What the Convergence is
A goal is not a molecule. “Brain fog” is not a receptor, and no single plant is a treatment for it. Underneath every goal on this site sit biological pathways — 567 of them, each named, each with its own page. A formula is built to cover pathways, not to out-dose a single ingredient.
That is the whole idea, and it is the reason a formula is many ingredients rather than one. Several ingredients that act on the same pathway converge on it. The question the Convergence answers is not “which herb is strongest” but “which combination covers the most of the biology behind what you asked for, on evidence that was graded before the result was known.”
How the combination is calculated
The calculation is deterministic. The same goals always return the same combination, because it is solved against a fixed graph rather than generated.
88 goals and conditions map onto the 567 named pathways beneath them. The mapping is published; every pathway has its own page.
A pathway opens only under the rule fixed in advance at How Evidence Is Graded. Grade A opens one alone. B+ opens one only with an independent corroborating driver. B and C never open a pathway by themselves.
946 graded ingredients sit across those pathways. Where several act on the same pathway, they converge on it — that overlap is the unit the builder optimises, not the dose of any one ingredient.
68 formulas are searched for the set that covers the most opened pathways for the goals given. It is a coverage problem with one answer, not a recommendation.
Why a small dose of many, rather than a large dose of one
Each formula is a 200:1 concentrate taken at an eighth of a teaspoon. The honest objection to that is obvious and worth stating plainly: every individual ingredient in such a formula sits below the dose used in the trial that earned it its grade. Anyone evaluating this should know that, and it is why the grade describes the ingredient’s published science rather than the finished formula’s effect.
Convergence is not synergy. Synergy is a specific pharmacological claim — that a combination produces an effect greater than the sum of its parts, demonstrated by combination index, isobologram or factorial design. This site does not make that claim about any formula. No finished formula here has been tested that way; the Limits Register records that 0 of 68 have been trialed as a complete formula.
Convergence is the checkable version. It means multiple independently graded ingredients acting on the same named biological pathway, where each ingredient carries its own published grade and each pathway its own resolvable citations. It describes how a formula was assembled — which pathways it targets and on what evidence — not what the assembly achieves in a person. What the literature does support is narrower, and worth stating exactly:
In gene-expression work on botanical hybrid preparations, the three-herb combination ADAPT-232 (Rhodiola, Schisandra, Eleutherococcus) deregulated 206 genes that none of its individual ingredients deregulated. A two-herb combination of Rhodiola with Withania deregulated 25 genes through interaction between the two. The work is in vitro, in cultured neuroglial cells; the review is narrative rather than systematic, and its authors note that in silico findings require in vivo validation. Panossian, Lemerond & Efferth, Pharmaceuticals 2024 — 10.3390/ph17040483
5′-methoxyhydnocarpin had no antimicrobial activity alone at any concentration above 500 µg/ml. At 10 µg/ml it cut berberine’s minimum inhibitory concentration against Staphylococcus aureus from 256 µg/ml to 16 µg/ml — a sixteen-fold potentiation — by inhibiting the NorA efflux pump that would otherwise expel berberine from the cell. Both compounds occur in the same Berberis plant. This is the clearest published case of an ingredient being worth carrying at a dose where it does nothing by itself, and it has a measured mechanism rather than an inferred one. Stermitz, Lorenz, Tawara, Zenewicz & Lewis, PNAS 2000, 97(4):1433–1437 — 10.1073/pnas.030540597. Antibacterial MIC work in vitro; nothing here establishes the same effect after oral dosing in a person.
The most comprehensive review of synergy in Chinese herbal medicine states that “evidence to support these synergistic effects remains weak and controversial”, that “the level of evidence remains low”, and that “the clinical relevancy of most of these findings is undetermined”. We publish that here rather than omit it. Zhou, Seto, Chang, Kiat, Razmovski-Naumovski, Chan & Bensoussan, Frontiers in Pharmacology 2016 — 10.3389/fphar.2016.00201
For one class of ingredient the small amount is not a compromise on the way to a larger one — it is the range in which the mechanism operates at all. That argument is hormesis, and it is set out with its citations, its seven documented hormetins, and the point where it stops, in The Hormetic Dose.
What this does not establish
Four things, stated directly, because an evaluation that omits them is not worth reading:
0 of 68 formulas have been trialed as a complete formula. The grades describe published science on ingredients and pathways. They are not a claim that any formula treats, cures, or is clinically proven for any condition.
The combination evidence above is in vitro. The 206-gene result is gene expression in neuroglial cells. The benefit it implies is predicted from that data, not demonstrated in people.
