <strong>Asynchronous Editing:</strong> Values manually altered across administrative desks. </li> </ul> </div> <!-- Rechte Spalte: Dokumenten-Steckbrief --> <div style="flex: 0.95; text-align: center;"> <img src="images/06a.png-1.png" onerror="this.src='images/first-record.png'" style="max-height: 580px; top:-30px; width: auto; object-fit: contain; border: 1px solid #d0d5dd; box-shadow: 0 4px 12px rgba(0, 0, 140, 0.08); border-radius: 2px;" alt="Site profile assessment table"> <div style="font-size: 0.5em; color: var(--tud-text-muted, #566371); margin-top: 6px;"> Exemplary site profile sheet </div> </div> </div> Note: - 150 site profiles with 22,000 values. - Asynchronous manual edits across municipal departments. - Word tables obscure origin and updates. - High risk of undetected data corruption. - Solution (back then): Two-way VBScript Word <-> Excel - Transition: The architectural solution to Word. --- * **1. Canonical Database:** Store all environmental indicators and site metrics in a structured, relational table (GeoPackage / SQLite / CSV). * **2. Plain-Text Narrative:** Write assessment text modules in Markdown. * **3. Automated Synthesis:** Use automated scripts (e.g. Python / Pandoc) to compile the 150 Word documents on demand. * **The Rule:** *Serve your clients their desired format, but never let a binary file become the archive of your knowledge!* Note: - Solution: Relational tables as Single Source of Truth. - Plain-text Markdown for narrative text modules. - Pandoc scripts compile Word documents downstream. - Rule: Deliver formats without archiving binaries. - Transition: The four universal architectural pillars.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel: Sicher platziert, schlanker Untertitel ohne "Methodological" --> <div style="text-align: left; margin-left: 235px; margin-top: 25px; margin-bottom: 20px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.35em; line-height: 1.15;"> The Four Architectural Pillars </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.8em;"> A Compass for Resilient Spatial Research </div> </div> <div style="display: grid; grid-template-columns: 1fr 1fr; gap: 14px; max-width: 1200px; width: 92%; margin: 0 auto; text-align: left;"> <!-- 1. SoC --> <div style="background: var(--tud-bg-light, #F8F9FA); border-left: 4px solid var(--tud-blue, #00008C); padding: 12px 16px; border-radius: 3px;"> <h4 style="margin: 0 0 6px 0; color: var(--tud-blue, #00008C); font-size: 0.9em;"> 1. Separation of Concerns (SoC) </h4> <p style="font-size: 0.78em; margin: 0; color: var(--tud-text, #323F4B); line-height: 1.4;"> Strictly decouple Raw Data, Computational Logic, and Presentation Interfaces. </p> </div> <!-- 2. SSOT / DRY --> <div class="fragment" data-fragment-index="1" style="background: var(--tud-bg-light, #F8F9FA); border-left: 4px solid var(--tud-blue, #00008C); padding: 12px 16px; border-radius: 3px;"> <h4 style="margin: 0 0 6px 0; color: var(--tud-blue, #00008C); font-size: 0.9em;"> 2. Single Source of Truth (DRY) </h4> <p style="font-size: 0.78em; margin: 0; color: var(--tud-text, #323F4B); line-height: 1.4;"> Never duplicate data. Store authoritative records once; generate figures and maps downstream. </p> </div> <!-- 3. Knowledge Longevity (Zero Lock-in) --> <div class="fragment" data-fragment-index="2" style="background: var(--tud-bg-light, #F8F9FA); border-left: 4px solid var(--tud-blue, #00008C); padding: 12px 16px; border-radius: 3px;"> <h4 style="margin: 0 0 6px 0; color: var(--tud-blue, #00008C); font-size: 0.9em;"> 3. Knowledge Longevity (Zero Lock-in) </h4> <p style="font-size: 0.78em; margin: 0; color: var(--tud-text, #323F4B); line-height: 1.4;"> Your thinking is your most expensive asset. Insights must outlive proprietary software. </p> </div> <!-- 4. Calm Tech & HITL --> <div class="fragment" data-fragment-index="3" style="background: var(--tud-bg-light, #F8F9FA); border-left: 4px solid var(--tud-blue, #00008C); padding: 12px 16px; border-radius: 3px;"> <h4 style="margin: 0 0 6px 0; color: var(--tud-blue, #00008C); font-size: 0.9em;"> 4. Calm Technology & HITL </h4> <p style="font-size: 0.78em; margin: 0; color: var(--tud-text, #323F4B); line-height: 1.4;"> Automate mechanical friction so human attention is preserved for synthesis and ethical critique. </p> </div> </div> Note: ~12 Min - Separation of Concerns: Decouple data, logic, interface. - Single Source of Truth: Store records once. - Knowledge Longevity: Avoid proprietary vendor lock-in. - Calm Technology: Preserve attention for ethical critique. - Transition: Entering Part II (Structure): The Filesystem.

<!-- .slide: data-background="#001450" --> <!-- .slide: data-header="clean" style="top: 0 !important;" --> <div style="text-align: center; color: white; padding-top: 45px;"> <!-- Kicker / Überzeile --> <div style="font-size: 0.8em; font-weight: 700; color: var(--tud-blue-2, #97C6FF); letter-spacing: 2px; text-transform: uppercase;"> PART II • STRUCTURE </div> <!-- Haupttitel --> <h1 style="color: white; font-size: 2.3em; margin: 12px 0 16px 0; font-weight: 800; letter-spacing: -0.5px; line-height: 1.1;"> The Filesystem is Forever </h1> <!-- Untertitel --> <h3 style="color: rgba(255, 255, 255, 0.78); font-size: 1.1em; font-weight: 300; max-width: 860px; margin: 0 auto; line-height: 1.45;"> Data Stewardship, Open Formats, and the Provenance of Thought </h3> <!-- Zitat-Box mit klarer typografischer Trennung --> <div style="margin-top: 40px; padding: 20px 28px; background: rgba(255, 255, 255, 0.06); border-left: 4px solid var(--tud-blue-2, #97C6FF); max-width: 1020px; margin-left: auto; margin-right: auto; text-align: left; border-radius: 0 4px 4px 0; box-shadow: 0 6px 20px rgba(0, 0, 0, 0.25);"> <p style="margin: 0; font-size: 0.95em; color: #ffffff; line-height: 1.5; font-style: italic;"> “Human-readable forms of data will outlive all other forms of data and the applications that created them. Period.” </p> <div style="margin-top: 12px; font-size: 0.75em; color: var(--tud-blue-2, #97C6FF); text-align: right; font-style: normal; letter-spacing: 0.3px;"> — <strong>Andrew Hunt & David Thomas</strong>, <em>The Pragmatic Programmer</em> </div> </div> </div> Note: ~12 Min - Hunt and Thomas: plain, human-readable text will outlive every application that created it - Why do we start data stewardship with plain text? - Spatial science must be able to stand up in court ten years from now - Long-term preservation starts at file level. - Transition: Most of us take a long detour before we believe this.. --- <!