What we will not tell you, and why

There are three sentences this site will not write: that your marriage will succeed or fail, that an event in someone’s life was caused by a dosha, or that a remedy works. Those are not disclaimers tucked under a page that says otherwise at the top. They are the reason the rest of the site is built the way it is.

What follows is what is left once you take them seriously, which turns out to be quite a lot, and to be checkable line by line.

What is the difference between a conclusion and an outcome?

One is a fact about a rule set and the other is a claim about two people’s lives, and only the first can be verified by anybody.“The Kerala tradition scores this pairing below its acceptance line” and “this marriage will be unhappy” are not two ways of saying the same thing.

The first can be checked by anyone holding the same tables. It is either right or wrong about what that tradition says, and if it is wrong someone can show that it is wrong in an afternoon. The second cannot be checked by anyone, including the person making it, and it is the one that gets repeated at dinner tables for years.

Everything in a report from us is of the first kind. We name the tradition, print the rule it applied, cite the text the rule comes from, and state how common that result is across all pairings. Where five traditions disagree, we print five verdicts rather than picking one and presenting it as the answer.

This is a real limit and it costs us something to keep. Plenty of people arrive wanting to be told that everything will be fine. We cannot give anyone that, and the substitute on offer — here is the full range of what the tradition actually says about you, with sources — is less comforting in the first minute and more useful in the second.

Why is that limit honest rather than merely cautious?

Because we ran the traditions against each other and measured how often they agree, and the answer is that they disagree on most pairings. The refusal to report one school’s conclusion as the truth is a finding, not a temperament.

Across all 11,664 ordered nakshatra-pada pairings, scored under five documented schools, 56.4% trigger a tested difference or veto across the combined census checks. That was computed over the whole population rather than a sample: there are exactly a hundred and eight positions, so a hundred and eight squared is every pairing that exists, and a second run produces the same figure to the decimal.

The Mars axis is worse, and it has to be measured differently because it needs real planetary positions rather than star names alone. On a seeded corpus of 2,000 generated charts, 61.4% are flagged under one school’s house set and clean under another’s. The narrowest reading — Astro-Vision (serious Kuja dosha only) — flags 11.7% of those charts. The widest, OP / SKU panchangas, flags 73%. Same charts, same planet, two published house lists that omit different houses because they descend from two different verses.

If two equally documented readings of the same chart disagree with each other on most charts, then no single reading of that chart can honestly be reported to you as a fact about your life.

Nobody made an error to produce that spread. Each list is faithful to its own source. The disagreement is inside the tradition, and a service that hides it in order to hand you one clean verdict has chosen confidence over accuracy. The piece on why two astrologers give opposite answers works through where the five schools part company.

Which choice actually moves a manglik verdict the most?

Not the house list everyone argues about — the reference point Mars is judged from, which moves the rate roughly ten times further. This is the least discussed finding in the census and the one that changes how the argument should be conducted — the piece on reference points works through both axes and the attribution between them.

Every manglik rule has two moving parts. The first is the list of houses in which Mars is held to cause the affliction, which is where the published disagreement sits: some texts count six houses, some seven, some eight. The second is the point those houses are counted from — the ascendant alone, or the ascendant and the Moon, or the ascendant, the Moon and Venus.

Hold the reference point fixed and swap between the house lists, and the flagged rate moves by about 3 percentage points. Hold the house list fixed and swap the reference point, and it moves by about 29. The axis nobody names in a consultation is the axis doing nearly all of the work.

That is worth saying plainly because it changes the useful question. Asking an astrologer which houses they count is a question about the part that barely matters. Asking whether they checked Mars from the Moon and from Venus as well as from the ascendant is a question about the part that does, and it is the difference between a chart that comes back clean and one that does not far more often than the verse dispute is.

Do we ever tell you that a remedy works?

No, and we do not sell one, take a commission on one, or refer you to anybody who does. That is a business decision as much as an editorial one, and it is the easiest of the three refusals to keep because we removed the incentive to break it.

