Posts for year 2008 (old posts, page 14)

Understanding Duck Dealer

I’ve played Jeroen Doumen and Joris Wiersinga’s Duck Dealer twice. It is a tough game to grok.

Update: In fact the game is sufficiently difficult to grok that almost all the below is flat out wrong. It is hogwash. Beware. I’ll be writing up a correct summary of the game after it reaches broad release.

  • Duck Dealer is a perfect and certain information logistics game very much in the spirit of Roads & Boats and Neuland but with its own unique take
  • Don’t be deceived by the light and funny theme
  • Duck Dealer is a deceptively direct re-interpretation of Merchant of Venus
    • The theme, core nouns and board-patterns are similar
    • The play decisions and game character are violently different
      • Duck Dealer is a mentally demanding forward-looking planning-festival. Merchant of Venus is a much lighter and simpler affair of delivery optimisation with occasional opportunities
  • Three players is probably the sweet spot, maybe four with experienced players
  • The first phase of the game is ship improvements
  • The opportunity cost of ship improvement is usually prohibitive later in the game
  • Each pattern of ship investments establishes a natural rhythm for that ship; the rhythm at which it takes energy and then acts
  • The balance of colours on the ship likewise establishes a natural activity-type rhythm
  • None of the choices of combinations for game-start ship configuration are obviously bad but each one strongly suggests a different ship development tack and thus resulting rhythm
    • I suspect that the game’s initial turn order plus the ship configurations of the player’s ahead of you in turn order is a key factor to smart initial ship configuration decisions
    • I also suspect this reflects strongly into suggested ship configurations
      • It is clear that the random distribution of initial mines and the consumer tile ordering affects this and in fact defines these choices
      • It is not at all clear to what extent this is true in practice given human’s struggle with the decision space of the game
  • Coarsely (there are exceptions) the larger/more equipped a ship is the longer its rhythm
  • The heart of the game is managing your ship’s natural logistical rhythms against the emergent opportunity costs as the game/board develops and new buildings (tiles) are placed
  • It is viable to stay with an undeveloped 8-speed ship with two discs and a cargo hold
  • The undeveloped ship strategy is the most programmatic (fewest significant decisions), most risky and most subject to interference from other players
    • Undeveloped ships are dependent on rapid short-term opportunity exploitation which in turn requires frequent actions and results in a short/fast game
    • If the undeveloped ship dawdles the better equipped ships will out-pace them
    • If multiple players do run undeveloped ships they will drive a short/fast game that will likely seem to be decided arbitrarily
      • I find longer games with more developed ships more interesting
  • The more ship improvements the players do, the longer the game will be and the more of the board will be developed
  • Just one undeveloped ship moving fast and staying focussed can drive a fast/short game
  • The value of Build/yellow decreases during the game, sometimes precipitously.
  • Build/yellow can be worth a lot early in the game
  • Exception: Heavily built ships with two movement can reliably profit from Build/yellow energy to place Galactic Infrastructure (cubes on routes) to help manage their naturally short (2) movement distance
  • The bigger your ship is the more often Galactic Infrastructure (cubes on routes) is more useful than privilege markers on tiles
  • Teleports are remarkably hard to place usefully
  • Clever teleport placement may invert some of these observations
  • I have yet to see a clever teleport placement
  • With rare exception there are more VPs for building factories and flipping consumer tiles than doing milk runs stuffing goods into an already flipped consumer
  • Establishing and then running a production loop is almost invariably less profitable than focussing on building factories and flipping consumer tiles
    • This is directly opposite from Merchant of Venus’ reward patterns
  • Exception: Ships with huge hold-spaces (7+ capacity?) can profit enormously from a well-optimised production loop (ie privileges claimed on all steps) on a 25/10 consumer.
  • The usual reason players act is that they become unable to sustain comprehension of their intended move. Their mental stack overflows, they lose track and so they act instead of continuing to search for the optimal risk/value return for energy accumulation. This is often (almost always) a mistake.
  • Usually getting more energy and doing more later is markedly more efficiently profitable
    • Unless someone gets there first
  • The game is all about acceleration curves against game length
  • Each player’s ship investment has a pay-off curve
    • A undeveloped ship is faster to build, runs more quickly but makes fewer points per natural action
    • A big ship takes longer to build and runs more slowly but makes more points per natural action
  • The players emergently determine the game length
    • Which player’s ship’s reward-curve/opportunity-exploitation-rate/game-length sums best when the game ends is the core question of the game