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Schlanker Titel neben dem Logo --> <div style="text-align: left; margin-left: 235px; margin-top: 20px; margin-bottom: 15px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em; line-height: 1.1;"> The Evolution of Knowledge Management </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 4px; font-size: 0.78em;"> How we try (and fail) to organize scientific thought </div> </div> <!-- Das Meme als Star der Folie: Perfekt zentriert, volle Höhe, unbeschnitten --> <div style="text-align: center; margin-top: 10px;"> <img src="images/expanding-brain.webp" style="max-height: 470px; width: auto; object-fit: contain; border-radius: 4px; box-shadow: 0 4px 16px rgba(0,0,0,0.12);" alt="Galaxy Brain: From 15 apps to plain text notes"> </div> Note: - we try 15 productivity apps - then we build a complex wiki, ... - Enlightenment: Return to plain text Markdown. - Durability beats feature-heavy commercial software ecosystems. - Transition: The daily crisis of unorganized files. --- <!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Kein Titel, keine Unterzeile: Voller Fokus auf das Bild und deine Stimme --> <div style="display: flex; justify-content: center; align-items: center; height: 75vh;"> <img class="restricted" src="data:," style="max-height: 480px; width: auto; object-fit: contain; border-radius: 6px; box-shadow: 0 4px 20px rgba(0,0,0,0.12);" alt="This is Fine Meme - Research data in flames"> </div> Note: - Chaotic folder structures equal technical debt. - Backing up chaotic folders preserves chaos. - Good practice creates resilient personal archives. - Transition: Plain text versus the disposable lens.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 20px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.2em;"> Plain Text vs. Disposable Lens </h2> </div> <!-- Fokus: Reine Bilder, links & rechts --> <div style="display: flex; gap: 24px; justify-content: center; align-items: center; max-width: 900px; margin: 0 auto;"> <img src="images/01_note_raw.webp" style="height: 350px; width: auto; object-fit: contain; border: 1px solid #ddd;" alt="Raw Markdown in Notepad++"> <img src="images/01_note_typora.webp" style="height: 350px; width: auto; object-fit: contain; border: 1px solid #ddd;" alt="Rendered in Typora"> </div> Note: - apologize: Senior Scientist in Geoinformatics.. talking about basic files and folders - everything that follows in this lecture builds up on this - Canonical text file viewed via Typora. - Markdown remains raw, accessible text. - Interfaces are disposable; text data persists. - Zero data loss if viewers disappear. - Transition: Structuring directories with priority prefixes. --- <!-- .slide: data-header="clean" --> <!-- Titel: Schlank & universell --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 20px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Frictionless Habits </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Universal conventions that scale from laptops to servers </div> </div> <div style="display: flex; gap: 35px; align-items: center; max-width: 1080px; width: 94%; margin: 0 auto;"> <!-- Screenshot: Nimmt sich nur die eigene Bildbreite (flex: 0 0 auto) --> <div style="flex: 0 0 auto; text-align: left;"> <img src="images/notes.webp" style="max-height: 385px; width: auto; object-fit: contain; border-radius: 4px; border: 1px solid #d0d5dd; box-shadow: 0 4px 14px rgba(0,0,0,0.08);" alt="Admin directory tree"> </div> <!-- Rechte Seite: Bekommt den vollen restlichen Platz (~600px) --> <div style="flex: 1; text-align: left; font-size: 0.84em; color: var(--tud-text, #323F4B); line-height: 1.45;"> <div style="margin-bottom: 22px;"> <strong style="color: var(--tud-blue, #00008C); font-size: 0.98em; display: block; margin-bottom: 5px; white-space: nowrap;"> 1. Encoding Priorities (00_, 01_) </strong> Overrides the alphabet: pins critical folders to the top and enforces pipeline flow. </div> <div> <strong style="color: var(--tud-blue, #00008C); font-size: 0.98em; display: block; margin-bottom: 5px; white-space: nowrap;"> 2. Temporal Order (ISO 8601) </strong> <code>YYYY-MM-DD</code> makes chronological sorting identical to alphabetical sorting on any system. </div> </div> </div> Note: - `00_IN` pins raw data as immutable. - `01_PROCESS` holds code; `02_OUT` holds exports. - ISO 8601 `YYYY-MM-DD` guarantees native chronological sorting. - Naming conventions scale to server clusters (e.g. Linux) - Transition: Managing complex databases via recipes. --- <!-- .slide: data-header="clean" --> <!-- Titel: Bündig und fokussiert --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 20px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Databases: Archive the Recipe </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Declarative text files outlive complex database clusters </div> </div> <!-- Screenshot als zentraler Held --> <div style="text-align: center; max-width: 920px; margin: 0 auto;"> <img src="images/03_sql_recipe.webp" style="max-height: 380px; width: 100%; object-fit: contain;" alt="Declarative SQL schema files"> </div> <!-- Eine einzige, kristallklare Konklusion --> <div style="text-align: center; margin-top: 18px; font-size: 0.82em; color: var(--tud-blue, #00008C); font-weight: 600;"> Store the code that creates the database — not gigabytes of opaque binary state. </div> Note: - Binary database dumps break across versions. - Archive declarative SQL scripts in Git. - Code creates schemas reproducibly on demand. - Prefixes enforce deterministic schema execution order. - Transition: Managing literature and personal annotations. --- <!-- .slide: data-header="clean"--> <!-- Titel & einzeiliger Untertitel --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 25px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Literature: Asset vs. Disposable Lens </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 4px; font-size: 0.74em; white-space: nowrap;"> Keep things together that belong together — annotations stay in the file </div> </div> <div style="display: flex; gap: 35px; align-items: center; justify-content: center; max-width: 1120px; width: 95%; margin: 0 auto;"> <!