(29-01-2026, 01:18 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.Mixing the words will destroy the LAAFU properties of the text
But that is my point, again. If you mix the words and then format them into paragraphs with the trivial line-breaking algorithm,
you will still get significant "LAAFU properties". Or if you keep the order of the words but format them using a max line length that is 1.6180559 times the current one, or 0.6180559 times the current one.
Will they be the same as the "LAAFU properties" of the the unscrambled text? Or equally strong? I don't know; but I won't be impressed by LAAFU claims unless they come with such control experiments.
All the best, --stolfi
(29-01-2026, 02:51 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.But that is my point, again. If you mix the words and then format them into paragraphs with the trivial line-breaking algorithm, you will still get significant "LAAFU properties". Or if you keep the order of the words but format them using a max line length that is 1.6180559 times the current one, or 0.6180559 times the current one.
Will they be the same as the "LAAFU properties" of the the unscrambled text? Or equally strong? I don't know; but I won't be impressed by LAAFU claims unless they come with such control experiments.
I'm not sure what difference this makes here. There is clearly some positional dependence within the lines, which could explain some of the vertical patterns.
(29-01-2026, 03:28 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view. There is clearly some positional dependence within the lines
But, again, the point is that such
positional dependencies are created by the trivial line-breaking algorithm. For any text, any language, any page width. Even for algorithmically generated gibberish. As long as the words come in different sizes. Because a line break will be more likely to be inserted before a long word than before a short one.
In fact, the first word being longer than average is indirect evidence that the Scribe was copying some draft text, ignoring its line breaks, and putting new line breaks on his own. It rather excludes the possibility that the Scribe was generating each word at random after deciding to start a new line. Or that the line breaks were determined by the Author, and he created each line the same way.
All the best, --stolfi
(29-01-2026, 03:43 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.But, again, the point is that such positional dependencies are created by the trivial line-breaking algorithm. For any text, any language, any page width. Even for algorithmically generated gibberish. As long as the words come in different sizes. Because a line break will be more likely to be inserted before a long word than before a short one.
This is obviously true, but I'm not sure how this is relevant. No matter how the positional dependence is created, it can affect vertical patterns.
(29-01-2026, 04:02 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.This is obviously true, but I'm not sure how this is relevant. No matter how the positional dependence is created, it can affect vertical patterns.
Ah, OK, sorry. I though you were defending the LAAFU theory.
All the best, --stolfi
(29-01-2026, 04:21 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.Ah, OK, sorry. I though you were defending the LAAFU theory.
I agree that "LAAFU" is a bit strong term for simple positional dependence, it implies some "functional" role. But I think most people understand what features I refer to when I mention LAAFU, so I keep saying LAAFU. Since I consider the cipher hypothesis the most likely, I think that the positional dependence is an emergent properly of the cipher used (say, if encoding it reset at the beginning of a line or somehow wrapped at the end of the line). The line-breaking of a normal plaintext could explain some positional dependence as well.
(29-01-2026, 04:38 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.But I think most people understand what features I refer to when I mention LAAFU
I suppose a lot of the LAAFU research and discussion has happened during the 20 years I have been hibernating, so I am not sure I know what those features are, exactly.
I understand that the plot of the frequency of X as a function of the position along the line has bumps and dips near the beginning of the line and near the end of the line, for various choices of X. Is that all?
I suppose that the stats consider only paragraph text minus the head and tail lines, correct?
And I suppose that the position can be measured in words, EVA characters, "elements" (like "aiin", "ch", "cth"), or other, and divided by the length of the line, correct?
(I came to believe that measuring line length and position EVA characters may be the best option. Counting words has the problem that transcribers will make errors when deciding whether a space is a work break, and those errors probably vary near the ends of the line. Counting "elements" makes the results potentially dependent on the definition of "element" that one uses... And anyway the choice should have a small impact on the relative possitions.)
All the best, --stolfi
(29-01-2026, 08:10 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.I suppose a lot of the LAAFU research and discussion has happened during the 20 years I have been hibernating, so I am not sure I know what those features are, exactly.
I understand that the plot of the frequency of X as a function of the position along the line has bumps and dips near the beginning of the line and near the end of the line, for various choices of X. Is that all?
I don't think it's only bumps and dips, often there is some gradual change in the frequency of X as the function of the position.
I think one good summary is this article by Patrick Feaster: You are not allowed to view links.
Register or
Login to view.
I remember studying these a few years ago, but I couldn't find any good explanation, other than the generic thought that for a one-to-many cipher this could be something natural. Say, if T could be encoded as
ok qok or
qoke, the encoder would use the shorter
ok near the beginning of a line and then to avoid repeating same patterns would switch to
qok/
qoke later in the line.