Launchpad Research and Placement Rationale
Sources inspected 2026-10-27. This page explains the revised boundary between preparation and programming. Sources inform the design; none validates this exact curriculum, its hour estimates or a promise of becoming a particular percentile of engineer.
The placement decision
Launchpad should support a learner who can begin without programming experience. Reading claims carefully, asking useful questions, planning realistic study and recovering from interruptions are useful from the first day. Those three modules move here with their existing practice and assessment, adapted to files, schedules and short passages. This transfers 17–26 hours from Semester 00; it does not add those hours again.
Digital Systems, Safety and Recovery adds a concrete no-code mental model and an actual restore. AI for Learning and Everyday Work adds a no-code assistance and verification workflow. Their 14–21 hours are new. They contain three concept clusters and six lessons each, with cluster practices, cases, labs, mistakes, fresh assessments and calibration.
Programming-dependent work stays in Semester 00: learning calibration with actual code attempts, data automation, HTTP exchanges and executable evaluation of AI suggestions. This boundary is our prerequisite-based design inference, not a placement rule claimed by an external framework.
What the sources support
Learning transition and study support
Princeton's Making the Transition to College guidance emphasizes active engagement, planning and seeking help in a more independent learning environment. We use these as practical preparation goals, not a fixed daily routine or a trial-proven formula. The existing orientation research documents learning-method research and the Cornell/Waterloo source routes retained in the moved modules.
Computer models before commands
CS50 Understanding Technology explicitly targets people who use technology without understanding its underlying behavior. Its hardware and Internet scope supports including mental models before specialized development tools. However, the official page states that the course was retired on June 30, 2024. We label it as an optional archive for concepts, not a current certification or source of current security-product instructions. New local lessons and current official guidance provide the required route.
CS50's Introduction to Cybersecurity is an optional deeper course for account, device and data protection. Full completion is outside the Launchpad estimate. We do not turn an orientation module into a prerequisite security specialization.
Recoverability and safe everyday decisions
NCSC's everyday-security collection supplies account, authentication, update and password-management guidance. Its backup guidance provides the practical recovery context. Our disposable wrong-edit lab and deliberately incomplete-copy control are original teaching exercises. A local restore demonstrates its narrow contract, not protection against every failure.
Mozilla's connection-identity guidance supports interpreting connection state. Our destination and provenance exercises keep that separate from judging a site's honesty. Browser icon shapes and product menus can change; understanding and evidence are assessed instead of screenshot memorization.
AI literacy before AI programming
UNESCO's AI competency framework for students spans human-centred thinking, ethics, technical/application knowledge and system design, with increasing depth. We use this breadth to avoid reducing AI literacy to prompt tricks. The local Launchpad route concentrates on understanding, bounded use and judgment; later programming and production modules deepen implementation and evaluation.
AI Fluency: Framework and foundations, developed by Rick Dakan, Joseph Feller and Anthropic, provides a practical framework for deciding what to delegate, specifying assistance, judging responses and taking responsibility. It is vendor-supported educational guidance. Its usefulness as a design reference does not establish independent evidence that our module improves learning. Select relevant sections; the complete course is optional.
The University of Helsinki's computing-course overview identifies Elements of AI as a beginner starting point without computer-science experience. It is an optional conceptual course, not a hidden coding prerequisite. The course landing page itself returned an access error to the research tool, so registration, progress tracking and completion were not verified.
Assisted success is not independent competence
Bastani and colleagues' randomized classroom study studied AI-supported high-school mathematics practice and later unassisted performance. It found that the form of support mattered: stronger assisted performance did not necessarily imply stronger independent learning. This supports examining both outcomes rather than treating completed answers as learning evidence. It does not establish the same effect size for adult computing learners, every AI product or our authored offline exercises. Our hint, fresh-transfer and assistance-disclosure requirements are design responses whose effectiveness still needs learner evaluation.
What is deliberately not required
No new book purchase, paid AI service, full external course or upload of private study material is needed. Existing book routes remain selective. Local lessons and fixtures are original teaching material; resources are linked rather than copied into redistributed book text. The course requires a real disposable restore but permits offline AI fixtures, which demonstrate workflow judgment rather than live model performance.
The old GCF computer-basics route redirected to a general learning page during inspection, and CISA pages/PDF returned access errors to the research tool. They are not used as if their full current contents had been inspected. Accessible NCSC and Mozilla primary guidance supports the required security route instead.
Workload and evaluation plan
Launchpad becomes 92–138 focused hours, or 103–156 if all four conditional repairs are needed. Semester 00 becomes 210–309, excluding those repairs. Combined core preparation is 302–447, versus the previous 288–426; only the two new modules increase the combined estimate.
Use one evidence packet and record reading, practice, assessment and repair time separately. Inspect the transferred records at Semester 00 entry; do not repeat whole modules merely because they moved. A learner who already completed equivalent work can demonstrate a fresh scenario and reuse the record.
Before calling the design learner-validated, collect independent content review and beginner observations: confusing prerequisites, actual completion times, failed/fresh assessment results, needed accessibility adaptations and later unaided use. Record non-completion and repairs as well as successes. Automated content checks, arithmetic checks and a working website verify narrower properties; they cannot establish motivation, durable learning or educational effectiveness.
Content-review correction
The subsequent review found legacy assessments that required algorithms before programming, forced daily/identity-based study rules, conflicting fixed timelines and an overextended resource list. These were corrected to prerequisite-matched tasks, explicit task scoring, one core study route and a small inspected source selection. Study resources distinguishes the author's habit framework, university advice and evidence-rated research. These corrections improve curriculum coherence; they do not establish learner effectiveness. The newer staged AI assessment also separates independent planning from candidate exposure and uses unequal review durations to test transfer.