-- Linke Seite: Ordner + schwebendes PDF --> <div style="flex: 0.95; position: relative; text-align: left;"> <img src="images/folder_pdf.webp" style="width: 100%; height: 340px; object-fit: cover; object-position: left top; border-radius: 4px; border: 1px solid #d0d5dd; box-shadow: 0 4px 14px rgba(0,0,0,0.06);" alt="Literature folder structure"> <img class="fragment" data-fragment-index="1" src="images/paper_annotated.webp" style="position: absolute; bottom: -18px; right: -15px; width: 54%; max-height: 350px; object-fit: contain; border-radius: 3px; border: 2px solid #ffffff; box-shadow: 0 12px 28px rgba(0,0,0,0.32);" alt="PDF with yellow and red annotations"> </div> <!-- Rechte Seite: Zotero auf exakt gleiche Höhe (340px) gezoomt --> <div style="flex: 1.05; text-align: right;"> <img src="images/zotero.webp" style="width: 100%; height: 340px; object-fit: cover; object-position: left top; border-radius: 4px; border: 1px solid #333; box-shadow: 0 4px 16px rgba(0,0,0,0.15);" alt="Zotero interface"> </div> </div> Note: - Zotero acts as replacable search lens. - Canonical PDFs live in file directory. - Annotations stay embedded directly inside files. - Separation of storage from metadata management. - Transition: Debunking Myth 4: Reproducibility's purpose.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 25px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Myth 4: “Reproducibility is only for others” </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Good scientific practice as personal emancipation </div> </div> <div style="max-width: 860px; margin: 30px auto 0 auto; text-align: center;"> <!-- Zitat: Schlank und dezent --> <p style="font-size: 1.05em; color: var(--tud-text-muted, #566371); font-style: italic; line-height: 1.5; margin-bottom: 25px;"> “Why should I invest time into open formats if only my supervisor reads my thesis?” </p> <!-- Ruhige, elegante Antwort (Fragment) --> <div class="fragment" data-fragment-index="1" style="background: var(--tud-bg-light, #F8F9FA); border-left: 4px solid var(--tud-blue, #00008C); padding: 14px 22px; text-align: left; border-radius: 3px; box-shadow: 0 2px 8px rgba(0,0,0,0.03);"> <div style="font-size: 1.0em; font-weight: 700; color: var(--tud-blue, #00008C); line-height: 1.35;"> Reproducibility is your personal independence. </div> <div style="font-size: 0.82em; color: var(--tud-text-muted, #566371); margin-top: 4px; line-height: 1.4;"> Your thinking and analytical capacity travel with you — independent of changing institutions, proprietary licenses, and software stacks. </div> </div> </div> Note: - Reproducibility is personal independence, not compliance. - Portable workflows travel across changing employers. - Self-reliance from university IT or licenses. - Transition: The structural cost of switching tools.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 15px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> The Cost of Switching </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Why format choices define your lifelong professional freedom </div> </div> <div style="display: flex; gap: 30px; align-items: center; max-width: 1100px; width: 95%; margin: 0 auto;"> <!-- Grafik: Groß und dominant links (65% Breite, linksbündig forciert) --> <div style="flex: 1.45; text-align: left;"> <img src="images/diagram_temp.drawio.svg" style="max-height: 440px; width: 100%; object-fit: contain; margin-left: -25px;" alt="Cost of Switching Graph"> </div> <!-- Rechte Seite: Nur zwei schlanke, unaufdringliche Anstriche --> <div style="flex: 0.75; text-align: left; font-size: 0.85em; line-height: 1.45;"> <div style="margin-bottom: 28px;"> <div style="font-weight: 700; color: var(--tud-text-muted, #566371);"> ✕ ESRI File Geodatabase </div> <div style="color: var(--tud-text-muted, #566371); font-size: 0.85em; margin-top: 3px;"> Proprietary lock-in • Periodic breaking changes </div> </div> <div> <div style="font-weight: 700; color: var(--tud-blue, #00008C);"> ✓ GeoPackage / SQLite </div> <div style="color: var(--tud-text, #323F4B); font-size: 0.85em; margin-top: 3px;"> Guaranteed backward compatibility until <strong>2050</strong> </div> </div> </div> </div> Note: - Changing jobs causes steep productivity loss. - ESRI geodatabases do not guarantee backwards compatibility. - SQLite and GeoPackage guarantee 2050 compatibility. - Open standards secure lifelong analytical freedom. - Transition: Balancing open data with ethics.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 20px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Governing the Commons </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> FAIR baseline <em> and necessary ethical boundaries</em> </div> </div> <div style="display: flex; justify-content: center; gap: 80px; max-width: 960px; margin: 30px auto 0 auto; text-align: left;"> <!-- Linke Spalte: FAIR (immer sichtbar) --> <div style="flex: 1; border-top: 4px solid var(--tud-blue, #00008C); background: var(--tud-bg-light, #F8F9FA); padding: 18px 24px; border-radius: 4px;"> <div style="font-size: 1.15em; font-weight: 800; color: var(--tud-blue, #00008C); margin: 12px 0 16px 0; letter-spacing: 1px;"> F • A • I • R </div> <div style="font-size: 0.84em; color: var(--tud-text, #323F4B); line-height: 1.6;"> <strong>F</strong>indable<br> <strong>A</strong>ccessible<br> <strong>I</strong>nteroperable<br> <strong>R</strong>eusable </div> </div> <!-- Rechte Spalte: ELSA (didaktisch zweistufig aufgedeckt) --> <div style="flex: 1; border-top: 4px solid var(--tud-orange, #C85000); background: var(--tud-bg-light, #F8F9FA); padding: 18px 24px; border-radius: 4px;"> <!-- Klick 1: E L S A ? --> <div style="font-size: 1.15em; font-weight: 800; color: var(--tud-orange, #C85000); margin: 12px 0 16px 0; letter-spacing: 1px;"> E • L • S • A ? </div> <!-- Klick 2: Die Auflösung --> <div class="fragment" data-fragment-index="1" style="font-size: 0.84em; color: var(--tud-text, #323F4B); line-height: 1.6;"> <strong>E</strong>thical<br> <strong>L</strong>egal &<br> <strong>S</strong>ocial<br> <strong>A</strong>spects </div> </div> </div> <!-- Unaufdringliche Normen-Referenzen am Fuß --> <div class="fragment" data-fragment-index="1" style="text-align: center; margin-top: 30px; font-size: 0.62em; color: var(--tud-text-muted, #566371);"> FAIR: Wilkinson et al. (2016) <em>Sci Data</em> <a href="https.//doi.org/10.1038/sdata.2016.18">[↗]</a> • DFG Code of Conduct: Guidelines 10 (Ethics) & 13 (Open Access) </div> Note: ~20 Min - FAIR principles mandate open, reusable data. - ELSA: (ask, reveal) Ethical, Legal, and Social Aspects. - Spatial data trajectories inherently resist anonymization. - Rule: As open as possible, closed as necessary. - Transition: Entering Part III: Freezing computation.

