No problem with that change if unpopulated controls go back to having the same difficulty as before... let's have an opportunity to actually direct a defense away from populated areas instead of just playing whack-a-mole.
Analysing the Thargoid Simulation
- Thread starter Ian Doncaster
- Start date
Most likely reading I think is that the previous order was
"pick the system with the highest Tastiness factor" (exactly what determines that we haven't figured out)
and the new order is
"pick the inhabited system with the highest Tastiness factor, or if that's not possible, the highest uninhabited system"
By speeding up their outbound expansion it might mean that they get to the point where their surface has so many targets that they don't instantly re-invade, perhaps?
"pick the system with the highest Tastiness factor" (exactly what determines that we haven't figured out)
and the new order is
"pick the inhabited system with the highest Tastiness factor, or if that's not possible, the highest uninhabited system"
Yes, though that was pretty likely for most systems anyway, since if it was Tastiest earlier, in general all six weeks of fighting over it will achieve is that every system around it gets captured or invaded as well, so it'll be the only free one in that subregion.
By speeding up their outbound expansion it might mean that they get to the point where their surface has so many targets that they don't instantly re-invade, perhaps?
Okay, I wasn't expecting that to be what they meant: do the same number of inhabited systems as normal, but just don't take the 57 extra uninhabited systems they'd previously have the budget for. The actual number of inhabited systems hit is actually down to its lowest since 9 February.
And when they are hitting uninhabited systems, they're not necessarily good ones - Hadad has picked Col 285 Sector BU-X b15-1, for example, which would be an okay use of a spare alert point to complete the budget under the old rules but seems massively frivolous when they only get 5 total.
And when they are hitting uninhabited systems, they're not necessarily good ones - Hadad has picked Col 285 Sector BU-X b15-1, for example, which would be an okay use of a spare alert point to complete the budget under the old rules but seems massively frivolous when they only get 5 total.
Last edited:
Are there any current alerts that don't have any systems within 10LYs (besides the control system that triggered them) ?
Did a quick check on the (lack of) expansion of Thargoid space.
This is (roughly) systems which have had more than 20% pushback activity each week, so mainly inhabited ones. Still, there haven't been very many inhabited systems further out attacked at all.
It's incredibly flat - the Thargoids seem to be very strongly prioritising in-fill of their existing space, so that's where the contested systems end up. There's also, of course, a lot of re-attacking of previously defended systems.
Based on that I'd suggest a slightly broader theory of Thargoid Alert placement: very difficult to rigorously prove, but probably a decent working model.
This is (roughly) systems which have had more than 20% pushback activity each week, so mainly inhabited ones. Still, there haven't been very many inhabited systems further out attacked at all.
It's incredibly flat - the Thargoids seem to be very strongly prioritising in-fill of their existing space, so that's where the contested systems end up. There's also, of course, a lot of re-attacking of previously defended systems.
Based on that I'd suggest a slightly broader theory of Thargoid Alert placement: very difficult to rigorously prove, but probably a decent working model.
- Alerts must be placed within 10 LY of an existing Control, rounding to 1/32th resolution (near certain)
- Controls can only place an Alert at most once every two weeks (near certain) and not in their first week (near certain)
- Controls with a choice of multiple targets will usually pick the closest target (strong evidence, though "usually" is doing a lot of work)
- Where a Maelstrom has more usable Controls than it needs - i.e. have at least one target, weren't used last week - it will prioritise ones which placed their last Alert less recently (the new speculation)
- Thargoids tend to re-attack "inner" systems as soon as they get chance (immediately after Invasion-Recovery, 3 weeks cooldown after Alert) - which would make sense as they'll be the only remaining target for several Controls which won't have fired recently
- Thargoids tend to expand their sphere cautiously and prioritise in-fill, except in Leigong/Thor where connectivity limits them - again, the more inward control systems are only going to be able to reach out to 25LY when they fire, whereas a Control at 30 LY has a decent chance of its next closest system being "backwards" or at least no/not much further out.
- "Outer" systems don't tend to be reattacked in the same way - lots of systems > 25LY out have not been reattacked once cleared. There are by surface area a lot more outer controls, they have a wider choice of targets (including in-fill) and it may take quite a while to get around to them again, especially in those maelstroms where inner systems are being repeatedly cleared.
The problem with those observations is that there's less systems per uncontrolled system on the edges than inside so even if the attacking systems were picked randomly those outcomes could be more likely.
Having cooldowns seems messy and I still think any AI should be able to take just the systems/states and make a move based on that (Controls not alerting on their first week would be because they're placed after the alerting phase is done).
But there is precedent for that type of stuff in the BGS I guess.
It would actually be more interesting if all progress wasn't just a progress bar with a binary outcome and you could do something other than achieve total victory in a system to prevent it from being a source for an expansion next tick.
I can't see any other explanation for the early Alert oscillations going between 20 points and 0 in alternate weeks. Without some sort of cooldown a maelstrom should never have no targets.
Having Control states have an invisible "Control-Used" state would still keep the state machine simple.
That's true, but on the other hand the number of control systems at a radius goes up with the square of the radius, so pure random selection should be picking outer ones a lot more often than inner ones.
