Why testing needs someone else's chart
Finding a birth time from life events can't be tested on yourself: if you don't know your time, you have nothing to compare the result against. Only a chart where the time is written in a document will do — then it can be hidden from the algorithm to see whether it finds it.
Steve Jobs has such a record: February 24, 1955, 19:15, San Francisco. In public astrological data collections it has the highest reliability category — that is, it comes from a document, not from relatives' recollections. That is exactly why his chart is so often used as an example when people talk about rectification.
Let's say right away what is not in this article. We don't analyze Jobs's character from his date of birth and don't intend to: the site says outright that none of these systems is science, and pretending that one reads the personality of a stranger would contradict ourselves. Here he is a set of dates.
What we took and what we didn't
Five events, all from the public biography: the founding of Apple (April 1, 1976), the birth of his first child (May 17, 1978), leaving Apple (September 17, 1985), the wedding (March 18, 1991) and the death of his adoptive father (March 2, 1993).
We deliberately did not take his illness, although it is in the event catalog: someone else's medical history is not the material on which to demonstrate how a calculator works.
We also dropped a sixth event — the founding of NeXT in November 1985. And we didn't drop it ourselves: the engine warned that it stands within half a year of leaving Apple. In half a year the solar arc shifts by exactly its tolerance, and such a pair gives the same hit twice, inflating the match. We fell into this while preparing the article — exactly as anyone will who collects events in a row from one turbulent year.
What came out
The algorithm's verdict: no signal. In that case the page shows no time interval at all.
The numbers are these. The best point of the day scored 2.43. The random-background threshold is 3.35: that's what the same five events score if you randomly shuffle them in time. That is, the peak turned out 0.73 times as strong as what chance produces. The probability of getting such a result by chance is 0.63 over two hundred shuffles. For comparison: a result is considered noticeable when this value is below 0.05.
And a separate detail that would be nicer not to write. The actual minute of birth, 19:15, scored 0.08 — almost zero. The best point of the graph fell at 19:48. Thirty-three minutes of difference sounds like “almost hit it,” but against this background it's a coincidence, not a hit: a peak would have appeared there even without any events.
Why the internet confidently says “19:02”
Jobs's chart is a textbook example in the rectification literature, and there you often meet a refined time such as 19:02 instead of the documented 19:15. It looks convincing: thirteen minutes, a neat justification from events.
The difference between that analysis and ours is not in the arithmetic but in one step. Any set of events gives some peak — simply because the day has places where there are more hits because of how the chart is built, not because of your biography. Until the peak is compared with the random background, it means nothing. We make that comparison, and that's why we sometimes have to say “we don't know.”
This doesn't mean that someone else's result is wrong and ours is right. It means that on these five dates it can't be told apart from chance — and it's more honest to say so than to name a minute.
What this means if you are the one who doesn't know your time
The method works, but more modestly than it is said to. On charts with a documented time it cuts off most of the day on average — narrows it to a few hours. It can't find the minute.
The practical conclusion is simple: the more events and the more precise the dates, the higher the chance of seeing something. Five events, some of which are remembered as “somewhere in spring,” will almost always give “no signal.” And that is a normal answer, not a broken page.
If you got a refusal from us, you got an honest result. A site that names a minute in such a situation simply isn't doing the check against chance.
How to repeat this yourself
Everything described above is calculated on our own birth-time search page. Open it, enter the date 24.02.1955 and the city of San Francisco, add the same five events with their dates — and you'll see the same verdict and the same numbers.
By the way, while we were preparing this article it turned out that the site understood “San Francisco” as a small town in Argentina: the real city was in the directory under the machine transliteration “San Fransisko.” The calculation didn't crash, it silently used a different longitude and a different time zone. We fixed it — together with Philadelphia, Montreal, San Diego and San Jose, which had the same trouble.