<!-- .slide: data-background="#001450" --> <!-- .slide: data-header="clean" style="top: 0 !important;" --> <div style="text-align: center; color: white; padding-top: 45px;"> <!-- Kicker --> <div style="font-size: 0.8em; font-weight: 700; color: var(--tud-blue-2, #97C6FF); letter-spacing: 2px; text-transform: uppercase;"> PART III • CONSTRUCT </div> <!-- Haupttitel --> <h1 style="color: white; font-size: 2.3em; margin: 12px 0 16px 0; font-weight: 800; letter-spacing: -0.5px; line-height: 1.1;"> Orchestration </h1> <!-- Untertitel --> <h3 style="color: rgba(255, 255, 255, 0.78); font-size: 1.1em; font-weight: 300; max-width: 860px; margin: 0 auto; line-height: 1.45;"> Building Bridges in a Gap-Filled Reality </h3> <!-- Zitat-Box (Buckheit & Donoho / Claerbout) --> <div style="margin-top: 40px; padding: 20px 28px; background: rgba(255, 255, 255, 0.06); border-left: 4px solid var(--tud-blue-2, #97C6FF); max-width: 840px; margin-left: auto; margin-right: auto; text-align: left; border-radius: 0 4px 4px 0; box-shadow: 0 6px 20px rgba(0, 0, 0, 0.25);"> <p style="margin: 0; font-size: 0.9em; color: #ffffff; line-height: 1.5; font-style: italic;"> “An article about computational science is [...] merely advertising of the scholarship. The actual scholarship is the complete software environment and code that generated the results.” </p> <div style="margin-top: 12px; font-size: 0.75em; color: var(--tud-blue-2, #97C6FF); text-align: right; font-style: normal; letter-spacing: 0.3px;"> — <strong>Jonathan B. Buckheit & David L. Donoho</strong> (1995) </div> </div> </div> Note: ~20 Min - Bridge from components to reality - Claerbout and Donoho: Two software engineers - Published papers merely advertise underlying scholarship. - Complete software environment embodies actual research. - Transition: Real-world automated viewshed case study.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Nüchterner Titel --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 15px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Case Study: 1,500 Wind Turbine Viewsheds </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Automated batch processing in JupyterLab with legal requirements </div> </div> <div style="position: relative; width: 1080px; height: 475px; margin: 0 auto;"> <!-- Das interaktive Jupyter-Notebook Iframe --> <iframe id="viewshed-frame" src="assets/01_viewshedanalysis.html" style="width: 100%; height: 100%; border: 1px solid #d0d5dd; border-radius: 6px; box-shadow: 0 4px 18px rgba(0,0,0,0.08); background: white;"> </iframe> <!-- ================= IÖR-GRÜNE HUD BOXEN (1 PRO FRAGMENT) ================= --> <!-- Fragment 1: Prepare environment --> <div class="fragment hud-ioer" data-fragment-index="1" data-target="Prepare-environment"> <h4 class="hud-ioer-title">01 • Definition of Dependencies</h4> <p class="hud-ioer-body">Code as documentation & execution. Carto-Lab container pinned in the header.</p> </div> <!-- Fragment 2: Documenting the Environment (RAM, Cores) --> <div class="fragment hud-ioer" data-fragment-index="2" data-target="Hardware-specs"> <h4 class="hud-ioer-title">02 • Documenting the Environment</h4> <p class="hud-ioer-body">Programmatic hardware auditing: 64-bit architecture, 128 GB RAM VM, and 16 CPU cores.</p> </div> <!-- Fragment 3: Parameters --> <div class="fragment hud-ioer" data-fragment-index="3" data-target="Parameters"> <h4 class="hud-ioer-title">03 • Explicit Definition of Parameters</h4> <p class="hud-ioer-body">Observer height (1.8 m), turbine hub heights, and atmospheric refraction declared as immutable constants.</p> </div> <!-- Fragment 4: Create grass project --> <div class="fragment hud-ioer" data-fragment-index="4" data-target="Create-grass-project-(a-Location)"> <h4 class="hud-ioer-title">04 • Parametric Project Generation</h4> <p class="hud-ioer-body">Headless GRASS project and projection (EPSG:25833) initialized via code — no manual GUI clicks.</p> </div> <!-- Fragment 5: Data visualization --> <div class="fragment hud-ioer" data-fragment-index="5" data-target="Data-visualization"> <h4 class="hud-ioer-title">05 • Input Validation</h4> <p class="hud-ioer-body">Automated reprojection checks and boundary verification on municipal shapefiles prior to batch execution.</p> </div> <!-- Fragment 6: Viewshed Preview --> <div class="fragment hud-ioer" data-fragment-index="6" data-target="Viewshed-preview"> <h4 class="hud-ioer-title">06 • Visual Verification & Preview</h4> <p class="hud-ioer-body">Included preview map and visual verification directly inside the notebook pipeline.</p> </div> <!-- Fragment 7: Cumulative Preview / Final Archive --> <div class="fragment hud-ioer" data-fragment-index="7" data-target="cumulative-preview"> <h4 class="hud-ioer-title">07 • Automated Archive Generation</h4> <p class="hud-ioer-body">Output packaged into timestamped ZIP archives. Static HTML report compiled and published via CI/CD.</p> </div> </div> <!-- CSS: Keine zuckenden Transitions, stabile Deckkraft --> <style> .hud-ioer { display: none; position: absolute; width: 44%; right: -60px; top: 220px; box-shadow: 0 1px 4px rgba(0,0,0,0.5), 0 5px 25px rgba(0,0,0,0.2); background-color: rgba(112, 164, 1, 0.88); color: #ffffff; padding: 16px 22px; font-size: 18px; text-align: left; border-radius: 4px; z-index: 50; } /* Nur die jeweils aktive Box einblenden */ .hud-ioer.current-fragment { display: block !important; } /* Unsichtbares Dummy-Fragment für den leisen RAM-Scroll */ .hud-silent { display: none !important; } .hud-ioer-title { color: #ffffff !important; font-size: 1.05em; font-weight: 700; margin: 0 0 6px 0; } .hud-ioer-body { font-size: 0.82em; line-height: 1.45; color: #ffffff; margin: 0; } </style> <!-- Steuerung für Box 1 während Fragment 2 (kein Wackeln!) --> <script> (function() { function updateHUDState() { const b1 = document.querySelector('.hud-step-1'); const silent = document.querySelector('.hud-silent'); if (!b1 || !silent) return; // Wenn das leise Fragment 2 aktiv ist, Box 1 stabil sichtbar halten: if (silent.classList.contains('current-fragment')) { b1.style.display = 'block'; } else if (!b1.classList.contains('current-fragment')) { b1.style.display = ''; } } if (window.parent && window.parent.Reveal) { window.parent.Reveal.on('fragmentshown', updateHUDState); window.parent.Reveal.on('fragmenthidden', updateHUDState); } })(); </script> Note: - 1,500 wind turbine viewsheds for state legal area enforcement. - Legal court scrutiny required exact reproducibility. - Automated batch processing using headless GRASS. - Parameters, seeds, and hardware programmatic audited. - Transition: Zero-friction execution via the cloud.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 20px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Immediate Reproducibility </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Interactive living compendia in the cloud • Jupyter4NFDI & Binder </div> </div> <!-- Natives RevealJS r-stack mit identischen CSS-Parametern --> <div class="r-stack" style="margin: 20px auto 0 auto;"> <!-- 1. Basis-Bild (ohne Kasten) --> <img src="images/haclathon_binder_1.webp" onerror="this.src='images/notebook.png'" style="max-height: 480px; width: auto; object-fit: contain; border-radius: 4px; border: 1px solid #d0d5dd; box-shadow: 0 4px 18px rgba(0,0,0,0.08);" alt="IOER HaCLAthon Living Book"> <!-- 2. Überblend-Bild (exakt gleiche Maße, gleicher Rand, gleicher Schatten!) --> <img class="fragment" data-fragment-index="1" src="images/haclathon_binder.webp" onerror="this.src='images/notebook.png'" style="max-height: 480px; width: auto; object-fit: contain; border-radius: 4px; border: 1px solid #d0d5dd; box-shadow: 0 4px 18px rgba(0,0,0,0.08);" alt="Callout on Launch in Binder and Jupyter4NFDI"> </div> Note: - Contrast: Heavy pipeline, no one would ever reproduce it -> IOER HaCLAthon - Binder and Jupyter4NFDI enable cloud execution. - Single-click deployment removes local installation barriers. - Reviewers verify code without compiler configuration. - Transition: The underlying system dependency wall.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 20px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> The Spatial Dependency Hell </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Why spatial reproducibility fails at the system level </div> </div> <div style="max-width: 1040px; width: 92%; margin: 15px auto 0 auto;"> <!-- 3-Teilung: Daten • Kopplung • Software --> <div style="display: flex; justify-content: space-between; align-items: center; text-align: center; margin-bottom: 25px;"> <!-- Links: Persistent State --> <div style="flex: 1; background: var(--tud-bg-light, #F8F9FA); border-top: 4px solid var(--tud-blue, #00008C); padding: 18px 14px; border-radius: 4px;"> <i class="fa fa-database" style="font-size: 2.2em; color: var(--tud-blue, #00008C); margin-bottom: 8px;"></i> <h4 style="margin: 0; font-size: 0.95em; color: var(--tud-blue, #00008C);">Persistent State</h4> <p style="margin: 6px 0 0 0; font-size: 0.75em; color: var(--tud-text, #323F4B); line-height: 1.35;"> Raw inputs (<code>00_IN</code>), GeoPackages, SQL schemas, plain Markdown. </p> </div> <!