What am I looking at once an analysis is in?

#The workspace

The first thing a project shows is the system itself. Three views of the same accepted knowledge sit side by side, and selecting an object in one selects it in the others.

• The map. Components and the interfaces between them, with external services at the edge. A large system opens as subsystems, grouped by the containment the analysis recorded, never guessed from names or folders. Each group says how many objects and connections it holds, and opens on request.
• The interface registry. One entry per logical interface, with its operations inside it: what calls what, over which protocol, at which address where that is known. A missing route is shown as missing.
• The components. The catalogue of everything the analysis found, searchable, with each component purpose and lifecycle.

The inspector beside the views shows the selected object: what it is for, its operations, the evidence and the open questions about it. A three-dimensional view of the same map is one switch away, for walking through a request level by level. It draws the same recorded facts and invents no traffic.

#Evidence under every claim

A statement in a project is never a loose paragraph. It names the thing it is about, the property it describes, the value, and where that was established, down to a file and lines where the analysis recorded them.

Each claim also says how it came to be believed: observed, meaning read directly in the system, declared by a document or a person, or inferred from naming and shape. An inference is labelled as one, because a confident sentence and a checkable one look the same on a page until somebody asks how anybody knows.

#What is not known

What an analysis could not establish is not filled in. It becomes an open question with the reason it is open and how much it matters, listed on its own page. Some will be answered by the next analysis. Some can only be answered by a person, and the page says which.

The same honesty applies to scope. Every analysis records its coverage, what was examined completely, in part, or not at all, and the project never treats an unexamined part of the system as an empty one.

#How much is known

One number on the project page says how much of the system is known, and the arithmetic behind it is fixed and shown. A question fully established counts as one, established in part counts as one half, and not established counts as nothing. An answer that something does not apply here is an answer and counts as one. A contradiction is not an answer and counts as nothing.

The number is a measure of the picture, not a grade of the system. Code alone does not answer everything, and the remaining questions are on the page as questions for a person rather than promised away.

#What changed

After a second accepted analysis, the changes view compares the two. It keeps three things apart: a change in the system, such as an interface that moved or an integration that appeared, something newly learned that was already there, and a description that was only reworded.

Something is reported as gone only when the newer analysis looked at the place it used to be. Elsewhere it becomes a question. And a claim that an earlier analysis supported and the current one no longer does is marked as no longer confirmed, with the date it was last supported.

The map can show the same comparison, with additions and removals drawn on the recorded structure, and every fact behind a mark one click away.

#The briefing

Each accepted analysis has a briefing: a short numbered agenda for a conversation with whoever builds the system.

1. What changed in the system

   The structural changes of this analysis, with their before and after.

2. What was learned

   Facts found for the first time that were already true, kept apart from changes.

3. What is still open

   The current open questions, by the priority they were recorded with.

4. Where to look next

   One area for the next analysis, with the reason it was suggested. The owner chooses.

The briefing can be copied as meeting notes, with links back to the evidence. The links give nobody new access: a reader without an account in the project sees nothing behind them.

#Taking it away

What the project knows can be downloaded from the project page as one Markdown file that any editor opens. It says which analysis it reflects and what that analysis looked at, how much is known, what the system is made of, what talks to what, where the data comes to rest, what is risky and what nobody has established, and it ends with a section on what the file does not contain. It is one button, not a format to choose or a job to wait for, and the project records who took a copy and when.