Individual doses are below trial doses. Coverage of a pathway is the design intent; it is not the same thing as reproducing a trial’s dose of a trial’s ingredient.
Combinations can also antagonise. Antagonism also occurs in these studies; some combinations underperform their parts. Convergence is a design principle under test, not a guarantee.
Two real dietary phytochemicals antagonised each other at low doses. In a formal combination-index analysis in human colon carcinoma cells, sulforaphane and 3,3′-diindolylmethane were antagonistic at low concentrations — a combination index as high as 4.2 at a 2.5 µM total — and became synergistic only at the cytotoxic 40 µM. The direction of the interaction is the inverse of the convergence idea at exactly the concentrations a diet or a formula would produce. We publish it because it is the strongest published argument against this page. Pappa, Strathmann, Löwinger, Bartsch & Gerhäuser, Carcinogenesis 2007, 28(7):1471–1477 — 10.1093/carcin/bgm044
Who this is for
This is not built for someone who wants a bottle and a promise. It rewards the opposite. Four kinds of reader get the most out of it:
Every grade names the study behind it, and every study resolves to a DOI, PMID or PMCID you can open. If you are the kind of person who follows the citation before believing the sentence, this site was built for you.
Concentrated traditional botanicals, described through the molecular pathways they act on, with the pathway named and its evidence graded. Both the modern map and the traditional one are published, and held apart rather than blended together.
These are dietary supplements, not drugs. Nothing here is evaluated by the FDA to diagnose, treat, cure or prevent any disease, individual results vary, and 0 of 68 formulas have been trialed as a complete formula. If that framing is acceptable to you, the rest of the site is useful. If it is not, it will not be.
Especially anyone with a diagnosed condition or taking prescribed medication. Several ingredients in this catalog interact with anticoagulants, sedatives and immunosuppressants, and those interactions are named on the pages that carry them.
Take the studies to your doctor
We would rather you did. Most supplement companies hope your physician never looks; this catalog is built to survive the appointment. Hand them these four and they can audit the whole claim in an evening:
The grading rule — published in advance, applied mechanically, so they can see what a grade had to earn before it was given.
The evidence against us — studies that contradict grades carried on this site, quoted in full. Start them here; it is the fastest way to judge whether the rest is honest.
The full corpus — every study cited anywhere on this site, in one plain-text file, with every identifier resolvable to its primary source.
The verification manifest — the SHA-256 of every machine-readable file published here, so they can confirm the file they read is the file we published.
Compare us to anyone
We would encourage it, and we will not tell you what you are going to find. We will give you the test instead, so you can run it on this site and on any other supplement company you like. Five questions. They take about ten minutes each:
Not a reference list at the bottom of the page — the specific study behind the specific sentence, with an identifier that resolves. Here, every graded claim carries one.
A rule written after the result is not a rule. Ours is published at How Evidence Is Graded and applied mechanically.
This is the question almost nothing survives. Ours is at The Evidence Against Us, quoted in full, including a trial that argues against a grade we still carry.
Anyone can list what they did. Ask for the other list. Ours is the Limits Register, and it states that 0 of 68 formulas have been trialed as a complete formula and 1 of 68 carries a published outside-laboratory certificate.
A site that means it will hand you the whole graph, not a brochure. Ours is graph.jsonld — every ingredient, pathway, condition and formula as one queryable file, with a published SHA-256 so you can confirm you received what we published.
Run those five on this site and on any other. We are not going to tell you what the answer will be — the point of publishing the test is that you do not have to take our word for the result either.
What we claim about accuracy, precisely. Not that this site is flawless — no corpus of this size is, and claiming it would be the least credible sentence on the page. What we claim is mechanical, and narrower than it first sounds: 47 automated checks run on every build and refuse to produce one when a figure they cover disagrees with the corpus — the grade census, the ingredient, pathway and formula counts, the citation totals, the cross-file figures, the sitemap and the published digests among them. They do not cover every sentence of prose on this site. Six figures had drifted when this page was last audited, and they were corrected rather than defended. So if you find a number here that disagrees with the corpus, it is an error and not a position — recompute it yourself from the full corpus, which is the fastest way to catch one. Where a study contradicts one of our own grades, we publish the study. Where we find an error, we correct it and the correction ships with the next build. That is a process you can inspect, rather than a promise you have to take.
Studies that contradict claims graded by this rubric are published in full at The Evidence Against Us. The full grading rule is at How Evidence Is Graded, and every pathway named here has its own page under Pathways.