Dumping Archon, creating a new tournament guiding app?

Sambomb wrote:

Lets say 1 and 2 drop early, and turn the tables on: (round 3 for 17 players edited to drop 2 players)
13 8 6 11 16
15 5 14 3 10
7 4 12 17 9

Instead of: ( normal r3 for 15 players )
13 15 2 5 11
14 3 10 1 6
12 8 4 7 9

Your proposition for R3 isn't good because 10 would have 15 as a prey twice.

This is a bad luck scenario, but what I would do is enter the 17 players, and override round 1 and 3.
Round 1 is easy: 3 tables of 5 players 1-2-3-4-5 / 6-7-8-9-10 / 11-12-13-14-15
Round 2 is defined by the archon: 9-14-10-15-7 / 5-12-4-2 / 3-13-17-6 / 16-1-11-8
Round 3 is the tricky one, because you don't want rules 1 and 2 to be broken. The default archon seating for 17 players and round 3 is:

13-8-6-11-16 / 15-5-14-3 / 7-4-1-12 / 2-17-9-10

Since players 1 and 2 drop, we remove them and consolidate the tables:

13-8-6-11-16 / 15-5-14-3-7 / 4-12-17-9-10

Now we must check that the rules are ok. We can see that 9 has 10 as a prey twice (on R1 and R3), which is not good.
Also, 9 and 10 (as well as 14 and 15) can't be on the same table once more because they were on the same table on R1 and R2.

If we swap 9 and 14, then 14 is the predator of 10 a second time.
If we swap 9 and 15, then 9 is the prey of 7 a second time.
If we swap 10 and 14, then 14 is the prey of 9 a second time.
If we swap 10 and 15, then 7 is on the same table as 10 for a third time but rule 1 works.

The 4th option is better, so I think we could reasonably stop there.

13-8-6-11-16 / 10*-5-14-3-7 / 4-12-17-9-15

But as you can see, it becomes very complicated to find a new seating for R3 because the seating of R2 doesn't help at all (it was created with other constraints that don't match 15-players constraints).

Sambomb wrote:

About the app, and if the app import optimal seating from the current archon and try to enforce it each round.
If late arrivals or early drops happens and tables need to be fixed, the skimflux algorithm can be used to fix them in a smother way.

Such an algorithm doesn't exist I think.
And trying wildly to find the optimal seating for each round is sometimes impossible (because it would have required the R1 or R2 seating to be different) or would take too much time.

Sambomb wrote:

And if we use the 15 people optimal seating again at round 3, replacing player 1 with 16 and player 2 with 17?

13 15 2 17 5 11
14 3 10 1 16 6
12 8 4 7 9

Then 7 has twice 9 as a prey.
Don't try to do some guesswork, it's wasting our time.