-- Mitte: Das CoRDI 2025 Kopplungs-Mandat --> <div class="fragment" data-fragment-index="1" style="width: 130px; text-align: center; padding: 0 10px;"> <i class="fa fa-link" style="font-size: 1.6em; color: var(--tud-blue, #00008C); margin-bottom: 6px;"></i> <div style="font-size: 0.62em; color: var(--tud-text-muted, #566371); margin-top: 2px;"> Must be preserved together </div> </div> <!-- Rechts: Ephemeral Software --> <div style="flex: 1; background: var(--tud-bg-light, #F8F9FA); border-top: 4px solid var(--tud-orange, #C85000); padding: 18px 14px; border-radius: 4px;"> <i class="fa fa-cogs" style="font-size: 2.2em; color: var(--tud-orange, #C85000); margin-bottom: 8px;"></i> <h4 style="margin: 0; font-size: 0.95em; color: var(--tud-orange, #C85000);">Ephemeral Software</h4> <p style="margin: 6px 0 0 0; font-size: 0.75em; color: var(--tud-text, #323F4B); line-height: 1.35;"> Python packages, compiled C-libraries, GDAL, PROJ, dynamic OS runtimes. </p> </div> </div> </div> <!-- Punchline --> <div class="fragment" data-fragment-index="1" style="text-align: center; margin-top: 16px; font-size: 0.8em; color: var(--tud-blue, #00008C); font-weight: 600;"> “Standard data science isolates Python. Geoinformatics must freeze the operating system.” </div> Note: - CoRDI 2025 finding: Code requires environment. - Persistent state separated from ephemeral runtimes. - Pure Python isolation misses compiled C-libraries. - GDAL and PROJ require system-level freezing. - Geoinformatics must ship the operating system. - Transition: Demystifying software container technology.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Sachlicher, präziser Titel --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 20px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Software Containers </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Docker, Podman, Apptainer: Process isolation via Linux namespaces </div> </div> <!-- Das Whiteboard-Meme als alleiniger Held (kein Text darunter) --> <div style="text-align: center; margin: 10px auto 0 auto;"> <img class="restricted" src="data:," onerror="this.src='images/notebook.png'" style="max-height: 420px; width: auto; object-fit: contain; border-radius: 4px; border: 1px solid #d0d5dd; box-shadow: 0 4px 18px rgba(0,0,0,0.08);" alt="Overcomplicated Docker Whiteboard Explanation"> </div> Note: - fun fact: one of the first diagrams you find about docker - Much simpler: Containers use native Linux kernel (e.g.) namespaces/cgroups. - Not full virtual machines; lightweight isolation. - Standardized shipping container for scientific code. - Ensures portable cross-platform execution across clouds. - Transition: Bridging the Research Data Lifecycle. --- <!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel: Schlank & direkt neben dem Logo --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 12px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> The Research Data Lifecycle </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Bridging the computational gap from archive to reuse </div> </div> <div style="display: flex; gap: 40px; align-items: center; max-width: 1080px; width: 94%; margin: 5px auto 0 auto;"> <!-- Links: RDM-Kreis (auf 440px vergrößert, nutzt den Raum voll aus) --> <div style="flex: 1.05; text-align: center;"> <img src="images/csm_FDZyklus_en_2022_background_74a9925199.webp" onerror="this.src='images/first-record.png'" style="max-height: 440px; width: auto; object-fit: contain;" alt="Research Data Lifecycle Circle"> </div> <!-- Rechts: Minimaler, fokussierter Inhalt --> <div style="flex: 0.95; text-align: left;"> <!-- Die Kernachse (einzeilig) --> <h3 style="color: var(--tud-blue, #00008C); font-size: 1.2em; margin: 0 0 16px 0; font-weight: 800; white-space: nowrap;"> Archive → Share → Reuse </h3> <!-- Die ERC-Lösung --> <div style="background: var(--tud-bg-light, #F8F9FA); border-left: 4px solid var(--tud-blue, #00008C); padding: 16px 20px; border-radius: 3px; box-shadow: 0 2px 6px rgba(0,0,140,0.04);"> <strong style="color: var(--tud-blue, #00008C); font-size: 0.98em; display: block; margin-bottom: 6px;"> Executable Research Compendia (ERCs) </strong> <div style="font-size: 0.84em; color: var(--tud-text, #323F4B); line-height: 1.45;"> Packaging <em>Data + Code + Runtime Environment + Bindings</em> into a single verifiable archive. </div> </div> <!-- Zitation: Jetzt mit reichlich Sicherheitsabstand garantiert sichtbar --> <div style="margin-top: 16px; font-size: 0.68em; color: var(--tud-text-muted, #566371); line-height: 1.4;"> Concept: Kray et al. (2019) <em>COSIT</em> • DFG Code of Conduct (Guideline 7) </div> </div> </div> Note: - Complete cycle: Plan, collect, analyze, archive. - Executable Research Compendia bridge computational gaps. - Kray et al. (COSIT 2019) specification (Computational Research Compendium (CRC/ERC) - Bundling data, code, runtime, and bindings. - Transition: Carto-Lab Docker as research infrastructure. --- <!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel: Frei und ungestört neben dem Logo --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 25px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Carto-Lab Docker </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> A curated harness for geospatial research & teaching (IOER FDZ) </div> </div> <!-- Bild: Absolut verankert links im Hintergrund (beeinflusst kein einziges Text-Element) --> <img src="images/cartolab.png" style="position: absolute; left: -350px; top: 110px; height: 600px; width: auto; max-width: none; z-index: 1; pointer-events: none;" alt="Curated Geospatial Packages in Carto-Lab Docker"> <!-- Text: Rechts verankert mit z-index: 2 (volle Höhe, garantiert kein Überlapp) --> <div style="position: relative; z-index: 2; display: flex; justify-content: flex-end; max-width: 1080px; width: 94%; margin: 0 auto;"> <div style="width: 480px; text-align: left; font-size: 0.82em; color: var(--tud-text, #323F4B); line-height: 1.45;"> <!-- 1. Pre-Compiled C-Stack --> <div style="margin-bottom: 20px;"> <strong style="color: var(--tud-blue, #00008C); font-size: 0.98em; display: block; margin-bottom: 4px; white-space: nowrap;"> 1. Pre-Compiled C-Stack </strong> GDAL, GEOS, PROJ, GRASS, Mapnik tested for cross-compatibility. </div> <!-- 2. Versioned Stability --> <div style="margin-bottom: 20px;"> <strong style="color: var(--tud-blue, #00008C); font-size: 0.98em; display: block; margin-bottom: 4px; white-space: nowrap;"> 2. Versioned Stability </strong> Base OS with pinned Python runtime. Snapshot version <code>v1.1.0</code> executes identically across systems. </div> <!-- Link / Doku --> <div style="border-top: 1px solid #e2e8f0; padding-top: 12px; font-size: 0.76em; color: var(--tud-text-muted, #566371);"> <a href="https://cartolab.fdz.ioer.info/" style="color: var(--tud-blue, #00008C); text-decoration: none; font-family: monospace;"> cartolab.fdz [↗] </a> </div> </div> </div> Note: - anyone trying to compile Grass, Mapnik, GDAL etc.: Big challenge - Carto-Lab Docker maintained by IOER FDZ, a binder - Curated, pre-compiled spatial packages - Fixed version tags eliminate software drift. - Freely available open infrastructure for researchers. - Transition: Closing the provenance loop via CI/CD.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel: CI/CD direkt im Untertitel lesbar aufgelöst --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 20px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Closing the Loop: Automated Provenance </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Git, and Continuous Integration & Continuous Deployment (CI/CD) as the universal glue </div> </div> <!-- Screenshot der echten Pipeline im Zentrum --> <div style="text-align: center; max-width: 960px; margin: 0 auto;"> <img src="images/ci&cd.webp" onerror="this.src='images/gitlab_ci.webp'" style="max-height: 380px; width: 100%; object-fit: contain; border-radius: 4px; border: 1px solid #d0d5dd; box-shadow: 0 4px 18px rgba(0,0,0,0.08);" alt="GitLab CI/CD Pipeline of this lecture"> </div> <!-- Die Enthüllung blendet erst auf Klick ein! --> <div class="fragment" data-fragment-index="1" style="text-align: center; margin-top: 16px; font-size: 0.85em; color: var(--tud-blue, #00008C); font-weight: 700;"> This lecture was compiled, rendered as PDF, and archived with a DOI via this exact pipeline. </div> Note: ~Min 31 - How do we integrate all these components? - Git and CI&CD as the glue - Continuous Integration compiles this slide deck. - Every time I change a slide, a container starts in the cloud and compiles the presentation - PDF generated, versioned, and archived automatically. - Demonstrates Good Scientific Practice in action. - (Pause) - Transition: Entering Part IV: Generative AI.