Reposed mumbling recovery from a shower

The recent play of Muck & Brass has been occupying my sleeping nights and bedraggled shower time. I’ve no great conclusions other than that 5-days-for-everyone is a Bad Idea (see below). The rest is just terrain mapping.

The standard action pattern in the recent game was a spate of Expands followed by a mix of Develops and Capitalises. Typically the final actions were chosen to either force the round closed, or to put the next player in a position to force the round closed. The only times Capitalise was selected early was when either all of a player’s companies were broke or in the late game in order to float more secondary companies in order to extend game length (it was set to end in the sixth General Dividend with only two companies left operating but we called the game early mid the 5th round due to other constraints).

Unlike Wabash Cannonball early capitalisations were not seen as critical early moves, in part due to the high opportunity cost of the Capitalise action (time), but also because there are simply so many shares available (10 unsold shares) and relatively little to differentiate among them. The result is that Muck & Brass is mostly not about temporary emergent alliances but about the old standards of arbitrage, opportunity and tactical position.

In a four player game the dance is straightforward:

  1. Stay low/early in the turn order to get more Expands for your companies
  2. Buy lots of shares (difficult when you’re low in the turn order)
  3. Buy shares that players lower than you in the turn order also have so that you profit from their activities
  4. Try to keep your own shares undiluted

The lack of a Chicago-style golden target to create a discontinuous value set is the main characteristic that precludes temporary emergent alliances. What alliances there are, are happenstance collisions of mutual interest rather than deliberately managed affairs. As such the player relationships are much closer to Preußische Ostbahn’s management of temptation and exploitation than Wabash Cannonball’s forthrightly collusive work-together-and-we-both-make-more.

In our game, because of the high opportunity cost for capitalisation, Capitalise was usually used as a round ender to re-invigorate a cash-less company or to setup turn order for the next round. Thus the first round looked something like:

  • (E7/D5/C3)
  • P1-E2/2 P2/E2/2 P3-E2/2 P4-D1/2 P4-E2/2 (E3/D5/C3)
  • P1-E2/4 P2-E2/4 P3-E2/4 P4-C3/5 (E0/D5/C2)
  • P1-C3/7 P2-C3/7 (E0/D5/C0)

The second round was a little more interesting as London started to be developed, thus freeing up a host of other possible Develops for other companies:

  • (E7/D5/C3)
  • P1-E2/2 P2/E2/2 P3-E2/2 P4-D1/2 P4-D1/1 P4-E2/3 (E3/D4/C3)
  • P1-E2/4 P2-E2/4 P3-E2/4 P4-C3/6 (E0/D4/C2)
  • P1-D1/5 P2-D1/5 P3-C3/7 (E0/D2/C1)
  • P1-P1-D1/6 P2-D1/6 (E0/D0/C1)

The third round was similar, but with even less Capitalisation:

  • (E7/D5/C3)
  • P1-E2/2 P2/E2/2 P3-E2/2 P4-D1/2 P4-D1/1 P4-E2/3 (E3/D4/C3)
  • P1-E2/4 P2-E2/4 P3-D1/3 (E0/D3/C3)
  • P3-E2/5 P4-C3/6 (E0/D3/C2)
  • P1-D1/5 P2-D1/5 (E0/D1/C2)
  • P1-D1/6 (E0/D0/C2)

Such a lower rate of capitalisation encourages companies to run with broke treasuries, for positions to be more heavily invested, and for strong player differentiation, both positional and financial (low cash high income versus lower income and high cash). It also encourages players to forgo Capitalise and instead Develop heavily as it has similar effects on income (and slows the game down). From a design perspective these are all desirable patterns, not least the slowing of the game.

