hoba is a public, versioned atlas of the hiring process and the places it breaks: 81 entries covering the observations a candidate can make, the barriers an application passes through, the mechanisms that can produce each observation, and the changes that would remove them. Free to read, in two languages, and queryable as data.
The problem it exists for
Nothing in the process is built to tell you which of them it was. Nobody inside it sees the whole of it — the recruiter cannot see how the panel scored you, the panel cannot see whether the budget still exists — so what reaches you is short, standardised and late. People fill that gap with the one explanation always available: that the problem was them. Often the arithmetic was settled before anyone read anything.
Every observation is recorded with what it can establish and, separately, what it cannot. A rejection at two in the morning establishes that a decision was recorded. It establishes nothing about your work.
For each observation the registry lists the mechanisms compatible with it — sixteen of them, for the most common rejection template. It does not assert which one fired, because from outside the system that cannot be known, and pretending otherwise is how the guessing starts.
Each mechanism is attached to the position that controls it, and marked with whether a candidate can do anything about it at all. For most, the answer is no. Knowing which is which is the difference between effort and grinding.
One observation, several causes
It is also the least informative. Here is what the atlas does with it.
Subject: Update on your application
Thank you for your interest in our company and for the time you invested in the process.
After careful consideration we have decided to move forward with candidates whose profiles more closely align with the requirements of the role.
We wish you the best in your search and encourage you to apply for future openings.
What to notice: Nothing in the text is specific to this candidate, this conversation or this role. The message is a routing decision rendered as prose.
That a decision was recorded and sent out as a template, on a date you can point to. Nothing more.
Every mechanism below can produce this exact message. The registry does not tell you which one it was — only which ones are compatible, and which of those you can test.
This is why the registry sorts by agency zone. Some of these you can probe in one message; some depend on an intermediary; some are budget decisions taken above the hiring process.
Open this observation →Diagnostic methodology
Four steps that replace going over it in your head with a set of separate, checkable facts.
Record only what you observed directly: emails received, the exact stage reached, gaps between events. No guessing at intent.
Output: observation nodes (A)
Find the barrier in the hiring funnel where it stopped: ATS parsing, the technical screen, a budget freeze.
Output: funnel barrier DAG (B)
List the mechanisms logically compatible with those facts. The honest baseline mechanisms stay on the list, alongside structural friction.
Output: compatible mechanisms (M)
Separate what you can act on from what depends on an intermediary and what is exogenous. The step ends in a few bounded, low-cost probes.
Output: probes and a stop condition
Named patterns
When two defensible decisions combine into a bind nobody chose, it can only be discussed once it has a name.
The same candidate is rejected as overqualified for junior and mid-level roles and as underqualified, or not specialised enough, for staff and leadership roles.
Trigger rule: The candidate is rejected as overqualified at level N and as short on depth at level N+1 within the same market cycle.
A candidate completes multiple late-stage interviews, is rejected with a generic message, and sees the identical role reposted weeks later.
Trigger rule: The role is closed and the candidate rejected, and a materially similar listing is publicly reopened within 60 days.
A requisition requires more years of hands-on experience with a specific framework or tool than that technology has existed in the open-source ecosystem.
Trigger rule: Job description or automated screen requires X years of experience where X > technology_age.
The candidate has to name a compensation expectation first: a high number ends the process, and a low one lowers the level the candidate is assessed at.
Trigger rule: The candidate is asked for a compensation number before the band for the role is stated, and the band is not stated in return.
Boundaries
Every entry on this site is a validated file in a public repository, served as static JSON, GraphML and CSV, and available to agents over MCP.
It runs in your browser. Nothing you type is sent anywhere.