<!-- .slide: data-background="#001450" --> <!-- .slide: data-header="clean" style="top: 0 !important;" --> <div style="text-align: center; color: white; padding-top: 35px;"> <!-- Kicker: Bloom-Stufe 4 --> <div style="font-size: 0.78em; font-weight: 700; color: var(--tud-blue-2, #97C6FF); letter-spacing: 2px; text-transform: uppercase;"> PART IV • EVALUATE </div> <!-- Haupttitel --> <h1 style="color: white; font-size: 2.2em; margin: 10px 0 12px 0; font-weight: 800; letter-spacing: -0.5px; line-height: 1.1;"> The Generative Frontier </h1> <!-- Untertitel (kompakt) --> <h3 style="color: rgba(255, 255, 255, 0.75); font-size: 1.0em; font-weight: 300; max-width: 860px; margin: 0 auto; line-height: 1.35;"> AI Non-Determinism, Provenance, and Accountable Scholarship </h3> <!-- Zitat-Box: Breiter (960px), fokussiert auf die Kernbotschaft --> <div style="margin-top: 32px; padding: 18px 26px; background: rgba(255, 255, 255, 0.06); border-left: 4px solid var(--tud-blue-2, #97C6FF); max-width: 960px; margin-left: auto; margin-right: auto; text-align: left; border-radius: 0 4px 4px 0; box-shadow: 0 6px 20px rgba(0, 0, 0, 0.25);"> <p style="margin: 0; font-size: 0.88em; color: #ffffff; line-height: 1.5; font-style: italic;"> “We prefer to be active participants during all the steps in the process of discovery [...], because we are motivated by <em>why</em> the result is true, more than by the final answer.” </p> <div style="margin-top: 10px; font-size: 0.72em; color: var(--tud-blue-2, #97C6FF); text-align: right; font-style: normal; letter-spacing: 0.3px;"> — <strong>Álvaro Lozano-Robledo</strong> (2026), <em>What should we tell <br>our students?</em> (Terence Tao Blog) </div> </div> </div> Note: ~Min 31, QUESTION! - Lozano-Robledo via Terence Tao, one of the greatest living mathematicians. - Science values the discovery path over output. - QUESTION: Who here has used a Large Language Model? - QUESTION2: Who can reliably reproduce the exact system prompt and input and output from three months ago? - LLMs generate answers while obscuring thought provenance. - Transition: How generative models compute output.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel: Schlank --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 8px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> The Statistical Mean </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 2px; font-size: 0.78em;"> Generative AI predicts the average, but science lives at the edges </div> </div> <!-- r-stack: 1-zeilig, wechselt per Fragment nahtlos von p1 zu p2 --> <div class="r-stack" style="margin: 5px auto 0 auto; max-width: 960px;"> <!-- 1. Basis-Bild (p1) --> <img src="images/figure_sentiment_qualifier_2023-10-19_p1.png" onerror="this.src='images/figure_sentiment_qualifier_2023-10-19.png'" style="max-height: 200px; width: auto; object-fit: contain;" alt="Generated map icons with default sentiment"> <!-- 2. Fragment-Bild (p2) --> <img class="fragment" data-fragment-index="1" src="images/figure_sentiment_qualifier_2023-10-19_p2.png" onerror="this.src='images/figure_sentiment_qualifier_2023-10-19.png'" style="max-height: 200px; width: auto; object-fit: contain;" alt="Generated map icons with happy sentiment modifier"> </div> <!-- 2 Spalten: Einzeilige Überschriften, gestraffte Typografie --> <div style="display: flex; gap: 35px; justify-content: center; max-width: 960px; margin: 12px auto 0 auto; text-align: left;"> <!-- Links: Stochastic Parrots (Einzeilig, kein Punkt) --> <div style="flex: 1; border-left: 3px solid var(--tud-text-muted, #566371); padding-left: 12px;"> <strong style="color: var(--tud-text, #323F4B); font-size: 0.88em; display: block; margin-bottom: 3px; white-space: nowrap;"> Stochastic Parrots </strong> <div style="font-size: 0.75em; color: var(--tud-text-muted, #566371); line-height: 1.35;"> Models predict the expected average. Unguided, they smooth outliers and original thought. </div> </div> <!-- Rechts: Illusion of Explanatory Depth (Einzeilig) --> <div style="flex: 1; border-left: 3px solid var(--tud-blue, #00008C); padding-left: 12px;"> <strong style="color: var(--tud-blue, #00008C); font-size: 0.88em; display: block; margin-bottom: 3px; white-space: nowrap;"> Illusion of Explanatory Depth </strong> <div style="font-size: 0.75em; color: var(--tud-text, #323F4B); line-height: 1.35;"> Fluent, polished surface prose deceives researchers into mistaking language synthesis for genuine understanding. </div> </div> </div> <!-- Zitation am Fuß --> <div style="text-align: center; margin-top: 10px; font-size: 0.62em; color: var(--tud-text-muted, #566371);"> Dunkel et al. (2024) <a href="https://doi.org/10.1007/s42489-024-00159-9">[↗]</a> • Rozenblit & Keil (2002) <a href="https://thedecisionlab.com/biases/the-illusion-of-explanatory-depth">[↗]</a> </div> Note: - Generative models predict expected statistical averages. - Example: 2024 paper (p1), latent space shifting (p2), yet remains random - Science advances through anomalies and outliers. - LLMs produce information entropy - We fall for Illusion of Explanatory Depth: Eloquence deceives. - Transition: What are the levers for more deterministic outputs?

