System Tax Report — Original SRD vs Multi-Phase Boss

How much does the new boss system "harden" the game? Three states of eight monsters were compared: (A) Raw SRD (single stat block, single HP pool), (B) Solo-adjusted SRD (action economy compensation, ~1.6× damage), (C) Our multi-phase boss (four tiers). Metrics: win rate, death multiplier, round count.

The headline finding up front: The system doesn't one-directionally "harden" the game — it standardizes it. Raw SRD bosses are chaotically scattered (some yield %16 win rate, some %100); our Normal tier gathers them all into a narrow, predictable band. The "Tax" isn't a harshness tax — it's a consistency tax.

Three Reference States

A and B are the two extremes of the original. Raw SRD as a solo boss is too easy (the party crushes the single target with action economy); the solo-adjustment is too harsh (1.6× damage drags most parties into a TPK). Real DMs tune between these two by feel. Our system does this tuning automatically and consistently.

Tax Table 1 — Win Rate: Raw SRD → Our Normal

Surprising result: The average win rate is nearly identical (%60 → %63). So on average, the system does not harden the game. But case by case, there are big swings:

• The too-easy ones got harder: Owlbear (%100→%69), Troll (%95→%69), Ghoul (%89→%64). These were monsters "too weak to be a solo boss" in raw SRD; the multi-phase structure made them a real threat.
• The too-hard ones got easier: Ogre (%16→%59), Red Dragon (%15→%55). These were "unfair" monsters in raw form that could drop the party in a single burst (breath/greatclub); the telegraph + phase structure made them fair.

So the tax runs both ways: it hardens the excessively easy, and softens the excessively harsh. The middle standardizes.

Tax Table 2 — Round Count (how much longer the fight got)

This is where the real tax is. Fights on average last roughly twice as long (4,4 → 8,2 rounds). This isn't a "difficulty" tax, it's a duration/experience tax: a boss fight is no longer a 5-minute volley, it's an 8-11 round, multi-act set-piece. This is a deliberate design choice — so the boss fight feels like a "moment to remember."

Note: Troll extended the least (1,1×) because its raw form was already long due to regeneration. Owlbear extended the most (2,8×) because its raw form ended in 2 rounds.

Tax Table 3 — Death Rate by the Four Tiers (100 − win rate)

Each tier's rate of killing the party, relative to the original (raw SRD):

Overall Tier Tax (relative to the raw SRD baseline)

Reading:

• Easy is even more forgiving than raw SRD (×0,6) — right for beginner tables.
• Normal is nearly on par with raw SRD (×0,9) — meaning the "standard boss" is exactly as hard as the original, but far more consistent and rich.
• Hard and Deadly are ~2× as deadly as the original — the real "hardening" happens in these two tiers, and it's deliberate there too.

Headline Finding — The Tax Isn't "Hardening," It's "Standardization"

Raw SRD bosses' win rate swings between %15 and %100 (a spread of 85 points). The same monsters' state under our Normal tier falls between %55-69 (a spread of only 14 points).

The system reduces boss-to-boss inconsistency by a factor of 6, while barely changing average difficulty (%60→%63). As a DM, you're no longer guessing "will this boss wipe the party in one hit, or die in 2 rounds?" — every Normal boss is a predictable challenge.

The true nature of the Tax:

1. Duration tax (~1,9× rounds): Boss fights get twice as long — deliberate, for the "set-piece" feel.
2. Consistency gain (spread 85→14): Predictability rose dramatically — this isn't a cost, it's a gain.
3. Average difficulty ~constant (Normal ≈ raw SRD): The standard boss is exactly as harsh as the original.
4. Tier flexibility (×0,6 – ×2,1): Raw SRD offered one difficulty; the system offers four.

Comparison with Solo-Adjusted SRD (State B)

Most experienced DMs strengthen a solo boss against action economy with ~1,6× damage. The result of this "house rule":

Solo-adjustment makes most bosses crushing (four bosses at %0-3). Because blindly multiplying raw damage by 1,6× without a fairness mechanism like telegraph/key produces a TPK. Our system meets the same "strengthen the solo boss" need without crushing — because it distributes difficulty into phase/key/telegraph structure instead of HP-inflation.

In other words: a DM's manual "solo boss strengthening" ends up either too weak (raw) or too harsh (solo-adjusted). Our system automatically finds the fair, consistent point between the two.

Boss-by-Boss Tax Profiles

Got harder (were "fake-easy" in raw SRD):

• Owlbear (%100→%69, death 31×): The biggest change. The raw owlbear was too simple to be a solo boss; the two-phase structure made it a real T2 threat.
• Troll (%95→%69): Regeneration was already strong even in raw form; the phase structure + Fire key turned it into a "find the key or it won't die" puzzle.
• Ghoul (%89→%64): The paralysis threat is more dangerous in multi-phase.

Got easier (were "unfairly harsh" in raw SRD):

• Red Dragon (%15→%55): The raw breath attack could drop the party in one hit; the telegraph + phase structure made it fair.
• Ogre (%16→%59): The raw greatclub burst + low HP was unbalanced; the structure made it reasonable.

Already balanced (small change):

• Marilith (%65→%65), Fire Giant (%59→%63): Their raw forms were already well-calibrated; the system only added phase/tier richness.

Assessment

How much harder did the system make the game?

Short answer: almost not at all at the Normal tier (average %60→%63). The real change isn't in difficulty, it's in three other dimensions:

1. Consistency (spread 85→14 points): Bosses are now predictable. The biggest gain.
2. Duration (~1,9× rounds): Fights are twice as long — a set-piece experience.
3. Flexibility (four tiers, ×0,6–×2,1): Four difficulties from one boss's data.

Is the Tax fair?

Yes. The word "Tax" is misleading — this isn't a cost, it's a transformation. At the same average difficulty, bosses are far more consistent, far richer (phases, keys, telegraph), and scalable across four tiers. The one real cost paid is duration (2× rounds), and that's a deliberate design choice.

The single most striking data point

Average win rate: raw SRD %60, our Normal %63 — statistically identical. The system didn't harden the game; it transformed the SRD's chaotic balance into a predictable, scalable structure at the same average harshness.

Methodology: Raw SRD states were simulated using actual stat block values (AC/HP/multiattack expected damage, including breath/recharge); the solo-adjustment was modeled with the common ~1,6× damage compensation. All three states were run against the same 6-party tier batteries, with preserved dice variance, 1,200-1,500 times per party. Death multiplier = (100−new win rate)/(100−old win rate). Absolute percentages are model-sensitive; what's reliable is the comparative differences (same engine, same parties). At extreme values (e.g., Owlbear 31×) the multiplier balloons from a low baseline (raw %1 death rate); the point difference is more readable. Tool files (engine.py, original.py, difficulty.py, bosses.py) in the output. This report is an extension of documents 10, 12-16.

License: Adapted monsters are based on SRD 5.2 (Wizards of the Coast LLC, CC-BY-4.0). This work includes material from the SRD 5.2 by Wizards of the Coast LLC (https://www.dndbeyond.com/srd), licensed under CC-BY-4.0.