Skip to content
Inside the Boundary

All notes

Everything here, grouped by subject. Most people arrive with one question, so the paths on the front page may be a faster route in.

Start from where you are

Four short paths through the collection, chosen for the situation you are in rather than the topic. They are curated entry points, not the full list — a note can appear on more than one path, and everything is below.

Deciding whether to buy

Size the problem, then find out whether the physics allows it at your sites. Both come before any demonstration.

  1. The Question to Ask Before Anything
  2. Buddy Punching
  3. Position Accuracy at Building Scale
  4. Methods That Work Better Indoors
  5. Not Doing It At All
  6. Whether It Was Worth It
  7. Choosing a Product

Setting it up

Measure the sites, set the radii from the measurements, and design the fallback before anyone is asked to use it.

  1. Where to Start
  2. Sizing the Boundary
  3. When the Punch Is Refused
  4. Consulting Before You Deploy
  5. The Impact Assessment
  6. Writing the Notice
  7. Introducing It
  8. The First Week

Something has gone wrong

Refusals climbing, complaints arriving, or a dispute about pay. Start with the site rather than the person.

  1. How These Deployments Fail
  2. Shift Change and Queues
  3. No Signal, No Battery, No Phone
  4. Punch Time Is Not Arrival Time
  5. When the System Decides Pay
  6. The Thing You Are Spending
  7. Requests to Refuse

All 50 notes

The complete collection, grouped by subject.

The basics

4 notes

Five steps, and the strain is entirely in one of them. Plus the question that decides whether this technology addresses the problem you actually have.

The boundary and the position error are the same size at building scale, which is a fundamentally different situation from tracking a van across a city.

Methods

4 notes

Four alternatives work indoors, which satellite position does not, and three of them collect no location data at all.

Checking a position once, at a deliberate action, is one of the easier things to justify. Two design choices throw that advantage away.

The boundary

6 notes

Four features turn a defensible design into something else, and each arrives as an upgrade rather than as a decision.

Measure before you announce. A deployment that opens with a week of refusals never recovers the goodwill, and the refusals were predictable.

Running it

8 notes

The project ends at go-live; the system generates work every week. The refusal rate is the number, and most deployments do not track it.

Reference

7 notes

What the record cannot support, the arrangements that answer the same question with nothing, and the end state as a checklist.