What we learned with the Atlas Shuffle (A Discussion)

So it’s been a couple months since the atlas shuffle was reverted, hopefully that means we can have a good discussion about it. My goal is to summarize what happened, what was good, and what was bad. Before writing this I’ve taken an extensive amount of time in line chats asking questions and listening to discussions about the shuffle. Please feel free to leave constructive comments about the shuffle below.

First, why did the shuffle happen in the first place? Atlas was stagnant. There was little (unplanned) turnover of castles. Many teams were also hard to access in atlas, being multiple bubbles deep.

Here’s PG’s official reasons as to why:

  • Have a home you and your team care about enough to defend.
  • Have valuable territories that are worth fighting over.
  • Rethink the incentives that make fighting over castles or territories worthwhile.

So what went wrong?

  • The ratio of accessible castles varied significantly between different teams
    • Some teams found themselves with more than 50% of their castles accessible, some had 10% or less accessible, and others had anywhere in between.
    • With such variability in the number of access castles, it was an unfair situation where some teams were stretched thin defending and others could be purely offensive.
  • 5 TAs were supposed to be grouped together
    • While most TAs were grouped together as intended, many found themselves divided or missing a team in the main cluster
    • This would have made it difficult for 5TAs to defend all their teams as easily as was intended
  • Pirate teams got more of an advantage
    • With the shuffle, pirate teams had access to more teams without ways to retaliate against them

What went well?

  • There was no Safe Zone access
    • Without SZ access teams could no longer instantly spawn next to an opposing team’s castles and hit with little delay
    • There is still NML access, so team’s can still attack opponents without receiving passage or bubbling in, but they can’t spawn directly next to the castle they’re attacking
  • Gold mines went away
    • The rewards from beasts add up, it may seem minor but the change from gold mines to beasts will benefit players with additional prizes in the long run, especially if players weren’t already attacking beasts
    • You also didn’t have to travel across the map to reach beasts with elements you wanted
    • Beasts indefinitely spawned in land
  • Teams that were not previously accessible could be hit
    • In the current map, some teams can’t be hit without receiving passage or bubbling in, the shuffle made retaliation against these teams easier
  • 5TAs were clustered
    • This hypothetically would have made defending your allies a lot easier because you wouldn’t have to travel across the map to get to each other’s castles

Now, ultimately the goal is to try the shuffle again. Which the kinks get worked out, this is projected to be sometime in early 2021.

8 Likes

Very nicely summed up!

Ngl, I’m in no rush for another shuffle, would rather we get other changes implemented first - but I do think that in light of the recent blockade times, offensive wise, it would be fun to pair it with a well implemented shuffle where access isn’t skewed towards some teams and others got to hide

So here’s to hoping that the next version whenever it comes along does better

4 Likes

I would add my personal what went wrong is that the map was no longer a map but a mass of hexes that showed very little variance. A rethink on how the spiral looks, feels and works is needed. Teams that are of equal strength can be found on opposite sides of the map.

Personally I think the map would benefit from more diversity and greater grouping of like teams to like teams while maintaining separation between higher and lower ranked teams… something like the gauntlet islands would visually and logistically work for separation and differentiation.

5 Likes

I like the idea – reduce the depth of some castles. Right now, some teams have all their castles 3 or more levels concealed, so they’re like pirates – able to attack without recourse. The spiral NML concept seems more ideal, with all castles closer to access. The reduced delay timers have also opened up a bit more fun access to concealed castles. If only everyone didn’t have a bot. :stuck_out_tongue:

I like the endless beasts. It would be great if you could farm beasts on your 5TA’s castles so you have more access to different shard types.

3 Likes

In my view you missed a few things on what didn’t go well.

  1. The map went from something that looked like a real world with oceans and a coastline into a hideous blob of colours.

  2. Castle text overlapped other castle text so that you could not read some castle names - it was just poorly crafted.

  3. Some teams were 100% inaccessible due to somehow being inside mountains. No way to access them whatsoever as there wasn’t even neighboring zones.

  4. Latency was terrible. My 5ta did some movement drills and atlas crashed on our devices. Prims did not move for ages it was very unresponsive with more than 5 prims moving.

  5. I think the issue of having bigger castles with immediate access needs to be addressed given they have bigger cooldowns and it becomes a full time job managing so many castles - not only your own castles but those of your 5ta.

  6. Lets also address the issue of rank #100 -200 teams or lower getting placed next door to rank #1-10 teams. There was an understanding that the map would spiral out with stronger teams in the middle.

  7. Which leads into the issue of weaker teams potentially being easily wiped out.

  8. Many teams used real money to trade up castles in safer areas. Or they gave up good castles on borders for positioning. Changing the dynamic like this placed a disadvantage on players that support PG with real money.

Then let’s address the elephant in the room. We didn’t even get to test the map combat engine to see if it worked or not.

