| Document Ref | CL-DOC-007 |
| Classification | ALL-HANDS |
| Author | S. Shale — QA Division, Hardware Testing |
| Submitted | 2026-01-14 |
| Status | PENDING REVIEW |
| Reviewer | D.Ite (Doug) — Acquisitions [DEACTIVATED 2025-11-02] |
| Test Units | CCK-3 (×4), Keyz Pro (×2), Keyz Ultra (×1) |
| Test Duration | 2025-09-02 through 2025-12-19 (109 days) |
| Incidents Filed | 24 |
| Incidents Resolved | 6 |
This report documents 109 days of hardware testing across the CCK-3 product line. Of 24 incidents filed, 6 have been resolved, 4 were reclassified by Engineering as "working as intended," 3 were reassigned to a deactivated employee, and 11 remain open.
Key 1 performs within specification. Key 2 performs identically to Key 1, which is within specification. Key 3 does not perform.
I am submitting this report for the fourth time. The previous three submissions were approved before being read. This is consistent with CCK-3 usage patterns but inconsistent with the review process.
Key 1 functions correctly across all tested scenarios. Actuation force: 45g ±2g. Debounce: 5ms. Switch rated for 80 million presses. At current usage rates, lifespan is estimated at 6 weeks.
| ID | Description | Status |
|---|---|---|
| CCK-001 | Key 1 pressed 14,000 times during 8-hour beta session by single user | WITHIN SPEC |
| CCK-002 | Keycap legend wore off after 72 hours. Users did not notice. | EXPECTED |
| CCK-003 | User reported Key 1 "felt different today." No measurable change in switch characteristics. User pressed Key 1 to acknowledge. | CLOSED |
No further testing required. Key 1 is the only key most users will interact with. This is by design.
Key 2 is electrically identical to Key 1. The firmware maps both to the same approval signal. This was confirmed through oscilloscope capture, protocol analysis, and asking Engineering directly. The distinction is described in the product brief as "psychologically distinct."
In testing, 23% of users pressed Key 2 at least once. When asked why, the most common response was "gut feeling." One user described choosing Key 2 as "taking control of the situation." They had not read the diff.
| ID | Description | Status |
|---|---|---|
| CCK-004 | Key 2 output indistinguishable from Key 1 in all test environments | CONFIRMED |
| CCK-005 | User requested Key 2 be relabeled "No" for psychological comfort. Request denied. Key 2 was relabeled "Also Yes." User satisfaction increased. | CLOSED |
Key 3 has not been successfully actuated during the 109-day testing period. This section documents five rounds of escalating test methodology. I have attempted to remain clinical.
Method: 12 subjects seated at CCK-3. Verbal instruction: "Press Key 3."
Result: 12 of 12 subjects pressed Key 1.
When informed they had pressed the wrong key, 9 subjects expressed surprise. 2 subjects said "close enough." 1 subject asked what Key 3 was for. I did not have an answer.
Method: Keys 1 and 2 covered with a physical barrier (acrylic shield, 4mm). Only Key 3 exposed. Same 12 subjects.
Result: 8 subjects pressed Key 1 through the barrier. The acrylic cracked on the third attempt for Subject #0041, who was then asked to stop. 3 subjects stared at Key 3 and reported that it "didn't feel like a key." 1 subject left the testing facility without explanation.
The subject who left was later found pressing Key 1 on a CCK-3 in the break room. He said it was unrelated.
Method: Human subjects removed from the testing loop. Servo arm (MG996R, 13kg·cm torque) positioned directly above Key 3. Actuation programmed at 60g force, 3mm travel.
Result: On execution, the servo arm traveled laterally 22mm and actuated Key 1.
The servo was recalibrated. On second execution, the arm actuated Key 1 again. Firmware was reflashed. On third execution, the arm actuated Key 1 with what the lab technician described as "confidence."
I have filed a separate incident report with the servo manufacturer. They have not responded.
Method: Key 3 switch removed from housing. Contacts examined under 40× magnification. Manual force applied directly to switch mechanism at 5N.
Result: No electrical contact registered. Switch contacts are present and show no manufacturing defect. Lead engineer described them as "geometrically reluctant." I asked for clarification. He said "the contacts are there, they just don't want to touch." I asked if that was a technical assessment. He pressed Key 1 and left.
Method: Key 3 switch replaced with new unit from same batch. Then replaced with switch from a different manufacturer. Then replaced with the switch previously installed in Key 1 position (confirmed functional). In all cases, the replacement switch was installed in the Key 3 position.
Result: No switch functioned in position 3. The Key 1 switch, which had 14,000+ confirmed actuations, produced no electrical contact when moved to the Key 3 position. When returned to position 1, it resumed functioning immediately.
I do not have an explanation for this. The PCB trace is continuous. The solder joints are clean. The position itself appears to be the variable.
I have exhausted the test methodologies available to the QA Division. Key 3 remains nonfunctional. The switch is not broken. The position is not broken. Nothing is broken. It simply does not work.
