Manual vs. Code

Immortal Barons is a from-scratch clone, so every mechanic has to be sourced from somewhere. Barren Realms Elite ships a manual, a help database and a changelog, and they are cheap to read and usually right — which is what makes the exceptions expensive, because nothing in the text marks the wrong line.

These are the places where the original's own documentation turned out to describe something the shipped program does not do, plus a couple of outright defects in the program itself. Each says what the clone does about it — and in almost every case the answer is that it follows the program, not the paperwork. The exception is the last entry, where the original is not wrong so much as terse, and readability won.

This page is not a list of the ways Immortal Barons differs from the original by choice; those are recorded per mechanic in the developer reference.

The order of preference behind all of this: a screen captured from a running copy beats a reading of the compiled code, a reading of the code beats the text strings inside it, and those beat the documentation.

Where the documentation and the program disagree

A quick strike hits harder than advertised

The attack help puts a quick strike at 110% of normal strength. The invasion resolver (BRE.OVR 0x4055a) switches on the arriving attack's type byte and multiplies the incoming force by a 6-byte Turbo Pascal real: 1.2 for a quick strike, 0.85 for an extended battle, and nothing at all for a normal one.

What settles the reading is the second constant each branch loads — the fraction of its force a side has left when it breaks off:

Attack type	Strength	Breaks off at	Losses
Quick Strike	1.2	0.92	8%
Normal Attack	1.0	0.85	15%
Extended Battle	0.85	0.80	20%

Those loss figures are exactly the three the same paragraph of attack.hlp publishes, which pins each branch to its attack type beyond argument. The help is right about three numbers and wrong about the fourth, written inches apart.

Immortal Barons uses 120%.

Declaring war is not the polite option

The manual presents a declaration of war as the way to end an agreement without causing trouble at home, and says the treaty does not really break until the other realm has been told.

Neither holds. On confirmation (BRE.OVR 0x01a838) the original divides both popular support and military morale by four and multiplies by three — integer division first, so 99 support keeps 72 rather than 74 — and clears the relation row on both realms immediately. Nothing waits on the notification, and the game's own on-screen warning about revolts contradicts the manual a few lines later.

The corollary is sharper: no attack path in the original ever looks at the relationship, so betraying a partner by attacking them costs nothing at all. The crown charges for the public act and leaves the private one to be punished by the other players.

Immortal Barons matches both halves. It used to charge for the betrayal and nothing for the declaration, which was the original's asymmetry exactly backwards.

Protective Trade has nothing to maintain

The manual says the pact makes trade deals cheaper to send and to maintain.

The send discount is real: create_trade_offer (BRE.OVR 0x0268bc) compares the recipient's relation against the Protective Trade value and divides the per-day transit cost by three, before the span is chosen. The other half describes nothing — the cost routine is called from two sites, charged once up front, and no recurring trade cost exists anywhere in the program.

Immortal Barons implements the discount and does not invent an upkeep to discount.

A High terrorism setting makes terrorism rarer, not commoner

The sysop help for the Terrorist Costs option (game/reset.hlp) says the choice sets what an interplanetary terrorist operation charges, and that "a High setting will make terrorism more popular".

It does the opposite. The pricing routine (BRE.OVR 0x02aca8) reads the option's byte and, for High, multiplies the price by three; for Low it divides by five, and for None it zeroes the charge. The byte's encoding is Medium 0, None 1, Low 2, High 3, which the settings screen's own name lookup confirms. Tripling the price of an operation makes it rarer.

The sentence reads as though someone described the setting from the sysop's point of view — a high setting turns terrorism up — while the code turns the cost up. Nothing else in the entry is wrong, which is what makes the line easy to take at face value.

Immortal Barons follows the code: None free, Low a fifth, Medium the base price, High triple.

Where the changelog describes a version you do not have

Worth separating from the two above. A changelog can be a perfectly truthful record of a build that is not the one that shipped, so an entry being wrong about v0.988 does not mean it was ever wrong.

Tanks do not defend against chemical missiles

The changelog announces that they do. No chemical, biological or nuclear routine in the shipped version reads the target's tanks, turrets or missile defence at all — the only things any of them look at are the realm's name, its region count, its population, its morale and its support.

Immortal Barons used to reduce all three weapons by the target's SDI. That was the clone's own invention and is gone.

Missile bases do not stop a neighbour's nuclear strike

Same shape. The local nuclear routine reads a realm's name and region count and nothing else. The "up to 50% of incoming missiles" the documentation advertises belongs to the interplanetary path, not to a strike from the realm next door.

Immortal Barons lets a local strike always land.

Defects in the original

Not a documentation problem — the program does something its author plainly did not intend. A clone has to decide case by case whether being faithful means reproducing the bug.

Spying ignores the defender entirely

The success roll for Send Spy is built from two calls to the same strength routine, one meant for the attacker and one for the defender. Both are passed the attacker. The instruction bytes are identical — 8A 46 10 at both 0x4BAE7 and 0x4BB03 — and only the mode byte differs.

With no alliances in play the two terms reduce to agents / kind and agents, the agent count cancels, and the roll collapses to a constant:

chance = 0.1 + 0.9 x 1/(1+1) = 55%

Fifty-five percent, whatever either realm holds. Buying agents does nothing for spying, and defending against a spy is impossible.

It reads as a slip rather than a design, because the same routine is called correctly at 0x4AA5E, passing a different empire with the other mode.

It also has a consequence its author cannot have wanted. A Terrorist Prevention treaty raises the defending term — which, because that term is really the attacker's own agent count, lowers the holder's own spy success. Signing a counter-espionage pact makes you worse at espionage.

Immortal Barons reproduces this, defect included, as a deliberate choice. It was written here that only Send Spy went through the defective routine and that the effect operations kept a correct attacker-against-defender roll; walking the resolver's call sites showed otherwise. Every local operation reaches this one roll — the effect ones by way of the queue daily maintenance drains — so none of them is defended against by agents.

Mistaken for a defect

The traffic runs both ways, and calling something a bug closes the question early.

"Funding / Region: 0,000 Gold"

The missile-defence screen prints that line at every funding level, up to seven million. It was written down here for months as unexplained and probably a fault.

It is a display convention. The program stores the funding in whole thousands and the screen appends a literal ,000 to whatever it prints. The per-region figure is smaller than a thousand, so it renders as 0 and the line reads 0,000 — correct by its own rules.

It only became clear while deriving how much shield a given funding level buys, which is held in the same units:

strength = trunc( sqrt( funding / (10 x (regions + 1)) ) )     clamped to 0..100

(BRE.EXE 056d:1139). That formula reproduces all sixteen captured screens exactly — and only does so if the stored funding is read as thousands.

Immortal Barons prints the gold figure the label describes — the one place on this page where the clone does not follow the program. Nothing is being corrected: the original is right by its own convention, and a line that reads 0,000 at every funding level simply tells a player less than the number does.

Reading this the right way

None of the above is an argument against the documentation. Sweeping it first is still the fastest way to learn what a mechanic is and roughly where to look — between them, the manual and the help gave the whole treaty catalogue, the group-attack timing bounds, and three of the four attack figures correctly.

The rule that came out of it is narrower: read the prose to find the question, and read the program for the number you are going to rely on. Where the two disagree, the program wins, and the disagreement is worth writing down next to the answer — otherwise the next person to look will quietly correct it back.