The players clearly valued Expand higher than the other actions. I’m not convinced they were wrong. Getting to London is critical for most companies and building out of London not only offers a larger income delta than any other activity (once London been developed once), but also positions for ready mergers with the three companies surrounding London. In our game the B&GR built 5 connections out of London. When London was developed up to $9, that was $45 income for the B&GR: $14/share just for London alone!

Consider a hypothetical 4th round of a game:

  • (E7/D5/C3)
  • P1-E2/2 (port/merger) P1-C3/5 P2-E2/2 (port/merger) P2-C3/5 P3-E2/2 P4-E2/2 (E3/D5/C3)
  • P3-E2/4 P4-D1/3 (E2/D4/C3)
  • P4-D1/41 (E2/D3/C3)
  • P3-E2/6 (float or merger) P3-C3/9 P4-E2/6 (float/merger) P4-C3/9 (E0/D3/C3)
  • General Dividend (two players past 6 days)

Four secondary shares were sold. There is potentially only one company left in the mix. The game could have just ended. Or, more likely there are less ports and mergers as there’s just not enough money in the other companies/players for the Expands required for ports and mergers. The players early in turn order are likely to have shares in companies with cash for ports and mergers:

  • (E7/D5/C3)
  • P1-E2/2 (port/merger) P1-C3/5 P2-E2/2 (port/merger) P2-C3/5 P3-E2/2 P4-E2/2 (E3/D5/C3)
  • P3-D1/3 P4-E2/4 (E2/D4/C3)
  • P3-D1/4 (E2/D3/C3)
  • P3-D1/5 P4-D1/5 (E2/D1/C3)
  • P3-D1/6 P4-C3/8 (E1/D0/C2)

Two secondary shares just sold and one arbitrary share was capitalised. Assuming that mergers were done instead of port builds there could be as few as 3 companies left in the game. Assuming instead that P4 had a share of a company that could merge with a single expand the last actions set could have been:

  • P3-D1/6 P4-E2/7 (merge) P4-C3/10 (E0/D0/C3)

In which case there could potentially be only two companies and the game would be over depending on what share P4 force capitalised. P4 has three choices:

  1. Capitalise the first share of any secondary company. This clearly makes sense if P4 will win as the game ends
  2. Capitalise the second share of a secondary company which is already connected to a company in which P4 is invested. This assumes that one of the earlier players capitalised the first share of that company. In practice this usually means either the NER, LNWR or GWR and only makes sense (again) if P4 is will win as the game ends
  3. Capitalise the second share of an unconnected company. This will put three companies back in the game causing the game to not (yet) end.

The choice-set is interesting because P1 and P2 control whether #2 is even present as a choice, and which company fits #3 (likely one near their investments). Neat!

More immediately useful is that without the forced 3 days for the merger there’s a reasonable potential for a merger storm in the fourth round that instantly ends the game. The real problem is that it is absurdly difficult to control or predict the values generated by such a merger storm from the preceding rounds making the game results essentially a crap-shoot. Scratch that idea!

The other perspective is that ports effectively offer several values to minor investors:

  • Cause a special dividend
  • Impoverish a company such that it can’t afford to drive future mergers
  • Ability to capitalise a share that encourages a merger-via-Capitalisation along with its related Special Dividend (ie a connected secondary company)

Mergers deliver all the above plus the following:

  • Accelerates game-end condition
  • Builds a combined treasuring which is likely large enough to afford another merger or port

Mergers are generally better for major investors as they provide two forums for special dividends for their heavily invested shares. Ports conversely offer only one future venue for additional secondary dividends but protect other investment’s special dividend capacity from predation by a steamrollering behemoth.

