Twelve Questions That Arrive While a Run Is Open
Short answers to the things operators ask sideways, mid-run, when the article they need is three scrolls too long. None of these repeat what the longer pieces cover; they are the questions those pieces leave in the margins.
Can a live run be watched from a phone?
Yes, provided the checks were designed for it. A phone can carry a signal sheet of six lines, a journal you append to, and one independent lookup of a transaction signature. What it cannot carry is diagnosis. If a reading crosses a threshold and the response is anything above a simple adjustment, get to a full screen before acting, or hand the run to somebody who already has one.
Does watching more closely make a run perform better?
No, and it usually makes the record worse. Extra checks do not change what the software does; they change how often you are tempted to change it. Each unscheduled intervention breaks the comparison between the readings before and after it, so the run ends without evidence you can calibrate against. Watch on a fixed cadence and spend the attention you save on the opening and closing phases.
What separates a paused run from a stopped one?
A pause preserves the option to resume: state is kept and the run is expected to continue. A stop ends the run and begins close-out. The practical difference is what each does to work already submitted, which varies between tools and should be established before you need it. A pause also decays, because the market keeps moving, so a long pause quietly becomes a stop you never declared.
Should one person or two watch a run?
One person watching with a second available covers most cases. Two people watching at once tends to produce either duplicated interventions or a diffusion of responsibility, where each assumes the other is deciding. What matters more than the headcount is that exactly one person holds the authority to stop at any given moment, and that everybody involved knows who that person currently is.
Is a quiet dashboard the same thing as a healthy run?
Not necessarily, and the ambiguity is dangerous because the comfortable reading is the wrong one. A panel stops changing either because nothing notable is happening or because the feed behind it has broken. Distinguish the two with something that updates even when nothing interesting occurs, such as a last-activity timestamp, and compare its age against a staleness threshold you set before opening.
What if the console and a block explorer disagree?
Treat the chain as the outcome of record and the console as a statement of intent. A disagreement is not automatically an error in either; it often means they are reading at different commitment levels or at different moments. Establish which by repeating both readings after settlement. If they still disagree, pause rather than adjust, because you no longer know which view your decisions are resting on.
Can the schedule be changed while a run is live?
It can, and the question is whether it should be. Changing pace is a legitimate adjustment when a confirmed signal calls for it and you write down what you expect the change to do. Changing it because the run feels slow is not. Either way, record the change with a timestamp, because every later comparison of readings needs to know that the conditions moved partway through.
How is a live run handed to another operator?
Three things transfer and all three have to be explicit. The plan, so the incoming operator knows what the run is for. The journal, so known quirks are not rediscovered as emergencies. And the stop authority, so it is clear whether that person may end the run or must escalate to somebody else. Write the handover as a journal line with both names and a timestamp.
What should be written down while a run is live?
Four fields per check: the time, what you observed, what you decided, and when you will look again. That is enough to reconstruct the hour afterwards and short enough that you will actually keep it. Add the expectation whenever you change something, because a change without a stated expectation cannot be evaluated later. Never edit an earlier line; append a correction underneath it instead.
Is restarting after a stop the same as continuing?
No. A restart is a new run and needs a plan written for the state you are actually in, including partially completed sequences, unswept balances and wallets that may now sit below their floors. Before restarting, confirm that the condition which caused the stop has been addressed rather than merely stopped being visible, or the restart becomes a second attempt at the same failure.
How do I tell a finished run from a stalled one?
A finished run has completed its plan and has nothing outstanding: no queued work, no unresolved transactions, and a balance snapshot that stays still when you read it twice. A stalled run has work remaining and nothing happening. The distinguishing check is whether anything is pending rather than whether the screen is quiet, and pending work is visible from the chain regardless of the tool.
Why does this desk not publish target numbers?
Because a threshold depends on your budget, your tolerance for interruption, the venue you chose and what else that capital could be doing, none of which are visible from here. Publishing a number would make it look measured when it would in fact be invented. Each page instead states what setting a line too tight costs and what setting it too loose costs, and leaves the choice with you.
Answers are written by The Run Desk and cover procedure rather than outcomes. If a question here has an answer that depends on a number, the number is yours to choose, and both directions of that choice are described in the signals worth watching. Missing questions are welcome by email; the address is on the contact page.