The conversation was also an end in itself. I think both Paul and I always knew the odds were against Alexander ever seeing the light of day, but in true Ponarvian spirit, we attacked the project with a highly professional dedication and sense of purpose. The secret truth is that Alexander was more fun to design than it ever would have been to play, and for years we had a blast doing pure design. Our detailed dialog, often supplemented by charts, screenshots, and prototypes, remains as a fine case study of the creative process.

Most of that conversation is preserved in the following collection, gathered from a variety of sources. Note that I usually signed my name as "Epicurious J.", my pen name from a column Paul and I co-authored in our junior high school paper. Paul sometimes refers to himself by his pen name, "Smendrick T.", but during this period more often prefers the name "P. King Duk", or just "Duk".

Paul 1989-02-27	Paul opens the dialog with a list of possible improvements to Strategic Conquest. New pieces, reporting options, and startup options.
John 1989-03-08	World generator and AI opponent. Multiplayer games, customized production, enemy strategy, path tracking, names, and textual records.
John 1989-07-09	First priority: generating a random world map
Paul 1989-08-07	No to swamps and mountains. Still need name for the project.
John 1989-08-25	Capital city, production schedules, tracking enemy activity, first mention of a command language
Paul 1989-09-15	Piece characteristics matrix. Charactistics vary by city. Preliminary syntax for command language.
John 1989-10-09	First edition of the Konquest Journal
Paul 1989-10-12	Names, weather, and more command language ideas
John 1989-11-18	Naming scheme. First reference to a spy. Events and sample scripts for the command language. Continent generation.
Paul 1989-12-21	Data Dictionary. Air Transport piece. Multi-player games approved. Startup options. More work on continent generation.
John 1990-02-18	Sample maps
Paul 1990-04-06	Possible project name: Phoenix
John 1990-06-16	Second Joint Conference. Cities, bombs, roads, new pieces. Afterthoughts on multi-grid cities.
John 1992-04-26	First Online Summit. Ideas for a better interactive journal.
John 1992-05-03	Design of journal. Need text file map to test smoothing algorithms.
John 1992-05-17	Second Online Summit. John to work on journal, Paul to work on code.
Paul 1992-06-12	Progress on continent generator.
Paul 1992-06-14	First draft of continent generator.
Paul 1992-06-15	Generator now produces text files for input to HyperCard.
John 1992-06-17	Third Online Summit. Progress on HyperCard maps. Identifying continents. Bugs.
Paul 1992-06-19	Bug reports.
John 1992-06-25	Fourth Online Summit. No breakthroughs.
John 1992-07-01	Fifth Online Summit. How big can maps get?
John 1992-07-08	Smoothing algorithm proposed. Micro continents.
Paul 1992-07-12	Questions about smoothing algorithm.
John 1992-07-13	Questions and answers about smoothing.
John 1992-07-15	Sixth Online Summit. Paul agrees to give new algorithm a try, but it may take awhile.
Paul 1993-08-04	After a long pause, Paul returns to work on his continent generation program
Paul 1993-08-10	Questions and answers about the current state of the continent generation program
John 1993-08-26	Bug reports and initial graphics: commemorative painting and 3D globe
Paul 1993-08-28	Response to bug reports
Paul 1993-09-02	Progress report, trouble implementing smoothing
Paul 1993-09-06	Development continues at a snail's pace
Paul 1993-09-07	Conversion to C++ continent generator. Puzzle: How big does the spark list get?
John 1993-09-07	Graph showing size of spark list uncovered in the archives. Initial thoughts at refining the algorithm.
Paul 1993-09-08	First sketches of the one-pass smoothing algorithm, plain english recap of continent generation
Paul 1993-09-11	Two-pass smoothing, grid size
Paul 1993-09-18	Alexander icon revealed, grid size
John 1993-09-19	Bits or bytes? Landscape or portrait?
