Hacker News Reader

Comment by jack_pp | Hacker News Reader

Absence of evidence actually is (weak) evidence of absence. https://www.lesswrong.com/posts/mnS2WYLCGJP2kQkRn/absence-of...

View on HN

There is no absence of evidence, not by a long shot. There are fewer written records in the traditional sense (books, letters, legal documents concerning things like transfer of property) but there is no shortage of inscriptions on various sorts of things, for example -- not to mention every other kind of historical evidence and, of course, the fact that some aspects of early medieval Western chronology can be cross-checked against contemporary sources in North Africa and the Middle East. We call it "the Dark Age" by comparison with, say, Late Antiquity, but even so, we have vastly more evidence about what happened in this age in Western Europe than we have about what happened in various ages of Antiquity all over the place, and no one thins those were made up.

As for the article you linked to, that kind of probabilistic approach sort of works for recent events. Applying it to things that happened a long time ago, where the presence or absence of evidence is correlated with a whole lot of other things, from court historians making shit up to cows tripping up over a lamp light and burning whole cities, is a pretty futile exercise.

> There are fewer written records in the traditional sense

But still more written than you could ever hope to read in your lifetime.

You're ignoring the method by which these things are dated. First, if a date was written on the page (extremely rare), it may well have been after the date change happened. The vast majority of writings from the time are dated to known events, like eclipses, or by reference to other known things ("20 years ago, when Joe was born" and then finding Joe's birth records).

If calendars really did advance 300 years, we are living past the advance. That means that when we see a writing reference an eclipse that we know happened in what we call 900, it doesn't actually tell us anything about the calendar having moved or not. It only tells us that something was observed 1122 years ago. That could be 900 years after 0AD, or it could be 600.

Actually, almost every instance of dating anything back then depends upon few relative date calculations which are "known". The eclipse happened 1000 years ago, so that is year 1022, and this writing referencing it is from 1022, and so this scholar writing a generation later is from 1082. This inscribed rock was found under 1100 years of dust, so the people living here were from 922, and their language existed at that time, and all records we find from this civilization are now known to be from the early 900s. Now we can claim to have evidence that there was thriving commerce in the 900s with hundreds of written records about it, so of course the dates weren't changed.

If you dig into the records people cite as evidence against the phantom time hypothesis, the vast majority of them (literally every one that I've seen) is using indirect dating in this way, which doesn't mean anything.

> (weak)

It's actually as strong as your expectation of the evidence being there. If you are absolute sure of the evidence, look, and don't find anything, this is very strong evidence of the absence.

But well, on this case the absence is quite expected too.

Is there such thing as weak evidence?

Evidence that irrefutably proves something is very strong evidence.

Evidence that kind of points towards something, but inconclusively, is weak evidence.

For example, if 100 witnesses saw the accused murder the victim, and they're all credible, and they tell corroborating stories, that's pretty strong evidence.

If 1 person thinks he saw the accused murder the victim, from the other side of a car park, but he has bad eyesight anyway and it was dark and raining, then that's weak evidence. It's not nothing, it's still evidence, but it is weak.

Sort of.. at least there is in the scientific tradition.

Your theory predicting reality is generally weak evidence. It opens your theory to invalidation if the prediction fails, but does not rule out hypothetical alternative theories that also make the same prediction.

Theories aren't evidence at all, what could you mean by this?

A particular theory is a glue potentially connecting some number of observations. The observations + theory are evidence.

The reason that a messy pool of blood on the floor is evidence of violence is the theory that the blood came out of a person in a messy way.

If you're bayesian.

Wouldn't that mean the quantity vs quality of conspiracies though?

This seems to be equivalent to asking if there is any such thing as strength of evidence at all, and so yes, of course. The evidence that 100mg of cyanide will kill you is very strong, the evidence that seed oils cause inflammation is weak, even though there's a lot of it.

Reply on news.ycombinator.com