The general purpose of the KOI was to allow Russian text to be transferred between a computers and to be anyway readable. Thus, every Cyrillical character was placed in order to math a Latin one. This was resulted in a way when the Russian text is not decoded as Russian then you can see a Latin characters phonetically mathing an original, e. g. the text is still readable. However, not every character had got a Latin analogue, due to the Latin alphabet is smaller than the Cyrillic, but they are most rare and minimal.
For computer's internal use there was the Binary Code for Information Computing, but he is began to be superceded by the new KOI. Thus, what's would be happen, is happening. The USSR makes the Dialogue Computing Complex, which was introduced a reversed situation € an English text being decoded with the Cyrillic characters. Due to the KOI was do not designed for this case, such "transliteration" which you take looks very specificaly.
When your system falls to decode a 7-bit text as an ASCII-only text, every byte is to be interpreted with an ASCII notation. Then a caracters are tend to change in the following way:
¥ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¨
¡áâ÷çäåöúéêëìíîïðòóôõæèãþûýÿùøüàñ¡
¥ ¾ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ µ
¡upcas¡abWgdeVzijklmnoprstufhC~{}delYX|`Q¡
°€€€€€½€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€Š€Š€Š€Š€Š€‰€Š€Š€Š€´
¡locas¡ Reversed case ^[]_Rev\@Rev
« » ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ®
NOTE that here all uppercase begins lowercase and all lowercase begins uppercase. In this table every important character changes (e. g. every non-intuitive change) are typed in uppercase. All others appear just as they will do practically € in uppercase from lowercase. Anyway there is no the letter IO, thus I can't say anything about it.
When it is transliterated vice versa the only four characters are changed not as it may be the best, but they four are able to cause a pretty comical view. To have know how it can be seen look at this table:
¥ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¶ ¨
¡ABCDEFGHIJKLMNOPQRSTUVWXYZ¡
¥ ¾ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ µ
¡upcas¡ÁÂãÄÅÆÇÈÉÊËÌÍÎÏÐñÒÓÔÕö×øùÚ¡
°€€€€€½€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€‰€´
¡locas¡ Reversed case ¡
« » ®
The case handling is as in the previous situation. For some other characters you can see the previous table.
Here you can see that several letters are being replaced not as phonetically intended. The C letter, which usually means same as K, appears as a cse. The V letter appears as zhe. The X and Y begins a soft sign and a hard kind of an I. The presentation of Latin words gets changed and with a policy Russians writes a consonal words "what you hear is what you write" such mismatches are to be spoken. That's why there was a meme "iNZHALID DEZHICSE" thet is generaly based on a transformation of the letter "V".
The Dialogue Computing Complex writes to the display a seven-bit data in a KOI-7 notation. A current character set is set by a switch (may be wrong, I cat't proof). It is not controlled by a software. The software may output either Cyrillic and Latin text data. A videocard can output only one of that. As a result, in one state only partial text, Cyrillic or Latin, can be correcnly displayed. To read a text on an other language you need to toggle the switch. And just this is caused the meme to appear € that's how "other text" is looked-like.
These codings are practically feeled similarly and your software can be confused with they. However they are different encodings and mistakable decode makes a document unreadable. A ranges of character placement is similar, only the case is reversed from the other. But in those fields characters are distributed differently. I currently will not write here an example because I want a some source that is currently unavaible. In a fact when you pass some Cyrillical stritg 20 times from the KOI-8 to the Windows-1251 it gets back readable.