How we evaluate local knowledge

What is the record?

It is 272 dated weekly fishing reports for the Port Aransas area, from working local guides. The window runs October 2016 to August 2026, and every report in it is a guide describing the week he just fished.

We evaluate the published text of those reports where they were originally posted, and we leave them there. We publish none of that text and take no content from it. What the site keeps is a table of monthly percentages.

The sample splits into two independent stretches: 2016–2022 (155 reports) and 2023–2026 (117 reports). A single calendar month holds between 6 and 18 reports in a given era.

How do we test a pattern?

We count presence, not mentions: for each report we record only whether a term appears at all, then take the share of that month's reports that contain it. Then we run the same count separately in the two eras. A pattern is publishable only when both eras show it in the same direction.

Presence rather than frequency is deliberate. A guide who says a word six times in one report is not evidence of six weeks of fishing, and counting words would let one talkative week carry a month. One report, one vote.

The two-era split is the honest part. Anything that shows up in only one stretch is a change in how the reports were made, or noise, and we cannot tell which from inside the record. Three concrete outcomes of that rule:

• September bait is not published. The two eras disagree sharply about live croaker in September, so no September croaker claim appears anywhere on this site.
• A jetty-season claim was retracted. An early read suggested the jetty was a winter story here. It failed to replicate across the eras and reads as a change in reporting style rather than a change in the fishing, so it is not published.
• A seasonal snapper pattern is held back. It has not been reconciled with the federal season dates, and a fish with a regulated season is exactly where a record like this must defer to the regulator rather than lead.

What do we deliberately not publish?

Spot names taken from reports, any technique or rigging detail, and anything with a legal answer. The record is used for one thing only: what a month here usually looks like. It is never used to tell you where to stand, how to rig, or what you are allowed to keep.

Seasons, size limits, and bag limits are not patterns and never come from this record. They come from Texas Parks and Wildlife and NOAA, cited at the claim, with the date we checked them. Those pages are the species and limits pages and licenses and regs.

We also publish no shopping list. Knowing that a bait is seasonal is useful to a visitor; being told to go buy it is not, because in the weeks that matter it is often not on the shelf. That is a reason to fish with someone who already has it, and we say so plainly rather than pretending otherwise.

How do citations read?

Three tiers, and you can tell them apart on sight. A regulated fact cites the regulator. A live reading cites the feed and the time it was measured. A pattern from the record states its sample size and its window inside the sentence, so you can weigh it without leaving the page.

• Regulated fact. A size or bag limit, with a Texas Parks and Wildlife link and the date we verified it.
• Live reading. A tide, wind, or water temperature, stamped with when it was measured and which NOAA feed it came from. The grading rule behind the verdict is documented separately on how we grade fishing conditions.
• Pattern from the record. A sentence such as: live croaker is mentioned in a given share of August reports across 272 weekly Port Aransas reports published between October 2016 to August 2026, split into two eras. The numbers travel with the claim.

What are the limits of this method?

Three, and they are worth stating plainly. The record is a small number of local guides, not the whole fishery. Some months rest on a handful of reports. And the reports are spoken, not written, so a word can be misheard in the text we evaluate.

• A narrow source. A small number of local guides reporting weekly. They fish this water for a living, which is why the record is usable at all, but it is their water and their habits, not a census of the fishery.
• Thin months. A single month in a single era can rest on as few as 6 reports. That is why nothing is published off one era alone, and why we state the per-month counts in the sentence.
• Spoken, not written. The reports are spoken, so a word can be misheard in the text we evaluate. Presence counting absorbs some of that, and the two-era rule absorbs more, but it puts a floor under how precise any single percentage can be.

Where a pattern is close to a threshold, we leave it out. The record is here to add context to live conditions, not to replace them.

Where does this sit next to the conditions method?

This page covers the record: human reports, evaluated for patterns. The conditions engine has its own method page, covering the live side: which NOAA feeds are read, the breakpoints that grade them, and how the daily verdict is composed. They answer different questions and neither depends on the other.

Read how we grade fishing conditions for the live side, or see both working together on the Port Aransas fishing report, where today's reading sits above what the month is typically like.