Flattery is the easiest feature in this category to build and the only one that would have destroyed the product. Here is the argument, including the part where we were wrong.
Early in this app's development somebody asked for a reasonable thing: could the readings be warmer? Could a difficult chart be told it looked good?
The instinct behind it is not stupid and it is not cynical. People arrive at astrology at bad moments. Telling someone at a bad moment that their chart is full of trouble is a genuinely unkind thing to do, and the request was really about that.
The answer was still no, and it is worth writing down why — because the version of this product that said yes would have been more popular, and would have been worthless.
The argument against warmth
A mirror that flatters is not a mirror. The single thing this app has that a horoscope column does not is that it can be wrong — and therefore can be right. Both of those disappear the moment the output is tuned toward what the reader wants to hear. What is left is a very expensive way of generating compliments, and there is a free one on every social network.
It contradicts the engine's own rules. The reasoning underneath this app is explicit that a chart describes terrain, not a verdict; that a period which demands more than it returns must be named as one; and that a difficulty disclosed is a difficulty you can plan around. An interface that softened all of that would be arguing with its own brain in public.
It is found out. Not in the first session. In the fourth, when the app tells you a hard period is a good one and you are living through it. The credibility you spend on flattery is not spent evenly — it is all spent at once, on the day it matters.
The part where the objection was right
The objection was right that a hard finding delivered alone is cruel, and "we do not flatter" is not a licence for that.
So what got built is not permission to be blunt. It is a constraint:
No hard finding is ever shown alone. Every one is paired with a real mitigating factor drawn from that same chart — not a platitude, not a generic reassurance. Something specifically, checkably true about you.
That is a function in the source called compensator(), and it is not optional. It searches your chart for a genuine strength: an exalted planet, a lord in a good house, the house that carries the most weight in your chart. If it cannot find one it does not invent one — it names the strongest thing there actually is and says so plainly.
Comfort comes from completeness, not from lying. A hard chart told fully — including what the difficulty buys, and what in it is unambiguously working — is more steadying than a soft chart told falsely. That is the whole design.
What this looks like on the screen
- Periods are named for what they do, not graded on a curve. A period that asks more than it returns is described as one, along with the date it ends.
- Strength is described as strong or weak, never good or bad. Strong and weak still leave the significations room to move; good and bad hand down a sentence.
- Nothing is ranked relative to other people. There is no percentile, no score out of a hundred against a population, no "rare chart" badge. Those exist to make you feel selected and they measure nothing.
- Where the research is incomplete the app says training needed rather than producing plausible text. Why that matters more than it sounds.
- And it will never tell you that something is wrong with you and then offer to fix it for money. Ever.
The uncomfortable admission
This is written down as an open disagreement in the project's own documents, not as a settled victory — because the person who asked was the person building the thing, and being overruled by your own tooling is an odd position to be put in.
It stays written down for the same reason the bug that put every planet six hours behind the sky stays written down. A product that only publishes the arguments it won is doing marketing. The useful ones are the arguments where somebody had a real point and the decision went the other way anyway.
You are welcome to think we got this wrong. If the app tells you something hard and you have lived the opposite, file a disagreement inside it — that goes into the research and has already changed readings. That is the deal here: we do not flatter you, and you get to argue back, and the argument is recorded.