Classical sources prescribe a great deal, but our verified registry currently contains cancellation rules and matching vetoes rather than a sourced catalogue of observances. We therefore do not publish a remedy reference or present a veto as a remedy. The matching-veto reference names exactly what it contains and stops there.

What we will not add is the sentence at the end saying it will help. We have no way to establish that, and neither does anyone selling it. A report that finds a dosha and then offers to sell you the solution has a problem no amount of sincerity fixes.

What is a cancellation rule, and why is it not a remedy?

A cancellation is a condition written into the rule set itself, so whether it applies to your two charts is a question with an answer. A remedy is something a person does afterwards. The two get conflated constantly and they are not the same kind of claim at all.

A text that declares a dosha usually also declares the circumstances under which it does not apply. Same Moon sign but different birth star. The two Moon-sign lords being mutual friends. Mars in its own sign. These are not interventions; they are clauses, and they were written by the same authority that wrote the clause that flagged you.

So we compute them and print the result, including the ones that would have cancelled a finding and did not, with what each would have required. A reader who can see that four cancellation rules were checked and three came close is in a very different position from a reader handed one word.

The distinction also decides what a second opinion can honestly offer. Whether the Moon signs of two charts are identical is a matter of fact. Whether a fast on a Tuesday changes anything is not, and no amount of care taken over the first question earns anybody the right to answer the second. Keeping the two apart is most of the editorial discipline on this site, and probably most of the complaints it will attract.

The registry is also where the tradition is least tidy, and the report says so. One widely quoted verse cancelling nadi dosha when the Moon signs agree has not been located in a named printed edition of any of the texts it is attributed to, and we mark it as traditional and unverified wherever it fires. One school in our set accepts no cancellation on nadi at all. Reporting that honestly means sometimes printing “we could not source this” next to something a reader was hoping would rescue them, which is uncomfortable and correct.

What are we willing to assert?

Three kinds of claim, and every one of them is checkable by somebody who does not trust us. That is the test each has to pass before it goes on a page.

• Where the planets were. Positions are computed against the published JPL planetary and lunar ephemerides and validated against them, and anyone can check a single position independently in JPL Horizons. This is astronomy. It establishes exactly one thing: where the bodies were at a moment.
• What each tradition concludes. Rule tables transcribed from named sources, cell by cell, and cross-checked against a second independent statement of the same rule wherever one exists. The engine cites the text each rule came from, so a disputed cell can be argued about against the book rather than against us.
• How common each conclusion is. Computed exhaustively over the space of pairings rather than estimated. The average Ashtakoota total is 20.1 out of thirty-six, and 66.7% of pairings clear the conventional pass mark of eighteen — context a score on its own cannot give you. The base rate tables carry the rest.

How accurate does the astronomy have to be before accuracy stops being the argument?

Far less accurate than it already is — the worst disagreement we measure against the reference ephemeris is 0.44 arcseconds, and no matching rule can resolve anything close to that. Precision is not what two astrologers are disagreeing about, and saying so removes the most common excuse for treating one verdict as final.

Put the numbers next to each other. A nakshatra is thirteen degrees and twenty minutes wide, which is forty-eight thousand arcseconds. The Moon’s worst measured error in our engine is 0.37 arcseconds, checked at 2,000 sampled instants spread across the years 1900 to 2100, against a tolerance of 2. The Moon covers that error in well under a second of its own motion.

The recorded birth time is a different story. The Moon moves about thirteen degrees a day, so a minute of clock error moves it far more than the ephemeris error ever could, and an hour of error — a hospital note rounded, a remembered time, a time recorded before the national time standard reached a particular district, or a place whose historical offset in the time zone database is not what people assume — is most of a pada. If you want to know why two reports disagree, the clock and the ayanamsa are far more promising suspects than the planetary tables, and the explainer on the ayanamsa covers the second one.

None of that is an argument for astrology or against it. It is an argument about where the uncertainty actually lives, and it happens to sit almost entirely in the rule choices and the input data rather than in the arithmetic.

