Monday, September 25, 2017

Castle Construction Costs

One of my Top 5 major adjustments to D&D is in using a "silver standard" for the economy (see the sidebar). That is: I simply read all the costs, nominally in gold pieces, as silver pieces instead. This is particularly easy to do in OD&D where all costs are just in a single unit.

One major reason for this is that it generates a more realistic set of prices that given in D&D in terms of gold pieces. We get a better simulation. When we want to expand our gear selection, we can just look up real-world historical information, and use that with some confidence. It avoids the awkward rationale in AD&D that all of the campaign area is suffering a hyperinflation economy. Plus, it effectively increases the coin-value that adventurers can haul out of a dungeon. At some point they can level-up to gold and carry out increased density in value and XP. So it's also good gaming design.

As related previously, I usually think of the D&D copper, silver, and gold pieces as equating to English pence, groats, and nobles, respectively (link1, link2). Note that groats are one-third of a shilling, and nobles are one-third of a pound in worth (recall that shillings and pounds were not ever medieval coins). So this means that if I do find a real-world price list in shillings, then I need to multiply the shillings by 3 to get actual silver pieces (i.e., groat coins). In other words, 1 noble = 20 groats, and 1 groat = 4 pence (note that copper pieces still have reasonably legitimate worth by using the real-world coins, much as in OD&D; they are not entirely rubbish as in AD&D).

On to the main point here: Let's see what kind of estimated conversions we can male, based on what we see in the OD&D books (Vol-3, p. 20, 5th print), versus some real-world data on castle construction costs. Fortunately Wikipedia has some information and links under its Castle: Construction article on different types of castles. I always count Wikipedia as good enough for gaming inspiration; although you can follow up on the citations to these data there if you wish, of course.
Even a very small tower, such as Peveril Castle, would have cost around £200. In the middle were castles such as Orford, which was built in the late 12th century for £1,400, and at the upper end were those such as Dover, which cost about £7,000 between 1181 and 1191.[131] Spending on the scale of the vast castles such as Château Gaillard (an estimated £15,000 to £20,000 between 1196 and 1198) was easily supported by The Crown, but for lords of smaller areas, castle building was a very serious and costly undertaking. It was usual for a stone castle to take the best part of a decade to finish.

I followed up on the links to articles on each of those castles, documents sizes and construction details, did some Googling for extra maps of the castles, roughly estimated measurements for wall lengths where I could, etc., then tallied up costs as indicated in OD&D Vol-3. For example: Peveril Castle is simply a single keep, 40' square, and 50' tall (or at least that's the only part that we have construction costs data for). Looking in Vol-3, I compared this to the 40' round tower, 40' tall; as per the adjustment note, to add 10' height costs +20%, giving a D&D cost of: 10,000 × 1.2 = 12,000. And so on for the other, more complicated castles.

Of course: In D&D this is officially in gold pieces and vastly different from the 200 real pounds indicated in the article. So as an experiment, let's see what happens when we convert the real price in pounds to silver groats (i.e., multiply by 60):


Conclusion: I'd say it's completely uncanny how well the D&D costs correlate to real-world costs, assuming we express them in silver groat coinage. For example, the cost for Peveril castle is exactly the same as our estimate from the D&D rulebook above. The cost for Orford castle is off by only about 1%. My estimated cost for the enormous Dover castle is at least on the correct order of magnitude. Initially, I did expect such a close correlation!

What can account for this? Are the castle prices in OD&D invented from whole cloth, and just accidentally match the real-world cost in silver (and so too all the other prices for almost everything)? Is there some prior work based on historical research that expressed prices in silver, and was converted by fiat to "gold pieces" for D&D? At any rate, it rather stunningly gives added confidence to the very handy idea of reading prices in D&D in silver (groats) instead of gold.

A few other points: Note that time (in years) does not scale linearly with cost. Based on the three data points here, it's actually more exponential in nature (R² = 0.95). The other thing is that I left out the reference to Château Gaillard in France; while described as "vast" and supposedly costing between 15,000  and 20,000 pounds, by my eye it is actually smaller than Dover Castle (although in a more precarious location), and took a lightning-fast 2 years to build (perhaps due to personal, eager oversight by Richard the Lionheart). So that doesn't correlate in any way with the prior data points.