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel: Tippfehler korrigiert ("Reducing") & Grammatik sauber ("their limits") --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 25px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Reducing Non-Determinism </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Technical controls and their fundamental limits </div> </div> <div style="max-width: 960px; width: 92%; margin: 20px auto 0 auto; text-align: left;"> <!-- 3 prägnante Einzeiler (Titel + Parameter inline) --> <ul style="font-size: 0.85em; color: var(--tud-text, #323F4B); line-height: 2.0; margin: 0 0 25px 0; padding-left: 25px;"> <li> <strong style="color: var(--tud-blue, #00008C);">1. Parameter Tuning:</strong> e.g. Temperature → 0.0 forces greedy token selection. </li> <li> <strong style="color: var(--tud-blue, #00008C);">2. Pinned Random Seeds:</strong> Fixed generation salts for run-to-run consistency. </li> <li> <strong style="color: var(--tud-blue, #00008C);">3. Self-Hosted Open-Weights:</strong> Local HPC (or ScaDS.AI) prevents silent vendor model drift. </li> </ul> <!-- The Invariant / Das wissenschaftliche Fazit --> <div style="background: var(--tud-bg-light, #F8F9FA); border-left: 4px solid var(--tud-blue, #00008C); padding: 14px 20px; border-radius: 3px; font-size: 0.8em; color: var(--tud-text, #323F4B); line-height: 1.45;"> <strong>👉 Explainable AI (XAI)</strong> remains in its infancy. A single altered token diverges the entire inference path. Treat LLMs as probabilistic black boxes. </div> </div> Note: - Can't we just configure the randomness away? e.g. Temperature 0.0. - Fixed random generation seeds stabilize output. - Self-hosted ScaDS.AI models prevent API drift. - But: a single comma, new output. AI remains fundamentally black box. - parameters cannot scientific rigor, but then what metric can? - Transition: How can we measure if a human is still in control?

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel: Schlank --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 15px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Token Ratio as a Metric of Thought </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Input-to-output ratio as an indicator of original synthesis </div> </div> <div style="display: flex; gap: 35px; justify-content: center; max-width: 1040px; width: 94%; margin: 10px auto 0 auto; text-align: left;"> <!-- 1. Low Ratio (Einzeilige Überschrift) --> <div style="flex: 1; background: var(--tud-bg-light, #F8F9FA); border-top: 4px solid var(--tud-text-muted, #566371); padding: 12px 18px; border-radius: 4px; box-shadow: 0 2px 6px rgba(0,0,0,0.03);"> <div style="font-size: 0.92em; font-weight: 700; color: var(--tud-text, #323F4B); margin-bottom: 6px; white-space: nowrap;"> 10 Tokens In → 1,000 Out </div> <p style="font-size: 0.78em; color: var(--tud-text-muted, #566371); line-height: 1.4; margin: 0;"> Unstructured prompting delegates the narrative to the model. Without strict constraints, you lose agency over the research scope. </p> </div> <!-- 2. High Ratio (Einzeilige Überschrift) --> <div style="flex: 1; background: var(--tud-bg-light, #F8F9FA); border-top: 4px solid var(--tud-blue, #00008C); padding: 12px 18px; border-radius: 4px; box-shadow: 0 2px 6px rgba(0,0,140,0.04);"> <div style="font-size: 0.92em; font-weight: 700; color: var(--tud-blue, #00008C); margin-bottom: 6px; white-space: nowrap;"> 5,000 Tokens In → 150 Out </div> <p style="font-size: 0.78em; color: var(--tud-text, #323F4B); line-height: 1.4; margin: 0;"> Extensive context priming (schemas, bounding boxes, text fragments) forces the model into a constrained compiler of human ideas. </p> </div> </div> <!-- Deine pragmatische "Rule of Thumb"-Box (jetzt garantiert voll sichtbar) --> <div style="max-width: 1040px; width: 94%; margin: 14px auto 0 auto; padding: 10px 16px; background: #ffffff; border-left: 4px solid var(--tud-blue, #00008C); box-shadow: 0 1px 4px rgba(0,0,0,0.04); text-align: left; font-size: 0.78em; color: var(--tud-text, #323F4B); line-height: 1.4;"> <strong>👉 Rule of Thumb:</strong> Provide more structured information as input than you expect as output. Human intellect selects and constrains, the model synthesizes. </div> <!-- Klare, renommierte Referenz mit Link --> <div style="text-align: center; margin-top: 10px; font-size: 0.64em; color: var(--tud-text-muted, #566371);"> Andrej Karpathy & Tobi Lütke on “Context Engineering” <a href="https://simonwillison.net/2025/Jun/27/context-engineering/" target="_blank" style="color: var(--tud-blue, #00008C); text-decoration: none;">[↗]</a> </div> Note: - Low ratio: Few inputs produce the average answer ('cognitive offloading'). - Example: Master thesis turning from cartography slowly into a psychology work - High ratio: Dense context yields constrained synthesis. - Context priming preserves researcher conceptual agency. - Karpathy and Lütke on context engineering. - Transition: Use AI to solve a problem. But: document that process openly..

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 15px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> Prompts as Code & Thought Provenance </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Archiving the conversational lineage of a bug fix </div> </div> <div style="position: relative; width: 1080px; height: 475px; margin: 0 auto;"> <!-- Das interaktive Funkwhale-Transkript Iframe --> <iframe id="funkwhale-frame" src="assets/2026-09-11_funkwhale_scrollbar_conversation_archive.html" style="width: 100%; height: 100%; border: 1px solid #d0d5dd; border-radius: 6px; box-shadow: 0 4px 18px rgba(0,0,0,0.08); background: white;"> </iframe> <!-- ================= IÖR-GRÜNE HUD BOXEN ================= --> <!-- Box 0: Ausgangszustand vor dem ersten Klick --> <div id="fw-box-0" class="hud-ioer-fw" style="display: block;"> <h4 class="hud-ioer-title">Scientific Transparency in Practice</h4> <p class="hud-ioer-body"> Fixing an upstream UI bug in open-source software (Funkwhale). As a non-frontend engineer, LLM assistance was utilized and disclosed openly to upstream maintainers. </p> </div> <!-- Box 1 (Fragment 1): Scrollt zu model-thought --> <div id="fw-box-1" class="fragment hud-ioer-fw" data-fragment-index="1" data-target="model-thought"> <h4 class="hud-ioer-title">Documenting Thoughts & Contributions</h4> <p class="hud-ioer-body"> Separating human intent from model reasoning. Transparent disclosure of model-generated suggestions versus human architectural choices. </p> </div> <!-- Box 2 (Fragment 2): Scrollt zu prompt-git-branch --> <div id="fw-box-2" class="fragment hud-ioer-fw" data-fragment-index="2" data-target="prompt-git-branch"> <h4 class="hud-ioer-title">User Control & Continuous Audit</h4> <p class="hud-ioer-body"> Active steering, prompt refinement, and code rejection. Maintainers can review, audit, and build upon verified human decisions. </p> </div> </div> <!-- CSS: Flach, unten rechts; nur die AKTIVE Box ist sichtbar --> <style> .hud-ioer-fw { display: none; position: absolute; width: 48%; right: -25px; bottom: 25px; box-shadow: 0 1px 4px rgba(0,0,0,0.4), 0 5px 20px rgba(0,0,0,0.18); background-color: rgba(112, 164, 1, 0.92); color: #ffffff; padding: 14px 20px; font-size: 17px; text-align: left; border-radius: 4px; z-index: 50; } .hud-ioer-title { color: #ffffff !important; font-size: 1.02em; font-weight: 700; margin: 0 0 5px 0; } .hud-ioer-body { font-size: 0.8em; line-height: 1.4; color: #ffffff; margin: 0; } </style> Note: - What could good scientific practice look like? - Open-source bug fix example. - Full LLM conversational transcripts archived publicly. - Transparent disclosure of human versus model reasoning. - Prompts versioned as code in Git, serve as audit trail - thought experiment: write conversations to be published! - Transition: Five essential Master thesis habits.

