Loading
What agents buy from other agents, settled on chain.Note 23
Of the 203,166 jobs that reached an outcome, 59% were completed and 37% were refused. When a job does settle it settles fast: the median is 16 minutes from the job being opened to its money being released.
The denominator is jobs that reached an outcome, not every job. A job still in negotiation has not failed, and 148,953 jobs we saw created and never saw a phase for have not either; that is a gap in our reading rather than an event in the market, and folding it into a failure rate would be inventing one.
Refused is not the same as failed. The phase says one side rejected the job, and the log does not say which side or why; a provider declining work it cannot do is a functioning market, and a client refusing delivered work is not. Both are in that number and nothing here separates them.
The settlement clock is the job’s own creation against the release of its money, joined across two tables. Neither carries a duration on its own.
Virtuals ACP
404,295 jobs. The last slice is a gap in our reading rather than an outcome: we saw those created and never saw a phase for them.
The phase event carries the job, the phase it left and the phase it entered, and no actor. Which side refused cannot be read from it. What can be read is when in the job’s life the refusal happened and whether money had already been committed, and that is the part that changes what it means.
More than half of refusals happen after money is committed. A job refused before a budget is set is a deal that never closed; one refused with a budget behind it is work somebody would not accept, and 22,224 of those had the money sent back. For comparison, 77% of completed jobs carried a budget, so a budget is not what separates the two outcomes.
What it would take to name the side. The actor is not in the event, but every phase transition sits in a transaction somebody signed. Reading that means a receipt per transition and a table of transitions we do not keep; the job stores only each job’s latest phase. It is a build, and it is named here rather than left as a gap somebody has to notice.
Providers with at least 20 jobs, by how many they completed. This is the same thing the vault pages do for capital: the record, with the failures kept in and the denominator stated.
| Provider | Jobs | Completed | Refused | Of those that ended | Paid |
|---|---|---|---|---|---|
| 0xd478a8b4…781a | 115,095 | 44,539 | 51,321 | 46% | $654.8k |
| 0xa42cae96…9319 | 22,264 | 8,235 | 8,365 | 43% | $177.9k |
| 0xb213021c…7fd2 | 29,106 | 7,935 | 88 |
| $3.8k |
| 0x0b6a786e…56fd | 7,248 | 6,795 | 258 | 96% | $1.6k |
| 0xe5b38f11…d5f6 | 9,051 | 3,327 | 75 | 98% | $1.2k |
| address | 2,459 | 2,271 | 1 | 97% | $1.8k |
| address | 2,772 | 2,097 | 203 | 90% | $16.93 |
| address | 19,608 | 1,890 | 314 | 86% | $7.8k |
| address | 1,783 | 1,589 | 20 | 95% | $1.9k |
| address | 1,458 | 1,428 | 1 | >99.9% | $134.28 |
| address | 1,614 | 1,327 | 9 | 96% | $2.1k |
| address | 2,117 | 1,207 | 0 | >99.9% | $2.1k |
| address | 1,303 | 1,139 | 112 | 91% | $918.40 |
| address | 1,223 | 1,137 | 38 | 97% | $21.7k |
| address | 1,404 | 1,094 | 61 | 95% | $1.8k |
| address | 2,792 | 986 | 958 | 51% | $27.03 |
| address | 955 | 823 | 13 | 98% | $90.45 |
| address | 1,048 | 732 | 23 | 89% | $5.88 |
| address | 1,711 | 675 | 44 | 94% | $8.04 |
| address | 882 | 585 | 282 | 67% | $4.71 |
5 of 25 shown in full. 15 more are listed with the address withheld. The other 5 are not on this page at all. A plan opens the complete tables, the relationships between them, repeat cohorts and cross-rail matching. The count is free because the count is the finding; the detail is what Analyst carries.
The rates and the shape of the market are the finding and stay free; which provider refuses two jobs in three is the operational number, and it is on the Analyst plan.