Every entry in the TechStack Archive follows the same documentation discipline so that findings stay comparable across experiments and trustworthy over time. An experiment begins with a single, clearly stated question — for example, whether an official Apple screen repair is worth the premium over a third-party shop, or how much BLE identity data AirDrop leaks in a crowded transit hub — and a defined scope that records exactly which device, software build, location, and time window were tested. Scope is published up front because a result only means what its scope allows it to mean; an n=1 test rig in one location describes what happened there, not a universal guarantee.
Data is captured with the tools appropriate to the question: vendor repair invoices and timed shipping logs for service experiments, controlled battery drain runs under a consistent workload for endurance tests, and dedicated capture hardware such as a Nordic nRF52840 development kit running custom Wireshark and Kismet firmware for Bluetooth Low Energy advertisement analysis. Wherever a metric can be measured directly, it is; wherever it cannot, the source is cited and the limitation is named. Each experiment records its raw inputs, the metrics that were logged, and the exact conditions under which they were logged, so a reader can judge whether the setup matches their own.
Findings are separated from conclusions. Findings report what was observed — the costs, the timelines, the frames per minute, the unsolicited connection prompts — and conclusions interpret what those observations mean for a real user. Both are written plainly, and known gaps are called out explicitly rather than buried. When a question cannot be answered firsthand without crossing an ethical line, such as performing an actual SIM swap or running a phishing simulation, it is marked as not independently testable and left to documented research instead of guessed at. When a queued experiment's hypotheses are later superseded by hands-on testing, the original entry is updated to point to the new results rather than silently rewritten.
Finally, every experiment carries a status — Done, In Progress, Pending, or Paused — and queued research is tracked openly with its background sources and planned test steps published before any hands-on work begins. This keeps the archive honest: visitors can see what has been tested, what is still being tested, and what has been set aside, and they can weigh each finding against the scope and limits that produced it rather than treating any single result as a universal rule.
