To read it, use a magnifying glass or zoom in with a digital camera.
(Use the "macro" mode if it has one and hold it a few inches from the screen)
The text can only be seen on standard LCD (flat-panel) monitors.
Wikipedia explains why.
What's "millitext"? Millitext is (what I believe) the smallest possible representation of readable text on a computer monitor. It produces characters that are a couple of millimeters in size (hence the "milli") and are nearly invisible to the naked eye. Some examples are here.
How exactly does it work? Each pixel in an LCD monitor is made of three colored strips, red, green, and blue. The strips (called subpixels) are small enough that the eye blurs them together, combining the colors into a single dot. A pixel can appear to have one of millions of colors by using different brightness values for the red, green, and blue subpixels. Using the right color values, one can exploit this property and treat each subpixel as its own unit, packing three columns of dots into one pixel. More information about this is here.
How do I read such tiny text? If you have really good vision, and get close enough to your monitor, you may be able to see the separate subpixels and read the text. If you don't have a magnifying glass, point a digital camera at your screen. To get it to focus properly, set it to "macro" mode, and hold the camera a few inches from the screen. Use digital zoom if possible. The thick white border around the text will help the camera focus on the subpixel grid. If you have a Canon camera (like me), the Digital Macro mode works very well.
I don't see text, or I see backwards garbage. Millitext is only readable on standard LCD (flat-panel) monitors. CRT monitors don't use a regular grid of pixels, and some LCD monitors order the subpixels a different way. Wikipedia explains why in more detail.
Did you invent this? What algorithm is used to make the text? Subpixel rendering has been around for years (most operating systems use it for smoothing text), and I'm sure I'm not the first one to encode text/images directly into subpixels. I was just thinking about it randomly, and decided to do some experiments to see if it was possible. No special algorithm is used to create the text. You can make your own with any paint program.