Paul 1993-10-21	Maps stack, multiple resolutions
John 1993-10-26	Attachment document, paratroopers, engineers, spies, patrol paths, automatic project schedules
Paul 1993-10-31a	Latest copy of the C++ generator, ideas for settings dialog
John 1993-10-31	"Mostly sea to mostly land" scrollbar, "chunky to stringy" scrollbar, grid lines, smoothing document, shoreline 100 times bigger than macro-level map
Paul 1993-10-31b	Improvements to world map display
John 1993-11-02	Problems with smoothing
Paul 1993-11-02	Memory allocation: "Go wild!"
John 1993-11-09	Attachment hierarchy, new pieces (infantry, rangers, helicopters, engineers), hovering vs. landing, planning missions, daydream mode
John 1993-11-13	Three strategies for storing shorelines
John 1993-11-14	Wish list: improvements to world map display, separate worldview window, smoothing
Paul 1993-11-15	Response to wish list
John 1993-11-19a	News flash: coastal smoothing achieved
John 1993-11-19b	First set of coastlines: "interesting but flawed"
John 1993-11-19c	Gridlines, wish list priorities
John 1993-11-19d	Second set of coastlines: "torn into powder"
Paul 1993-11-21	"Lacy" smoothing not working, need new approach
John 1993-11-21	Two ways to make gridlines
Paul 1993-11-23	Current method: painting gridlines over the surface
John 1993-11-24	Agreement on grid size, need to change current method, grid numbering, using sparks as cities, eight-way scrolling box
John 1993-11-28	Map colors
John 1993-12-03	Land and air pieces sharing the same space, mid-air refueling, armies vs. infantry vs. rangers, parachuting, helicopters, fuel dumps, engineering products (roads, bridges, canals, fuel dumps, artillery batteries, radar installations, anti-aircraft, land/sea mines), and a question: "Can ships sail under bridges?"
John 1994-01-21	Piece size, design, and color
John 1994-01-22	Piece design sketches - need larger size for beveled look
Paul 1994-01-23	Larger size approved
John 1994-01-28	Three piece styles (stone, wood, and steel) and seven ways to play (with three players). Forming and reforming attachment groups. Multiple pieces per cell on surface and air grids.
John 1994-01-30	Three person game with no neutral cities, map revealed from start. Library of pre-designed maps. Popup map pallette for easier navigation.
Paul 1994-02-07	New rev with larger grid size
John 1994-02-07	Sketch: twenty piece designs in six styles
Paul 1994-02-09	Confusion about surface and air grids
John 1994-02-09	Mockup of three-way battle on smoothed map
Paul 1994-02-11	Problems with non-destructive grid lines
Paul 1994-02-15	Status display dialog, graphics programs, sound effects during drags
John 1994-02-16	Mockup of attachment closeup window. Air grid increased to nine units per cell. Spy planes, unloading transports, ferrying infantry with helicopters, attaching helicopters to fighters, bombers with mixed cargo, mid-air refueling, automatic attack by attached subs, occupying vs. capturing neutral cities, fighters landing in a city vs. flying over it. Distinguishing between neutral, captured and enemy cities in close up view, scrolling within closeup view, modal vs. multiple views.
John 1994-02-20	Continents look like cat vomit - stringy, not beefy. Try globs instead of sparks. Window names. Further implications of an air grid - helicopter airlifts into a beseiged city.
Paul 1994-02-21	Response to attachment mockup. Confusion about which enemy pieces are visible. Spy planes and spy "attacks". Proposal: host plus all its attached pieces are allowed exactly as many attacks as the unencumbered host. Movement and attacks between surface and air grids. Proposal: allow armies to occupy as well as capture neutral cities. Rangers cannot jump from airborne helicopters. Definition of attachment still fuzzy: containment vs. coordination. Visualizing containment. Refueling. Automatic attack. Separate grid for subs? World view pallette, menus, window names.
