LumenChoice — Decisions, without the guesswork

Inspiration

Every day, people choose between jobs, learning opportunities, community projects, and tools. Most comparison apps offer an unexplained winner, while spreadsheets hide assumptions. We wanted to make a decision traceable: which priorities mattered, what tradeoffs were made, and how robust is the answer?

What it does

LumenChoice is a privacy-first, browser-based decision studio. Add alternatives, assign measurable 0–10 ratings, set the importance of each criterion, and instantly view an explainable ranking. The recommendation changes as your priorities change. A reproducible 900-run Monte Carlo simulation explores how the ranking might change when weights and estimates vary.

Features

• Live weighted-score leaderboard with clear score breakdowns.
• Editable criteria, weights, choices and option ratings.
• Balanced, budget-first and career-first scenarios.
• Transparent tradeoff matrix and plain-language explanations.
• Adjustable scenario-stress test with seeded, reproducible results.
• Local autosave and JSON/CSV export and import.
• Responsive, keyboard-accessible design.

How we built it

A dependency-free static app using HTML, CSS, JavaScript ES modules, and the browser's native APIs. A separate pure computation module handles weighted scoring, normalization, deterministic simulation and strict JSON import validation. The UI uses DOM updates to reflect changes, and localStorage stores decisions on the current device. No cloud backend, signup, analytics or external AI inference is needed.

The math: A choice earns a score out of 100 equal to the sum of each 0–10 rating multiplied by its normalized importance, then multiplied by 10. Stress testing perturbs weights and scores around the user's inputs. Win shares represent robustness under these assumptions, not real-world probabilities.

Challenges and learnings

The challenge was combining meaningful mathematical behavior with an approachable interface. We separated the scoring engine from the UI so the model is independently testable, made scenario sampling deterministic, and avoided presenting simulated outcomes as predictive certainty. This reinforced the importance of transparent explanations and honest uncertainty communication.

Testing

Nine automated checks passed for score calculation, ordering, reproducibility, import validation and explanations. The interface was exercised in desktop and mobile-sized Chromium viewports. The example scenario is illustrative and not validated against external outcome data.

What's next

Multi-person consensus comparison, evidence links and confidence annotations for each rating, decision-history comparisons, multilingual interface support, and accessible data visualizations.

AI usage

AI assisted with implementation and drafting during this hackathon. The product's scoring and simulation are deterministic local JavaScript algorithms, not an AI model. Users can inspect the open-source logic and understand the recommendation.

Technologies

HTML5, CSS3, JavaScript, ES modules, Node.js built-in test runner, browser localStorage, Vercel and GitHub.

Source

https://github.com/samertall/tools/tree/main/lumenchoice-next-byte-v5

Live app

https://lumenchoice-next-byte-v5-samertalls-projects.vercel.app

Original implementation for Next Byte Hacks V5 in October 2026.

Built With

• css3
• github
• html5
• javascript
• localstorage
• monte-carlo
• vercel