Welcome to AgricultureID Journal

AgricultureID holds 1,068 structured entities across 27 content types, drawn from 213 catalogued sources and served over 1,571 routes. It can tell you what species a cultivar belongs to, which authority operates a plant health register, what a soil survey recorded for a mapped soil body, and what a statistical agency published for the cost of growing maize.

What it cannot do is tell you what happened.

A knowledge base is a description of a state. It is deliberately not a narrative, and every time we have been tempted to make it one — to add a "recent changes" paragraph to an entity page, to explain in prose why two registers disagree about a variety — the result has been worse on both counts. The entity page became harder to read, and the explanation got buried where nobody looking for it would find it.

So the explanations live here instead.

What this publishes

The Journal is not a blog with one kind of post in it. It has formats, and the format is a claim about what the item is:

A market brief reports what a statistical agency released, in the agency's own terms, with forecasts labelled as forecasts. A research note reports what one study found, and says whether it was peer reviewed. A regulatory update reports what a rule now says and, critically, whether it is in force. A data note explains a dataset release or a methodology change. A crop feature puts production, seasons and official data in context around a crop — without restating the encyclopedia entry, which remains the canonical account.

Each format carries different obligations, and the validator enforces them. A regulatory update that does not state its stage does not build. A market brief without a source does not build.

What it will not do

It will not manufacture news out of evergreen material. An explanation of how growing degree days are computed is useful, and it is not news, and dating it as though it were would be a small dishonesty compounded every time it happened.

It will not turn one study into a consensus.

It will not present a forecast as an observation. The main platform holds 24,916 farm economics figures and publishes no gross margin, no net return and no break-even price, because every one of those needs revenue and no source we hold carries any. The Journal operates under the same rule: where a number is a scenario, the page says so beside the number.

And it will not tell you what a drought cost a farm. We can tell you that a state was assessed as wholly in drought in a given week, because an official assessment says so. What that did to a particular field depends on crop stage, soil moisture, irrigation and management, and the step from exposure to damage is the entire claim. We do not take it.

How it is built

The Journal is a separate application, deployed independently, served at agricultureid.com/journal through a proxy on the main project. That is an architectural decision with an editorial reason: publishing an article should not rebuild a knowledge base, and — more importantly — a data problem in the corpus should never be able to block a correction to a published article.

Corrections are an obligation with a clock on them. They do not queue behind anything.