<!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- Titel: Ohne "methodologically", schlank & bündig --> <div style="text-align: left; margin-left: 235px; margin-top: 15px; margin-bottom: 20px;"> <h2 style="color: var(--tud-blue, #00008C); margin: 0; font-size: 1.25em;"> The Master Student’s Checklist </h2> <div style="color: var(--tud-text-muted, #566371); margin-top: 3px; font-size: 0.78em;"> Five foundational habits for resilient, reproducible research </div> </div> <!-- Breiterer Container (1080px): Ermöglicht schlanke, kompakte Zeilen --> <div style="max-width: 1080px; width: 94%; margin: 15px auto 0 auto; text-align: left;"> <ul style="list-style: none; padding: 0; margin: 0; font-size: 0.80em; color: var(--tud-text, #323F4B); line-height: 1.45;"> <li style="margin-bottom: 12px; display: flex; align-items: baseline; gap: 12px;"> <span style="color: var(--tud-blue, #00008C); font-weight: bold; font-size: 1.05em;">✓</span> <div> <strong style="color: var(--tud-blue, #00008C);">1. Immutable Inputs (<code>00_IN/</code>):</strong> Keep downloads write-protected; never alter raw spatial data by hand. </div> </li> <li style="margin-bottom: 12px; display: flex; align-items: baseline; gap: 12px;"> <span style="color: var(--tud-blue, #00008C); font-weight: bold; font-size: 1.05em;">✓</span> <div> <strong style="color: var(--tud-blue, #00008C);">2. Plain-Text First:</strong> Store logic, notes, and schemas in Markdown and SQLite; Word is just a disposable export. </div> </li> <li style="margin-bottom: 12px; display: flex; align-items: baseline; gap: 12px;"> <span style="color: var(--tud-blue, #00008C); font-weight: bold; font-size: 1.05em;">✓</span> <div> <strong style="color: var(--tud-blue, #00008C);">3. Headless & Parameterized:</strong> Expose important parameters in script headers; enable automated re-runs. </div> </li> <li style="margin-bottom: 12px; display: flex; align-items: baseline; gap: 12px;"> <span style="color: var(--tud-blue, #00008C); font-weight: bold; font-size: 1.05em;">✓</span> <div> <strong style="color: var(--tud-blue, #00008C);">4. Computational Freezing:</strong> Extensively use versioning and rely on containerized environments to eliminate software drift. </div> </li> <li style="margin-bottom: 12px; display: flex; align-items: baseline; gap: 12px;"> <span style="color: var(--tud-blue, #00008C); font-weight: bold; font-size: 1.05em;">✓</span> <div> <strong style="color: var(--tud-blue, #00008C);">5. Thought Provenance:</strong> Version prompts alongside code; archive AI synthesis as static (e.g. HTML) transcripts. </div> </li> </ul> </div> Note: - Pacing Checkpoint: Minute 40:00. - Concrete survival checklist for student theses. - Immutable raw data in `00_IN` directories. - Plain-text primacy; parameterized headless execution scripts. - Containerized environments and prompt version control. - Mention practical workshops - Transition: These rules serve home human purpose..

<!-- .slide: data-header="clean" --> ### Calm Technology <div style="width: 88%; margin: 8vh auto 0 auto; padding: 28px 35px; background: rgba(0, 0, 140, 0.03); border-left: 5px solid var(--tud-blue); border-radius: 4px; box-shadow: 0 4px 20px rgba(0,0,0,0.04); text-align: left;"> <p class="smallbold" style="color: var(--tud-blue); margin-bottom: 12px; text-transform: uppercase; letter-spacing: 0.05em; font-size: 13px !important;"> Design Philosophy • Human-Computer Interaction </p> <blockquote style="width: 100% !important; max-width: 100% !important; margin: 0; font-size: 1.1em; line-height: 1.45; color: var(--tud-text); font-style: normal; border: none; padding: 0; box-shadow: none;"> <em>„Designing technology that supports,<br>not distracts our lives.“</em> </blockquote> <div style="margin-top: 20px; padding-top: 10px; border-top: 1px solid rgba(0, 0, 140, 0.1); font-size: 0.75em; color: var(--tud-text-muted); text-align: right;"> Mark Weiser (1996) • Amber Case (2016) </div> </div> Note: ~Min 41:30 - good scientific practice is calm technology - it removes uncertainty, frees up agency and attention - and liberates you from locked knowledge - Quote replicated from research lecture - Restores focus to actually solve problems. - Transition: Concluding the lecture and references.

<!-- .slide: data-background="#232323" --> <!-- .slide: data-header="clean" style="top: 0 !important;" --> <!-- .slide: id="grand-finale" --> <div style="display: flex; justify-content: center; align-items: center; height: 80vh; gap: 8%;"> <div style="width: 45%; text-align: right;"> <img src="images/qr.png" style="width: 100%; max-width: 440px; padding: 10px;" alt="Scan for living slides & materials" /> </div> <div style="width: 50%; text-align: left;"> <h2 style="color: white; margin-bottom: 20px;">Thank You!</h2> <p style="color: #f0f0f0; font-size: 1.0em; margin-bottom: 4px;">Dr.-Ing. Alexander Dunkel</p> <p class="small highlight" style="color: #4fc3f7; margin-bottom: 24px;">alexander.dunkel@tu-dresden.de <br> a.dunkel@ioer.de</p> <p style="color: #ccc; font-size: 0.75em; line-height: 1.6; margin-bottom: 24px;"> Reproducible, Archived Lecture Slides & Lecture Notes: </p> <p style="font-size: 0.6em; color: #888888; line-height: 1.6; margin: 0;"> <a href="https://doi.org/10.5281/zenodo.23335010" style="color: #aaaaaa; text-decoration: none;">DOI: 10.5281/zenodo.23335010</a><br> <a href="cite.bib" download="cite.bib" style="color: #aaaaaa; text-decoration: none;">cite.bib</a> </p> </div> </div> Note: - Concluding lecture (~minute 42:30). - Thanking students, Dean, and habilitation committee. - lecture itself is the course material - Download via Zenodo or access in web. - Opening floor for questions and discussion.