In summary: Feel free to use the Original D&D prices as pretty good estimates for real-world costs, as long you read them in units of silver pieces (groats). Thoughts?

21 comments:

  1. Assuming Perivile is 60,000 cubic feet of stone (ten feet thick shell wall). The cost of mining anything is an ounce of gold per ton and 76 ton per thousand cubic feet. Thats 4,560 ounces of gold (285 lb of gold) or 2,850gp. Thats the cost to quarry. 4x that is 11,400gp which should cover transport and construction.

    ReplyDelete
  2. If silver is better than gold, can you give us your opinion on going to a copper standard?

    I'm less interested in historical accuracy than I am with the coin as an artifact of a game.

    My thought is 1 cp == 1 USD. Then two things happen. First, gold becomes valuable in the mid and high levels again, and it's very easy to estimate and then gauge prices, since we all know what a dollar is worth (you can say 1 cp == £1 or whatever you like, it's almost irrelevant what the real world unit is.)

    ReplyDelete
    Replies
    1. I can actually see that as a game-balance design (perhaps more so than gold). However, I'm always hoping to succeed at both the game-balance and world-simulation challenges, so personally I wouldn't consider it on those grounds.

      Delete
    2. And as I recall, that's just what Gygax did in his later Dangerous Journeys/Mythus system (called it a "basic unit of currency" or a "BUC" to get the idea across).

      Delete
  3. Skerples from Coins and Scrolls made a system for build castles. He even used the Moathouse and the proper temple of the Temple of Elemental Evil as example.

    https://coinsandscrolls.blogspot.com.br/2017/09/osr-building-castles.html

    https://coinsandscrolls.blogspot.com.br/2017/09/osr-castle-of-elemental-evil.html

    ReplyDelete
  4. Also, I have two books by Lisa Steele (Fief and Town) which have some building prices listed.

    Cà d'Oro in Venice is estimated to have cost 23,000 lires around 1450 (1 lire is 20 soldi and, 5 Venetian soldi is 1 London pence, using a carpenter wage as cross-reference for soldi/pence conversion.)

    The rebuilding of walls of Farnham castle cost 97l 15s 5d over the course of 3 years (year not provided).
    More from Farnham castle in 1302: Pulling down the ruined tower cost 8s 6d; Stones for new tower cost 6l; Lime cementer for tower 47s 4d; Carting the stones for tower cost 6l 16s 2d.

    ReplyDelete
    Replies
    1. Those books sound very cool. I just put them on my personal wishlist, thanks for that!

      Delete
  5. Many of the OD&D costs appear to be based on "real world" records, or so I found when I was researching medieval costs for a game of my own design. I think the original reason for some of the exorbitant costs (1 gold piece for six torches, for example) was just a matter of streamlining for game purposes; i.e. let's get started playing and not worry about "making change" when equipping our characters (remember that "town" wasn't really part of the game in its origins, except as an abstract concept of "home base").

    I'd guess the "gold standard" (as opposed to silver) was simply due to it being more fun to find chests of gold as treasure.
    ; )

    ReplyDelete
    Replies
    1. Well, yeah, but one of the advantages of going to a silver standard is that gold actually BECOMES treasure, as opposed to mostly being an impediment that will weigh you down in your search for gems and jewelry (which make up 92% of all treasure value - or whatever that silly number is). :)

      Delete
    2. I agree with the "seems fun" thesis (which is usually a deadly design criteria, IMO). Jon Peterson also writes of a tradition in prior wargaming/literature of emphasis on the "gold pierces" end of things for fantastic treasure descriptions.

      Delete
  6. The sixty multiplier is completely arbitrary even if it is pleasing that it fits your monetary division. There are divisors aplenty, and coin, to fit any particular table. It is more significant that Gygax's ratios hold.

    Have you determined, how many men-at-arms are required to service a castle of various sizes, that at least would be interesting.

    ReplyDelete
    Replies
    1. Fief, page 78: "In war, the garrison doubled or tripled so that it could protect the fortification and harass opposing armies, foraging parties, and scouts. A major royal castle might have a wartime garrison of 500 soldiers and 150 knights."

      I think that the smallest castles (like keeps) would have only one knight, his family and as much people as he could afford during peace (usually only servants and no men-at-arms).

