JAVAScript ASCII art:
Simple RLE Compression

(Reference times: IE4.0 = 4sec. NN4.5 = 37sec.)

Text file size: 9.41 Kb
New file size:  3.8 Kb (Only decompression script and data)

This kind of compression is only efficient if the file has many character repetitions. The compression is low but therefore the algorithm is very fast. Nevertheless it takes with JAVAScript some time to decompress the files. After opening this file you have seen a empty document before the text was displayed. The download has finished after the black background is visible and the following time is needed to decompress the data.
A other algorithm might have higher compression rates (LZE...), but the decompression time would increase. I think best results (compression rate - decompression time) will be reached by using binary encoded data. Especially if the algorithm only calculates the used characters (mostly <30 in ASCII art). May be next time...
Netscape needs much longer (~4x) than IE to decompress the data.