Flow Monitor publishes one kind of material: explanations of how live Solana activity data is produced, transported, decoded and displayed. Not which token is moving, not whether to buy it, and not who is behind a wallet. The question this desk answers is narrower and more durable than any of those: when you look at a number on a screen, what actually happened to put it there.
The work is written by The Flow Monitor Desk. That is an editorial identity rather than a pseudonym for an individual, and the site deliberately publishes no author photographs, no invented credentials and no biography for a person who does not exist. If a claim on this site is wrong, the correct response is to check it against the primary documentation, which every page points at, rather than to weigh it against somebody's stated seniority.
What the desk covers
Three sections divide the subject. Sources covers transports: HTTP polling of an RPC node, WebSocket subscriptions, provider webhooks and validator-level Geyser streams, with the delivery guarantee each one carries. Latency covers confirmation semantics and pipeline delay, including where a lag is protocol behaviour and where it was added by software. Building a view covers the work of turning transactions into a number, from decoding a swap to designing an alert that stays quiet.
Within those sections the bias is towards mechanism. A page here would rather explain what a subscription guarantees than assert which provider is fastest, because the first is stable and the second is a fact about one afternoon on one connection. Where a mechanism has a documented constant, the constant is stated with its source. Where it does not, the page says the figure depends on your setup.
What the desk refuses to publish
- Measured latency figures. No milliseconds for any provider or endpoint, because such a number is unreproducible and would be read as a benchmark.
- Provider rankings. Vendors are named where naming them is the only accurate way to describe a mechanism, and are described from their own public documentation.
- Fabricated evidence of any kind. No invented statistics, no review counts, no testimonials, no aggregate ratings, no case studies with numbers nobody can check.
- Trading advice. Nothing here tells anyone what to buy, when to enter or how to size a position. The subject is data, not markets.
- Backdated publication. Dates on this site are the real build and revision dates, and are never adjusted to make material look older or fresher.
How a page gets written
A page starts from a question somebody actually asked, usually a variation of why two screens disagree. The mechanism is then read from primary sources: protocol documentation for anything the chain does, and the operator's own reference for anything a service does. Where those two disagree, the page says so rather than choosing.
Numbers are admitted under three conditions only. They may be protocol constants, such as the target slot duration or the lockout depth required for rooting. They may be values published by the party that runs the service being described. Or they may be arithmetic, shown in full and labelled as illustrative, so a reader can substitute their own inputs and get their own answer. Any number that is none of those three does not appear.
Every article carries an in-body word count, a real revision date and a stated section. Structural claims about a layout or a mechanism are checked against the documentation they came from before publication, and the full rule set is set out in the editorial policy.
Commercial relationships
This site links to a third-party volume console in its header and at the end of its articles. Those links are commercial placements and they open in a new tab. The desk does not operate that product, does not audit its output, and does not present figures from it as evidence for anything on this site. No provider named in an article has paid for or reviewed its inclusion, and no page ranks tools.
If a commercial link ever appears inside a paragraph, it is because the paragraph was already about the thing the linked page answers. Where a topic does not genuinely fit, there is no link, and several articles carry no in-body commercial reference at all for exactly that reason.
Corrections
Mechanisms change. Methods are added and deprecated, defaults move, program identifiers are superseded, and a page that was accurate a year ago can quietly become wrong. If you find something on this site that no longer matches the documentation, the fastest route to a fix is the contact page, with the page name and the source that contradicts it.
Corrections are made to the page itself and the revision date updates. The desk does not silently rewrite a claim that materially changed the meaning of an article; where that happens, the change is described on the page so a returning reader can see what moved.