The other advantage for forced 3 day capitalisations is that players are now encouraged to capitalise more heavily in the early rounds in order to get enough cash into their companies to afford special dividends be they from ports or mergers. However this early capitalisation rush exposes another phantom that pushes back on such heavily capitalisation. Imagine the following perhaps not-quite-so-fantastic evolution:

  • P1 is invested in the L&MR (possibly even a minor investor), Expands to a port in Liverpool and force-capitalises the NER
  • P2 is invested in the L&SR (possibly also minor), can’t afford a nearby port and so merges with the EUR or LB&SCR for a special dividend. P2 then force-capitalises the NER to make a L&MR/L&SR/EUR/NER merger for yet another special dividend.
  • P3 is desperately behind and merges the LB&SCR or EUR into the new huge agglomerate, thereby removing the merger possibility from the agglomerate and securing their share’s dividends. P3 Capitalises something. Anything.
  • P4 has a potentially remarkably ugly decision as there are only two companies left in the game, the huge agglomerate and the B&GR.

That’s two actions spent for 3 special dividends. Oof. What’s different is that huge gobs of money have just been thrown into the mix by the three mergers. P4 is really wishing he owned a good spread of the northern companies at this point. Double oof.


  1. By exploiting time-position P4 gets in two Develops 

Ploughing submarines

The new map fits on 18”x12” paper easily enough with broad empty swathes down the edges. It is a bit tetchy at that scale. A few of the short links such as York/Selby, Preston/Blackoool and Dover/Canterbury get to be awfully short – going smaller would present significant clarity problems. For the Europeans printing on A3 is the best bet despite being a smidge shorter than the not-quite Super-B I’m working with here. (I’d test that conjecture here if I could find a reasonable supply of A3 or Super-B paper about San Jose).

map-8

I’ve updated (increased) the port costs back to within reason. If ports are too cheap then they are used exclusively rather than mergers. As the goal of ports is to a provide way for minor shareholders to drain gobs of cash from companies and to thereby ensure that they are not able to cause mergers, the cost of ports has to be high enough to provide the needed drain and yet low enough to prevent casual use. (ie more than $30 but not too much more) Meanwhile other players push for mergers as they solidify share positions and aggregate treasuries for further dividending activities.

Unfortunately the empty spaces along the edges of the new map aren’t large enough to fit the income track or the bank pool spaces. Instead I’ve made a separate tracks page that will fit comfortably on letter or A4 for those details:

ScoreTrack

And of course some rules polishing. The changelog:

  • Ports and mergers cause a (forced) capitalisation
  • Forced capitalisations cost 3 days! (This is big)
  • Added rules section for forced capitlisation
  • Sized bank described as ~$4K
  • Pumped port values most of the way back up
  • Removed several ports
  • Clarified glossary
  • Clarified double build language
  • Development rule WRT London clarified to account for Liverpool and York.
  • Added scoretrack page.

The forced capitalisations for ports and mergers is a huge question. There are good reasons to think it is a great idea, as a control on late game velocity, but also good reasons to fear it slows down the effective rate of capitalisations in the late game which in turn delays the rate at which secondary companies enter the game and that is a problem. I’m finding modelling this problem fiendishly difficult; it is so sensitive to exact player positions and incentives that I’m not yet able to predict broad pattern behaviours.

In partial response I’m considering changing the round-end determinant yet again. Currently it is two exhausted actions or two players at or past 6 days. While that’s a nice pattern it is also a wee bit slow. The new thought toy is to end the round on two exhausted actions or when all players have used at least 5 days. In a 6 player game that ensures that all rounds will end on two exhausted actions. With lower player counts it will often be all players at 5 days. Hmmmm.

New rules.