Atlas 2.0 was a complete train wreck. The reality is Atlas 1.0 is superior and it is the one that should be tweaked - reducing blockade times has been a good start.

7 Likes

Visual bugs, they were in fact accessible but the roads overlapped with the mountains so weren’t visible (sometimes, 2 roads also overlapped so a castle one hop in was actually accessible when it appeared not to be)

This isn’t true - I’m not sure what the screenshots you provided are supposed to measure tbh, but before removal of gold mines, the breakeven point for gold from mines vs beasts was lvl 6
I had screenshots for this, I’ll try to see if I can find them, but no promises

3 Likes

Thanks Fiery - I’m sure your memory is better than mine, ill update the post.

( withdrawn post )

I never tried to goto any mountain zone just heard from others that they couldn’t get to them I don’t have any evidence to support this claim.

I don’t think most people knew about this :sweat_smile: Only people that dealt with game files etc were really aware of that and the true number of accessible castles people had - I wish I still had screenshots for this, but I don’t think I do

I did find the ones for mines vs beasts though, I’ll edit this post shortly with them when I switch to my phone

Edit: @Bloodsworn

Mines vs Beasts

Here’s the screenshots showing mines vs beasts before they went away, all compared in SZ for fairness (beasts/mines on your castle regions give you more stuff, I’ll add extra ss I took rn showing this)
You get to see that the breakeven was lvl 6 beasts, and you start netting positive beyond that :slightly_smiling_face:

NML vs Castles

Do note, both my ice and dark shard bonuses at the time of taking these screenshots are equal, with equal multis and beast lvls
Usually there’s gold differences too iirc, but it wont show with the rounded values for gold, I can double check this when I go through my multis

3 Likes

@FieryxFury

Found the post

It seems all beasts are now supposed to pay what a gold mine used to pay as a minimum.

Right now L1-7 mobs pay way less than a gold mine.

So I think its fair to say those beasts need to be increased in payouts.

We have different gold bonuses, so can’t directly compare your screenshot to mine, but I’ll try to snap a ss when I get to the same multi as in the ss I posted for comparison

Edit: To elaborate - I’ll see what a lvl 1 beast pays at a x25 multi when I get there to see if it has really been fixed
But it’s normal for you to get less gold than I do in the screenshots I posted due to differing gold bonuses (mine is maxed)

2 Likes

From the looks of how this happened, it looks like the ⅓ ratio covered the Atlas board as a whole, not to the individual teams. Meaning ⅓ of all castles were access.

If this was the problem, then the simple fix would be to convert the ⅓ ratio to be team specific, which for me is something that I would be wanting to see.

3 Likes

Shuffling that map reduces it to a season long event :nauseated_face:
It will only promote piracy and each shuffle will make it feel even more pointless!
So I hope pg hurries up so we can set it down as a option and hopefully start to actually address the mechanic that created the problems.

image

5 Likes


This button …I blame this button :rofl:
Never push that button :man_facepalming:

1 Like

When I saw the new map after shuffle, I just thought that someone who saw a hex density plot in Tableau thought “Hey, I could do the same on my game’s map, assign each hex a value based on [insert any arbitrary value we choose to assign to hexes] and randomise/aggregate based on that value”.

How it ended showed that, IMO, it had all the best intentions but was not properly tested to show fundamental flaws that contradicted the main goals they were trying to achieve.

So far, the best thought out alternative I have seen posted is likely to be MALIK’s proposal.

That’s basically what I thought it reminds me of the pick and place program i used for selective soldering and it has a optimize button and when you push it you tend to get the same results we got with the shuffle! :crazy_face:

So apparently the Russian guy who made the optimize button lost his mind and they had to have another guy finish his work and I believe he lost his mind as well. :man_shrugging:
Apparently the math needed to make an optimize button will drive a human mind to the brink of madness much like that shuffled map did. :rofl:

When I started working the board programs for the selective solder machine had run times of 15-20 min per board when I left I had the run times down to 2-3 min per board and I made a sticker that said see this button “optimize” never push this button it will not save you run time :rofl:

1 Like

This right here :point_up::rofl: it’s never optimal lol


I have hope pg takes a very close look at positional dominance and positional attacking as a solution.
And moving league rank to map supports and drives that escalation while incorporating core game and map completely for the first time and offers us a solution to teams sandbagging.
Adjust the drop density and bracket that density by level and our games economy will balance out as well :man_shrugging:We might actually see player retention rise at the lowest and highest levels of play.

Dear God… 2 minutes delays and no castles more than 2 deep (if they code that part right next time). That would be a defensive nightmare for teams with more than a couple of castles.

Can I buy stock in bullhorns somehow?

Yep offense and defense are in equilibrium and we can’t afford the burden of play constant defense will produce.
Nidogod posted a link to my thread outlining what keeps them in equilibrium.

Good. Let there be more castle turnover

4 Likes