      Delete
    2. Quite so. A force of 650 would be a lot even in a big castle like Dover. The smallest castles, such as the fortified manors in Scotland, might have only a country gentleman on horseback and his son as squire.

      Delete
    3. Yeah... I was being polite saying the "don quixote" guy was a knight and his "pancho" son a squire. Heheh

      Delete
    4. The 60 divisor wasn't arbitrary. A statistical correlation between real-world price lists and D&D prices shows that a shillings-to-pieces conversion of 3 is appropriate (and also matches the common "groat" coinage). A factor of 20 is of course the conversion from real-world silver to gold. 3 × 20 = 60.

      Delete
  7. BLUF -- I think you can estimate construction time with this simple formula:
    COST IN SP / 2 = MAN-WEEKS REQUIRED

    That does not capture the non-linear nature of the project. To do that you need a more complicated rule... something like...
    FIRST 10K IN SP (0-10K): 1 MAN WEEK = 4 SP OF WORK
    NEXT 10K IN SP (11K-20K): 1 MAN WEEK = 3 SP OF WORK
    NEXT 10K IN SP (21K-30K): 1 MAN WEEK = 2 SP OF WORK
    BEYOND (31K+): 1 MAN WEEK = 1 TO 2 SP OF WORK

    Another way you could do it if you prefer to roll dice is to say that each man-week of labor generates 1d3 SP of work, +1 for small projects, -1 for large projects.

    Then you just chuck a fistful of dice to determine progress each week.

    \\\RESEARCH FOLLOWS\\

    On this subject, there is a fascinating source to estimate construction times here:
    https://www.guedelon.fr/en/order-of-construction_86.html

    I estimate this project's available man-power as about 3K man-weeks per year (2600 full-time skilled laborers, the rest transient visitors).

    To build about 600 feet of curtain wall to 10' (a 23K SP project) it took the team four years... They were doing a few other things too, so this is 6-12K man-weeks.
    - The ratio of SP to man-weeks is between 4:1 and 2:1.

    To build a great hall quite similar to the 6000 SP building took about two years, again with other projects going on, so 3-6K man-weeks.
    - The ratio of SP to man-weeks is between 4:1 and 2:1.

    The "western tower," similar to a 5K SP round tower, took work over parts of three years. Call it about 3K man-weeks.
    - The ratio of SP to man-weeks is about 1.6:1.

    All of this leads to an interesting observation: As an order of magnitude estimate, construction time can be estimated as: COST IN SP / 2 = MAN-WEEKS REQ'D.

    Let's compare that to some other castles in the real world. We have to make some manpower assumptions.

    Large castle (Beaumaris) (2.75 SP to 1 man-week): 2600 laborers on site each summer, dropping off each year, for 5 years, to build 11K pounds (660K SP in D&D terms). So let's call it 8 months of labor each year, with an average of 1500 laborers for five years -- that's 240K man-weeks. Here we get a figure of one man-week for every 2.75 SP in spending.

    - English Burgh Hide System (2 SP to 1 man-week): Every hide provides a man for wall defense and fortress work. Four men are needed for every 16.5' section (rod). Thus a D&D 90' wall section needs 21 hides. These men presumably work part time, as needed to defend against vikings -- call it six weeks per year. That's 126 man-weeks. Most of our medieval castle costs indicate annual MX costs run about 5-10% of new construction... A new 3500 SP D&D wall would thus cost 175-350 SP of MX. Again we get a variable of about 1 man-week for every 2 SP of work.


    - Oxford Castle (1.6 SP to 1 man week): Domesday book records 193 households ("hides") for Oxford and the nearest surrounding villages. This is an 84K SP D&D castle, and it took 8 years to build. If we assume each household works six weeks per year, that's 1.2K man-weeks per year, or 10K man-weeks over eight years... Based on Beaumont, we know that a typical ratio of unskilled labor to paid skilled labor would be about 2:1, and the skilled labor likely works all year. So that's another 100-ish laborers for 52 man-weeks, 41K man-weeks over eight years. Put it all together and you have: 84K SP and 51K man-weeks. Awfully darn close to 2 SP per man-week.

    - Peveril Castle (4 SP to 1 man week): Domesday Book records around 41 households near modern day Bamford. Using the same math as above, we have 246 man-weeks of part-time corvee labor each year, and another 1040 man-weeks of full time labor each year for a total of ~1.3K man-weeks each year -- 2.6K total. The D&D castle costs 12K SP. Our ratio is 4 SP per man-week. Within an order of magnitude.