Peen hammer toccata

Changelog:

  • Reduced port fees (perhaps too much)
  • Auctions clockwise from capitalising player (faster and simpler)
  • Clarified multiple home stations for merged companies
  • Fixed action supply length and colours on map
  • Correction: Capitalise may be chosen for any available company or one of their own shares
  • Clarified that non-floated company treasuries are discarded when scoring
  • Clarified that merger-forced capitalisations take 3 days (Okay, added)
  • Clarified that merger-auctioneers are not required to bid
  • Fixed language about summing incomes for auto-merges

New rules.

Smelting for copper

Four of us had a go at Muck & Brass with the new action system this afternoon. In short, it was wonderful, really wonderful and far less unstable than the earlier version. Early conclusions:

  • It was more intuitive to describe the action system in terms of days than points. Actions take days (time) and the players take turns in the order in which they are available
  • Ending on 7 days or two exhausted actions is too rich. It wasn’t terrible but we always ended on actions rather than points. 5 days may be too short but it is clearly either 5 or 6 days
  • The secondary companies are odd. Simply, odd. It is tempting to have them float on one share, but that feels too large a change for their modest oddity. I’m not sure they need to be addressed but it is tempting
  • The Foreign ports are probably over-priced by ~30%
  • 3 Capitalisations, 5 Develops and 7 Expansions seems right
  • The requirement for London to be the most valuable city forces the early game into an expansion frenzy. This is good, but I’d forgotten it in my earlier thought cases.
  • One player strongly disliked the action/time track being way from the geographic map. That’s easy enough to fix.

In all, an excellent outing.

Variant: Draughting Emperor’s Reward Cards in Confucius version 2

Building on the previously discussed variant Ben Keightley proposed a twist on #bgdf_chat:

[2008-11-07/14:02] <cocadieta> JC, you around? I had a flash of inspiration this afternoon re: Confucius.

[2008-11-07/14:03] <cocadieta> The bonus card drafting variant (unplayed!) sounds good but really pushes the main problem further up the pipe–say two people are in line to run ships out to sea, the second person could either luck out or get hosed by the card draw.

[2008-11-07/14:04] <clearclaw> That’s why the variant shows the pipeline of cards as well as the current draft set.

[2008-11-07/14:05] <cocadieta> Idea: At the beginning of the game, shuffle the bonus cards and deal five to each side of the board. The left side are the bonus cards available to military conquests, the right side to boats.

[2008-11-07/14:05] <clearclaw> Oh cute!

[2008-11-07/14:07] <clearclaw> So there’s the possibility of the 6th fleet getting stuck.

[2008-11-07/14:07] <clearclaw> There are only 3 possible army cards

[2008-11-07/14:07] <cocadieta> Nahh, if the stock of 5 runs out, players can pull from the other side.

[2008-11-07/14:09] <clearclaw> So players pull any of the face up 5?

[2008-11-07/14:09] <cocadieta> Exactly.

[2008-11-07/14:09] * clearclaw ponders.

[2008-11-07/14:09] <clearclaw> That makes ER cards FAR more valuable

[2008-11-07/14:10] <cocadieta> They’re already that valuable, just at random times.

[2008-11-07/14:11] <clearclaw> But now it is perfectly controlled and can be timed to a nicety.

[2008-11-07/14:11] <clearclaw> In particular it makes the bribe cards even stronger

[2008-11-07/14:11] <cocadieta> A game where all of the bribe cards got dealt to the left of the board would be much different than a game where they were all dealt to the right. Definitely.

[2008-11-07/14:12] <clearclaw> Not just that, but the certainty that you could get the bribe card you wanted exactly when you wanted it

[2008-11-07/14:17] <cocadieta> The bribe cards are already the most valuable cards in the game. It seems inappropriate to hide them behind random draws of any kind.

[2008-11-07/14:18] <clearclaw> There’s some truth there.

I like this change.

Share lessons

I made paper shares for Muck & Brass tonight using the share PDFs I genned yesterday. I don’t need them, the glass bits I have been using work well enough, but I suspect that a slightly crisper presentation (and demonstration of personal investment?) will make acceptance of a very early stage prototype easier.