It also disposes of a claim made in both directions. One camp says the calculations are too crude to mean anything; the other says modern software has finally made them exact. Neither describes what is happening. The positional part became exact enough to stop mattering decades ago, and the part that decides your verdict — which houses, counted from where, under whose reading, with which cancellations allowed — is not the kind of thing precision can settle. More decimal places will never close a gap that is made of two different verses.

Why do we print the harshest verdict as well as the kindest one?

Because a service that quietly surfaces whichever reading the customer wants is not a second opinion, it is a laundering service with citations attached. The temptation is obvious and worth naming out loud.

Somebody arrives having been refused. Five schools are run. Two accept, three refuse. There is an easy version of this product that leads with the two, buries the three under a heading nobody scrolls to, and lets the reader walk away relieved. That product would sell better than this one. It would also be indistinguishable, in every way that matters, from the astrologer who gave a family the answer they wanted in the first place — and it would collapse the moment a relative asked a hard question about it.

So the strict readings are printed at the same size as the lenient ones, in the same order every time, whichever way they fall. If the school that refuses you is the one with the better-sourced verse, the report says so. If the school that accepts you rests on a cancellation we could not locate in any printed edition, the report says that too, next to the cancellation, where it is inconvenient.

There is a practical reason as well as a principled one. A family that senses it is being handled stops listening, and a report that only ever brings good news reads as advertising within about two minutes. The thing that survives an argument with a sceptical uncle is a document that visibly did not choose a side: five columns, five sources, and the reader deciding what to make of it.

What do we do with the frightening wording in the classical texts?

We preserve it in one clearly labelled appendix and let it appear nowhere else. Some texts state the consequence of a finding in language a modern reader would find genuinely alarming, and sanitising a source is its own kind of dishonesty.

So the wording stays, marked as the historical text and explicitly not as a prediction, and a check on every build stops it leaking into a report page or a summary line. Repeating a threat to somebody who is already distressed is not transparency, and putting it three clicks away behind a label is not censorship.

The vocabulary itself has a history that is worth knowing, because the words have drifted. The history of the word dosha traces how a term meaning a fault or defect in a configuration came to be read as a curse on a person, and that drift explains a lot of how these findings land.

Is astrology true, and why is that not the question here?

Because it will not be settled in time to help you, and because a narrower question can be. If a match has been refused on astrological grounds this month, the metaphysics is not the obstacle in front of you. Your family is.

The answerable question is: is this the only reading? Published Mars rules and matching conventions can produce different findings. The census combines several kinds of difference and veto; its headline figure is not the rate at which schools give opposite total-score verdicts. Check the rules applied to your charts to see whether a named alternative reading actually changes the finding.

There is a version of this argument we deliberately do not make, which is the one that says the whole system is arbitrary and therefore nobody should mind. Families do mind. Research on how Indian families weigh match criteria finds horoscope and caste filters treated closer to absolute constraints than to preferences to be traded off — Banerjee, Duflo, Ghatak and Lafortune measured that directly in the marriage-advertisement market. Telling somebody their parents are being irrational is not a second opinion; it is a way of losing the argument on their behalf.

A second opinion here is not a better verdict. It is a full account of how many verdicts exist, which apply to your two charts, and where each comes from. What a family does with that is theirs to decide, and we have no view on it. The page on what to do after a refusal is the practical version.

How can you hold us to this?

By reading any page here and looking for the four things we say are not on it. A rule nobody can audit is a slogan, so here is the audit.

Look for a sentence that predicts something about a person. Look for a figure that is not attributed to a computation you could reproduce. Look for a remedy we claim works. Look for a product we suggest you buy to fix something we found.

A check runs on every build to catch the first and the third, and a separate one fails the build if any number on a page was typed by hand instead of read from the engine. That is why a figure changing in the engine changes it on the page, and why no page can quietly keep quoting a rate that stopped being true two versions ago.

Both checks have the same limit, and it is worth stating rather than hiding: a check only catches the phrasings somebody thought to write down. A page could still say something careless in a construction no regular expression anticipates, and the person most likely to notice is the person it lands badly on. If you find one that got through, it is a defect rather than a difference of opinion, and we would like to hear about it — with the sentence quoted, so the check can be widened to catch that shape the next time somebody writes it.