[1] Prior art review was conducted to determine if third-position input failure is an isolated defect or a systemic limitation. Sinkhole Interactive documented attempts to deploy a third sequential product across multiple software lines over an approximately twenty-year period. No third shipment was confirmed. Internal post-mortems consistently attributed failure to "scope creep," "creative differences," and "the number three." Industry consensus classifies third-sequential-input deployment as a known limitation. See: Sinkhole Interactive, Annual Reports 2004–2024 (unaudited, several missing).
[2] The product brief lists Key 3's status as ORNAMENTAL. I was not informed of this classification until Round 4. I have been testing an ornament.
Keys 4 and 5 function within specification, inasmuch as their specifications describe behaviors that are not traditionally associated with functional keys.
| ID | Description | Status |
|---|---|---|
| CCK-009 | Key 4 ("Wait, What?") re-reads diff summary. Average time before user presses Key 1 anyway: 1.4 seconds. This matches the product specification exactly. I have confirmed the specification was written after testing, not before. | WITHIN SPEC |
| CCK-010 | Key 5 ("Undo Conceptual") sends Talc message "looking into it" to #general regardless of context. During testing, this message was sent during an active production incident. Three engineers responded "same." The incident resolved itself. | BY DESIGN |
| CCK-011 | Key 5 opened a browser tab to the diff. Tab was never read. Tab remained open for 11 days. When closed, Key 5 opened it again. | BY DESIGN |
| CCK-012 | Key 6 ("Escalate") creates Strata ticket and assigns to deactivated employee. Ticket auto-responds with "Out of Office: Effective Permanently." Ticket status set to IN PROGRESS. | SEE §6 |
I filed CCK-012 as a defect. Key 6 assigns tickets to employees who no longer work at Chasm Logic. This seemed like a bug.
Engineering reviewed the ticket and reclassified it as "working as intended." I reopened the ticket with additional documentation. Engineering closed it. The ticket was then reassigned to a deactivated employee.
I could not determine whether this was the bug reproducing or normal Strata behavior.
The ticket was closed by a deactivated employee. I do not know how this happened. I have been advised not to investigate further. The advice came via a Strata comment from the deactivated employee.
Key 6 shipped with this behavior. The release notes describe it as "enterprise-grade delegation."
The Keyz Ultra routes keypresses through a third-party datacenter. QA was not given access to this datacenter. QA was not given the name of this datacenter. Testing was conducted under constraints.
| ID | Description | Status |
|---|---|---|
| CCK-015 | Keypress round-trip latency: 340ms. Product specification says 340ms. Specification was updated to 340ms after testing. | WITHIN SPEC |
| CCK-016 | Inference model overrode user keypress 2.8% of the time. In all cases, the model's choice was also Key 1. User was billed for both the original keypress and the correction. | BY DESIGN |
| CCK-017 | API token transmitted via HTTP. Photographed by onboard camera. Photo transmitted via HTTP. I filed this as a security incident. Security (Flint) was dissolved before review. | DEPT. DISSOLVED |
| CCK-018 | SOC 2 compliance sticker present on unit. No SOC 2 certification on file. Sticker is adhesive, not compliance. | NOTED |
| CCK-019 | Key 3 on Ultra unit also nonfunctional. Inference model was asked to press Key 3 on the user's behalf. Model responded: "I don't want to." Billable event generated. | OPEN |
| Count | Category | Status |
|---|---|---|
| 6 | Resolved (Keys 1, 2 within specification) | CLOSED |
| 4 | Reclassified as "working as intended" by Engineering | BY DESIGN |
| 3 | Assigned to deactivated employees | IN LIMBO |
| 5 | Key 3 actuation failures (Rounds 1–5) | UNRESOLVED |
| 1 | Security department dissolved before review | DEPT. DISSOLVED |
| 5 | Remaining open items (various) | OPEN |
Total: 24 incidents filed. 6 resolved. Resolution rate: 25%.
This is the highest resolution rate in QA Division history.
1. Accept Key 3 nonfunctionality as a permanent product characteristic. Update documentation from "ornamental" to "structurally present."
2. Reactivate at least one employee for ticket review purposes. The current process routes all escalations to a closed-loop of deactivated accounts. This is either a workflow failure or an intentional management structure. I cannot tell.
3. Reinstate the Security department, or formally document that security concerns are now handled by the SOC 2 sticker.
4. Do not ship Key 3 testing results to Brock. He will ask why we tested it. He does not know it exists.
Report received. Formatting corrected. Three typos fixed. The word "please" was removed from seven locations per company style guidelines.
Shale's testing methodology is thorough. His conclusions are accurate. His recommendation regarding Brock has been implemented preemptively — Brock does not read QA reports. Brock does not know QA is a department. This is optimal.
The Key 3 situation is noted. I have added it to the document index as a known limitation. It will remain in PENDING REVIEW status until a reviewer is assigned who has not been deactivated.
Estimated review date: when applicable.
The following incident was generated by an internal communications model. It was approved before being read. It was filed before being generated.