A quick trip to an office supply store produced a pad of craft paper in 8 colours (sadly one of the colours was a very dark purple). A couple aisles over revealed two packs of copier paper in 5 colours, one set rather pastel and the other day-glo. I grabbed the craft paper and pastel sets, printed four sheets of each share sheet, cut them quickly with a Fiskar’s rotary cutter and slipped them into penny sleeves. 390 penny sleeves no less. However, they’re really quite nice. The paper colours are bold and clear, the black printing shows well, the company abbreviations on the ends of the shares allows them to be fanned easily (as often done in the 18XX) etc. Fair dinkum.

Meanwhile yesterday’s trip to the paper supply store revealed a business card cutter: feed printed stock in one end, turn the handle and out fall cut business cards. Cost ~$150. I suspect card production is not as trivial as it suggests and that a vice-based paper cutter remains the better choice there. Lastly, while business cards could make good shares/cards but there are narrow limitations on available stock colours and relatively high media costs. Bah.

The paper supply store also had a manual corner rounder for $300 (essentially a jig with a vice and a curved blade). It was a rather heavy-duty piece of equipment; likely to survive WWIII unscathed. it was tempting to be tempted. I have a hard time resisting such well-formed mechanical engineering. Thank the gods nobody is waving a Curta calculator in my face.

Lessons from tonight’s production run:

  • Colour copier paper in a wide range of colours (I need 10 colours for the 10 companies in Muck & Brass) is readily available outside my local office supply stores (Stapes, Office Max, Office Depot etc). eg this store
  • Sleeves are constant size. This can hide a multitude of sins surrounding cut shares not being of constant size.
  • Craft paper is nice stuff to work with
  • Craft paper easily jams in the printer. Treat it carefully and ensure the guides and feed are aligned well!
  • Coloured copier paper is not so nice to work with, tending to crease, crumple and stick
  • With a little care the rotary cutter will handle and cut stacks of 8 sheets well enough. That makes for 8 cuts for 64 shares. Not bad.

The goal is to play tomorrow at the Endgame Anniversary party. It isn’t likely, but worth a shot.

Falling practice hammers

Being in possession of a new printer, an HP K8600 whose main claim to fame is that it is can up to print A31 (roughly 13”x19” – useful for prototype maps/boards), I spent a little time this evening messing with another new tool: Scribus. Scribus is a page layout tool akin to Adobe’s Framemaker. There’s little Scribus can do that I could not do and likely do better with my long standing favourites of LaTeX and LyX, but for quick, experimental and mostly throw-away tasks Scribus has a lower barrier to entry. The fact that it also emits SVG and PDF is icing on the cake.

My first test project was to draw shares for the 10 companies in Muck & Brass (EUR, B&GR, L&SR, L&MR, LB&SCR, CR, GWR, LNWR, NER, SWR). The assumption is that each sheet would be printed 4 times, each company on a different colour of paper, and then one of the non-merger shares for each company would be discarded. The final shares would be slipped into penny sleeves, perhaps with a card backing. This would give 3 non-merger shares and 28 merger shares per company, which is just barely enough.

Winsome Games uses rather nice coloured paper in a wide range of colours for many of their shares. I’ve not seen anything quite so pleasant at my local office supply stores. I wonder what I’m missing?2

EUR-shares-page1

As Ben Keightley graciously reminded me on #bgdf_chat, it would be better to put small images showing the position of that company’s home city on each share. I plead gross laziness and a knowledge of English geography! I can’t quite be bothered to gen 10 maps of England with highlighted cities, not when the above sheets merely required producing one master and doing search-and-replace with XEmacs to produce the others.


  1. Would that I could find a local paper supplier that carried A3! 

  2. I did find very polystyrene clamshell cases at a local store, similar to the ones that Winsome Games uses but rather larger (~4” deep). Thought provoking.