We cannot produce anything like the complexity of the internet presentations as we are working with one of the simplest microcontrollers, however we can show how to put a program together to produce almost any effect on a row of LEDs and 7-segment display. 
The RPS program does not contain any strategy or "thinking" but it can be designed for effect. 
The three letters on the 7-segment show "R P S" to indicate Rock, Paper, Scissors. This is the simple part. The lead-up to this involves flashing the 8 LEDs.
There are 3 different sections of the program:
1. The 8-LED display
2. The random number generator
3. The overall sequencing of the program. 

The 8 LED display can produce a number of effects such as "running,"  "flashing" or "splitting" and the 7-segment display can produce almost all the letters of the alphabet.
The results on the screen need to be random and this can be achieved by a random number generator, a table or by human intervention. 
We will be making the program run automatically and thus a random number produced by human intervention will not be possible. 
As you will see in a moment, computers are not capable of producing truly random numbers and so the solution is to have a long list of numbers in a table and the micro increments down the list. 
The overall sequencing of the program is a TIMING issue. Its just a matter of producing a suitable delay for each effect and running through it at the right speed.  

RANDOM NUMBERS
Computers are not very good at producing random numbers. 
There is nothing in a microcontroller that can be used to produce an unknown number ON A REGULAR BASIS. 
If the program is allowed to run automatically, it would be difficult to generate a random number by looking at the contents of a set of files, for example. You have no guarantee of getting an "even spread of results".
On the other hand, if you produce a table, the spread can be absolutely even or a slight advantage can be introduced. 
Even the smallest "odds" can reap an enormous reward. 
Casinos are supposed to have a "house advantage" of only a few percent, and on the roulette table it is about 3%. 
Suppose this "advantage" is transferred to a "poker machine".  It simply means you will lose all your money for every 33 plays!
It's simple mathematics. There are times when you win, but this simply increases the number of "plays" before you lose everything.
The longer you play, the more guarantee you have of losing EVERYTHING. 
It's an absolute certainty. 
Mathematic doesn't lie. 
The table used in our RPS game is not "weighted" and theoretically you should win exactly 50% of the time. 
This is a good predictor of human behaviour. 
Everyone thinks they can "beat the odds" and predict an event with a result better than 50%. 
This is incorrect. Humans are not as good as 50%. They average about 45%. That's why the average winnings on a table is about 5%. 
Take the RPS game. Play it 100 times and see the result. It's a perfect challenge .  .  .

CHIPS
The complexity of this game would require at least 6 chips if using older-style circuitry. The overall cost would be higher than a micro design and development time would be substantially greater. 
With our design, additional features can be added very easily and at very little cost, whereas a project made with discrete components would require complete re-designing.  
There is really no comparison, however some readers still need convincing of the enormous advantages of designing a project around a micro. 
If you are still working your way through the routines, try reading them exactly how the micro interprets them. They have been presented in a layout called "linear," with very few jumps up and down the page. All the files and values are visible so you can see exactly what is going on.  It's as clear as we can possibly make it. 
Any programs you produce can be designed along similar lines. They don't have to be "high technology." Go through the programs we have already presented. They will help you get a program together much faster than doing it all yourself.

THE SUB-ROUTINES
When writing a program, individual sub-routines are created in any order, as you think of them, and placed in the program in alphabetical order. Only write the code when you have the inspiration to create it.
For each sub-routine, a result must be generated and placed in a flag file or in a file that is not used by any other sub-routine. Don't branch-off from a sub-routine to another sub-routine. Go back to main and use the result from the sub-routine or flag file. 
If you need to do another calculation or produce another effect, write CALL xxxx instruction in Main. Go to the sub-routine, carry out the instructions and return to Main.
Each sub-routine must have a RETURN to Main. 
If a sub-routine increments a file, you must be sure the file has been incremented only once otherwise the result will be incorrect.  
These are the main requirements.

THE PROGRAM 
The first effect on the display is called the "attract" mode. This draws the player to the game. In our version the "Attract Mode" is similar to "Split," where the LEDs are removed from the middle until all are removed. The action is then reversed. 
Push button "A" to start the game. 
The display will produce a random effect and then a letter on the screen. 
The screen shows the letter for about 3 seconds and the cycle repeats. 
Compare the choice you make (rock, paper, scissors) with the display and work out the winner. 
The program can be extended by allowing the button to be pressed one, two or three times for rock, paper or scissors and this will be recorded. The display will show the score after say 20 games. 
Each sub-routine is very easy to follow.  No complex instructions have been used. 
But a few clever ideas have been introduced, such as . . .
1. In the Attract sub-routine, to put FF into a file, it is cleared and decremented. 
2. In the Run sub-routine, the RRF and RLF instructions move a bit across the display and back again. To detect when a bit has "fallen off the end of a file," the carry bit is tested.  
3. Before each letter appears on the screen, an effect, such as "Flash," "Run," or "Toggle," appears on the 8 LEDs. Both the "effect" and one of the letters is determined by a random number table. To prevent the effect coinciding with the letter on the screen, the jump value for the effect is incremented twice in the Display sub-routine. To detect this double-jump, the end of table needs two FF values.  
4. To test for End of Table, bit 7 is tested. Any bit above bit 2 can be tested in our case, we have used bit7 for convenience. There are three values in the random number table. To test these three values, bits 0, 1 and 2 have been used. This allows bit-testing. Bit0 = 0000 0001=1 , bit1 = 0000 0010=2 and bit 3 = 0000 0100=4. Bit testing is the easiest way to identify three values.