+
+ 
+
+ The first step in solving this puzzle is diagnosing just what is wrong with + the file. Depending on your browser or text file viewer, you may be able to + read certain characters, but for the most part you will just see blanks or + replacement characters. The file begins with a UTF-8 byte order mark, but + the content of the file is not valid. The key is that the entire file is + encoded using what the UTF-8 standard calls overlong sequences.
+UTF-8 uses a variable number of bytes to encode each character. The number + of bytes used depends on the numerical value of that character. Characters + up to 127 are encoded in a single character, making it compatible with + ASCII. Characters a little larger than that are encoded with two bytes, + using bytes in the range 128-255, where the first byte also encodes how + many bytes make up this character. The next range of characters is encoded + in three bytes, and beyond that, four bytes are used
+Specifically, the way these multi-byte encodings work is as follows:
+-
+
- 110xxxxx 10xxxxxx encodes a two-byte character, where the x's + represent 11 bits of the character value. +
- 1110xxxx 10xxxxxx 10xxxxxx encodes character values up to 16 bits + in three bytes. +
- 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx encodes character values up to + 21 bits in four bytes. +
The standard goes on to define five-byte and six-byte encodings as well, + but subsequently a limit was placed on Unicode which makes those encodings + never valid. However, it is possible to use these (or the other encodings + above) to encode characters using more bytes than necessary. For instance, + you could encode an ASCII space, 00100000, as 11000000 10100000. The + standard declares such "overlong sequences" invalid.
+The file in this puzzle is a UTF-8 text file in which many characters are + encoded using overlong sequences which are just one byte too long. When you + manage to convert it to a valid text file, you get the following:
+
+Actually, you get several repetitions of that, each using different + lookalike characters which (to varying degrees) resemble the letters + they are supposed to be. When you manage to interpret this correctly, + what you get is:
++Some things are not as they seem to be. This holds true in many aspects +of life, as it does here. Sometimes, things are not as they seem due to +natural occurrences, but most often, somebody is trying to hide something. +For instance, people wear disguises when they want to appear as other +people, or simply not be recognized for who they are. You might say that +this text is a disguise, but it is an unusually helpful one, as by +reading closely, you may discover much about the nature of the true text +you have yet to read. "Bad, bad text," you say. You had to fix all those +errors. So many complaints about things being too long. Making a character +be the right length is nothing. Making it be too long is really something! +Don't go overboard, though; just one extra byte's enough. Each one of +us contributes just a little bit to the complete disguise, but in the +end it's all there. Finally, the disguise is too short, so it is repeated +many times over. This may help you deal with parts of the disguise you +can't figure out. ++
What this message is trying to tell you is that there is a hidden text + underlying the one you are reading. The encoding works as follows: + For each character encoded correctly, write a zero bit. + For each character encoded in an overlong sequence, write a one bit.
+The result of this is a list of Unicode character names, encoded using + lookalike characters again. Ultimately, this results in this list:
+-
+
- ARMENIAN CAPITAL LETTER AYB +
- HANGZHOU NUMERAL TEN +
- MALAYALAM LETTER TTHA +
- DIGIT SEVEN +
- CANADIAN SYLLABICS TA +
- LATIN LETTER BILABIAL CLICK +
- COMMA +
- SPACE +
- GEORGIAN CAPITAL LETTER PAR +
- HANGZHOU NUMERAL TEN +
- LATIN CAPITAL LETTER O +
- CANADIAN SYLLABICS CARRIER GHO +
- CYRILLIC CAPITAL LETTER IE +
- DIGIT SIX +
- COMMA +
- SPACE +
- N-ARY UNION +
- HANGZHOU NUMERAL TEN +
- APL FUNCTIONAL SYMBOL ZILDE +
- CYRILLIC CAPITAL LETTER ES +
- Z NOTATION BAG MEMBERSHIP +
- DIGIT SIX +
- COMMA +
- SPACE +
- TAI LE LETTER O +
- HANGZHOU NUMERAL TEN +
- CYRILLIC CAPITAL LETTER FITA +
- REVERSED DOUBLE STROKE NOT SIGN +
- DIGIT TWO +
- TAI THAM LETTER WA +
- COMMA +
- SPACE +
- YI RADICAL DDUR +
- HANGZHOU NUMERAL TEN +
- DIVIDES +
- DIGIT SEVEN +
- CHEROKEE LETTER GV +
- NKO LETTER EE +
- COMMA +
- SPACE +
- ANGSTROM SIGN +
- LISU LETTER NA +
- CANADIAN SYLLABICS CARRIER THE +
- SPACE +
- CANADIAN SYLLABICS TE +
- HANGZHOU NUMERAL TEN +
- APL FUNCTIONAL SYMBOL UP CARET TILDE +
- CANADIAN SYLLABICS CARRIER KHE +
- GREEK LETTER DIGAMMA +
- YI SYLLABLE TU +
- COMMA +
- SPACE +
- ARMENIAN CAPITAL LETTER BEN +
- BUGINESE LETTER SA +
- LISU LETTER ZHA +
- SPACE +
- TIFINAGH LETTER YADD +
- VECTOR OR CROSS PRODUCT +
- TAI THAM LETTER HIGH KHA +
- RUNIC LETTER EHWAZ EH E +
- REVERSED PILCROW SIGN +
- CANADIAN SYLLABICS MA +
- TAI LE LETTER TONE-3 +
- SPACE +
- LEFT PARENTHESIS +
- GEORGIAN CAPITAL LETTER CHAR +
- IDEOGRAPHIC ANNOTATION LINKING MARK +
- LISU LETTER SHA +
- SPACE +
- CHEROKEE LETTER TLE +
- EULER CONSTANT +
- CHEROKEE LETTER I +
- YI RADICAL ZZIET +
- VAI SYLLABLE TO +
- CANADIAN SYLLABICS TLHI +
- VAI SYLLABLE POO +
- RIGHT PARENTHESIS +
- FULL STOP +
Writing out all of those characters gives:
+
+or more readably:
+U+07C0, U+0AE6, U+0CE6, U+0F20, U+17E0, and U+ABF0, for example (six letters).
+The characters at the indicated code points are NKO DIGIT ZERO, GUJARATI + DIGIT ZERO, KANNADA DIGIT ZERO, TIBETAN DIGIT ZERO, KHMER DIGIT ZERO, + and MEETEI MAYEK DIGIT ZERO, so the answer which describes them in six + letters is ZEROES (not zeros, which is also a valid plural).
+ + + +