    ReplyDelete
    Replies
    1. Sorry to double tap this -- thought of one more dice mechanic to use on this as well.

      I also think the slower construction time on larger castles is largely due to diminishing returns from larger crews. As an example, the small, cohesive french team that uses relatively little unskilled guest labor achieves productivity as high as 4 SP per man-week. The medieval castles using large amounts of corvee labor are more like 1.5 to 2.5 SP of productivity per man-week. It makes sense that a large draft crew (1-2K laborers!) will have lower productivity than a highly skilled small crew.

      So with those considerations... Some rules.

      CASTLE COSTS
      Pounds and shillings are used to calculate construction time as a useful measure of value for large projects.

      60 OED or OD&D Castle Guide SP = 1 pound.
      3 OED or OD&D Castle Guide SP = 1 shilling.

      SYSTEM #1: PRECISE, MATHEMATICAL TRACKING
      BASE ASSUMPTION: 1 man-week of labor with a normal crew (<100 workers) results in 1 shilling of work being done.

      Thus:
      - 4 man-weeks of labor results in 4 shillings of work being completed.
      - 80 man-weeks of labor results in 4 pounds of work being completed.

      Some unskilled labor is available to Lords for free (40 days per year per household), but otherwise must be hired out. Skilled laborers must be provided; 1 mason & 1 carpenter per 5 unskilled laborers, 1 smith per 50 workers.

      Modify productivity for 4-man week or 80-man week units as follows based on the size of the crew:
      (+2 pound) One Man Act (1 worker)
      (+1 pound) Small Crew (<=10 workers)
      (+0 pound) Normal Crew (<=100 workers)
      (-1 pound) Large Crew (<=1000 workers)
      (-2 pound) Enormous Crew (<=10,000 workers)
      (-3 pound) Legendary Crew (<=100,000 workers)
      (-1 pound) Insufficient skilled labor available.
      (-1 pound) First month on the job

      Fantastic creatures and heroes provide one man's worth of labor for each hit die, but count only as one worker. Thus a crew of giants or highly skilled dwarves are quite valuable!

      Example: A small 200 pound castle is designed. To build this castle, a crew of 500 workers is brought in. Each week for the first month of work, 12.5 pounds of work are completed [500 / 80 * (4-2)]. Thus 50 pounds or 25% of the castle are completed in the first month. Work picks up and after the first month, 19 pounds are completed each week [500 / 80 * (4-1)]. At this rate it will take another eight weeks to finish the castle for a total time of three months.

      Now, of course, the local labor force owes the lord but 40 days of service each year (6 weeks), so construction may be drawn out over two or three years, unless of course the lord is willing to pay wages for unskilled laborers to stay on site.

      Delete
    2. SYSTEM #2: ROLL SOME DICE!
      ASSUMPTIONS: As above in System #1.
      NOTE: System #2 is slightly more favorable to the builder, resulting in about 6.4% faster construction. It is however more variable.

      For every 25 man-weeks of labor, roll 1d6. A roll of 3-6/6 indicates that 2 pounds of construction is completed; failure indicates that work fails to progress. Modify the die roll with the modifiers for crew size, etc above, thus a "large" crew succeeds 4-6/6.

      For very large or small projects, simply multiple or divide the above man-hour and pound progress figures keeping the ratios similar, i.e. 250 man-weeks of labor has a chance for 20 pounds of construction progress... 5 man-weeks a chance for 8 shillings.

      Example: A 200 pound castle is designed. To build this castle, a crew of 500 workers is brought in. The DM decides to use the 250 man-week/20 pounds ratio. Thus each week, 2d6 are rolled -- one for each half of the work force.

      Due the the size and inexperience of the workforce, the odds of production are only 5-6/6 for the first month. In the first month, dice rolled result in 2, 2, 3, 1, 5, 6, 5, 1. Three successes! 60 pounds of work are complete and the castle is nearly a third done.

      After the first month, production increases and success is indicated by 4-6. Another seven successes are needed. It will probably take the crew another two or three months to finish. Regrettably, the corvee work force will return to the fields after just another two weeks... Work will likely have to resume next year.

      Delete
    3. That's some excellent work, thanks for sharing that!

      Delete