An alternative explanation might be that the control systems are indexed in order of either capture or range, and it goes through checking the earliest/lowest first, which would again bias towards inner ones without needing as much tracking of state.
Couldn't that be explained by ongoing invasions?
It would seem that the expansion is more limited than our fevered imaginations projected.Did a quick check on the (lack of) expansion of Thargoid space.
This is (roughly) systems which have had more than 20% pushback activity each week, so mainly inhabited ones. Still, there haven't been very many inhabited systems further out attacked at all.
It's incredibly flat - the Thargoids seem to be very strongly prioritising in-fill of their existing space, so that's where the contested systems end up. There's also, of course, a lot of re-attacking of previously defended systems.
Based on that I'd suggest a slightly broader theory of Thargoid Alert placement: very difficult to rigorously prove, but probably a decent working model.
That last rule in combination with the others would explain a lot of the "big picture" observations.
- Alerts must be placed within 10 LY of an existing Control, rounding to 1/32th resolution (near certain)
- Controls can only place an Alert at most once every two weeks (near certain) and not in their first week (near certain)
- Controls with a choice of multiple targets will usually pick the closest target (strong evidence, though "usually" is doing a lot of work)
- Where a Maelstrom has more usable Controls than it needs - i.e. have at least one target, weren't used last week - it will prioritise ones which placed their last Alert less recently (the new speculation)
- Thargoids tend to re-attack "inner" systems as soon as they get chance (immediately after Invasion-Recovery, 3 weeks cooldown after Alert) - which would make sense as they'll be the only remaining target for several Controls which won't have fired recently
- Thargoids tend to expand their sphere cautiously and prioritise in-fill, except in Leigong/Thor where connectivity limits them - again, the more inward control systems are only going to be able to reach out to 25LY when they fire, whereas a Control at 30 LY has a decent chance of its next closest system being "backwards" or at least no/not much further out.
- "Outer" systems don't tend to be reattacked in the same way - lots of systems > 25LY out have not been reattacked once cleared. There are by surface area a lot more outer controls, they have a wider choice of targets (including in-fill) and it may take quite a while to get around to them again, especially in those maelstroms where inner systems are being repeatedly cleared.
I keep looking out the bridge windows of my carrier in Colonia, and wait for the "secret triangles" to arrive...
Doubt it - think about Leigong's first few weeks. (Similar oscillations were seen at most maelstroms in their first weeks, though)
Date | Alert | Alert empty | Alert full | Invasion | Controlled |
01/12/22 | 0 | 0 | 0 | 5 | 11 |
08/12/22 | 11 | 11 | 0 | 5 | 11 |
15/12/22 | 0 | 0 | 0 | 3 | 24 |
22/12/22 | 17 | 16 | 1 | 2 | 25 |
29/12/22 | 2 | 2 | 0 | 1 | 43 |
05/01/23 | 14 | 12 | 2 | 1 | 45 |
If there's no cooldown, those 3+ controls should have been able to hit targets on the 15th too (as the target set on 15th and 22nd will have been identical for them). So there needs to be an explanation for why they didn't.
A one-week cooldown after use is the simplest one I can think of (and the recent prediction for Thor would have incorrectly predicted more Alerts had it not been assumed)
You've probably already considered it, but does the presence (or absence) of Amonia worlds or Gas giant with ammonia-based life type bodies (possibly smaller/light atmo bodies with amonia atmospheres) in systems play into any of the calculations?
Ammonia Worlds are rare enough that there's generally only one or two (other than the Maelstrom itself) within their ~30 LY operational range anyway. They don't seem to be heading specifically in the direction of any others.
Ammonia atmospheres / AL gas giants are much more common but I've not seen any evidence that they're considered in general. (Though "closest first" isn't a complete explanation for their alert selection, so that could play a part in it)
This is sort of what i was getting at. Like if "closest with planet type X,Y,Z" (within x range) first, or maybe presence of x/y/z planet type affects the "point" cost (tat you folks here established earlier) of moving a system status (to alert, or to controlled, watever?)
Do we have past data on how many ports are attacked per system? Maybe doing an invasion where 2 ports are lost/week costs more than doing an invasion where only one port is lost? A bit of a nightmare to figure out the extra cost per attacked port is not a whole number.
Were more individual ports attacked this week than other weeks and did this cause the number of systems attacked to go down?
Were more individual ports attacked this week than other weeks and did this cause the number of systems attacked to go down?
This week: 40 alerts again, 24 uninhabited, 16 inhabited. It doesn't look like the priority order has changed at all - lots of the uninhabited ones are "backwards" direction still, so the cost increase for uninhabited alerts is just making it even easier for them to crowd out inhabited ones.
The Thargoids are "backfilling"? Odd. I thought they would have a line/axis of advance.
They appear - except for Leigong and Thor which are heavily constrained by connectivity - to be prioritising filling in the closest systems to the Maelstrom first, in general, and expanding approximately spherically regardless. So both in-fill of their existing "spheres" and placing of new Alerts significantly further away from the bubble than the Maelstrom itself are very common.
It's an extremely cautious strategy but therefore has the advantages of not really having exploitable vulnerabilities. The disadvantage is that it leaves almost all of our military infrastructure intact for quite some time too.