John 1994-02-22	In battle map, fixed heirarchy determines which piece is on top (visible); in closeup view, not clear. New behaviors emerge when we distinguish between surface and air grids. Battle map displays either surface or air pieces depending on currently selected piece. Disagree on proposal limiting attacks of an attached group. Adjacency: air pieces must fly OVER surface pieces to attack. Occupying vs. capturing. Display of occupied cities. Helicopters cannot attach to fighters. Currently no concept of containment - difficult because heirarchies would be unavoidable. Restrict bombers to one type of cargo? No need for undersea grid. Closeup window non-modal. Globs instead of sparks. Menus, window names.
Paul 1994-02-23	Display number of pieces per cell in battle map. Charge one movement point for attachment/detachment. Can enemy see a sub under a carrier? Arguments for containment. Establishing and displaying containment. Occupation. Show asterisk on occupied neutral cities. Bombers restricted to one type of cargo. Two grids sufficient.
John 1994-02-23	Summary of 14 agreed-upon points. Indication of multiple pieces. When does attachment provide concealment? Argument against restricting attacks of contained pieces. Argument against containment. World thumbnail should float above battle map.
Paul 1994-02-26	Implementing the floating world thumbnail. Argument for containment tabled for now.
John 1994-03-10	New generation algorithm.
Paul 1994-07-07	Report on Empire Deluxe. Piece and city names, orders menu, promotions, production map, paths mode. More bug fixes.
John 1994-07-07	Request for more info on Empire Deluxe.
Paul 1994-07-08	Detailed notes on Empire Deluxe.
John 1994-07-08	Seven new features gleaned from Empire Analysis: minimum flight time, production map, city and piece naming, moving patrol paths, "sentry" instead of "sleep", pre-defined behavior scripts, production time reduced for subsequent pieces.
Paul 1994-07-09	Agreement on all seven features. Two more Empire screenshots.
Paul 1994-07-13	Simultaneous movement instead of turn-taking
John 1994-10-15	Automatically generated text radio messages for every action that could be intercepted by spies. Radio silence and blackouts. Teams of players on each side.
Paul 1994-10-28	Expand radio message idea to become fundamental part of architecture. Message passing across object hierarchies.
John 1995-07-17	More ideas: let bombers exchange bombs for fuel (range), radar sites on ships, stealth aircraft, flexible repair times, amphibious seaplane.
Paul 1995-07-19	All except amphibious seaplane approved.
Paul 1995-12-28	Question about generation algortithm: upper bound on sparklist.
John 1996-01-05	Idea: assign elevation to each cell.
Paul 1996-01-06	A resurgence in Alexander activity.
Paul 1996-01-09	Request details on phased generation algorithm.
John 1996-01-10a	Details coming soon...
John 1996-01-10b	Details and code found in previous letter.
Paul 1996-01-10	Excellent!
John 1996-01-28	Rumors of Conquest 4.0
Paul 1996-01-31	New version has fractal worlds and two new pieces.
John 1996-02-03	Rumors confirmed. List of new features with terrain screenshots.
Paul 1996-02-07	Fractal worlds: the bar has been raised.
John 1996-06-19	More ideas: enlarge cell display to show both air and land/sea pieces, enforced automation (automatic production schedules, canned rules of engagement, pathways, phases of play with limited manual override), spies, variation within pieces, rewards for capturing continents/countries.
John 1996-06-25	Lessons learned from first Conquest 4.0 game.
John 1996-06-29	Screenshot of first victory.
John 1996-07-12	Share these notes with Conquest creator?
Paul 1996-07-14	Sharing OK.
John 1996-07-18	Stalemate in a Conquest 4.0 game. Spies could attach to enemy pieces, spies and cities; automatic roads, bridges, and tunnels.
Paul 1996-07-31	Spies and automatic roads approved.
John 1996-08-22	Giant Conquest 4.0 map
John 1996-12-05	3D flyovers. Our land turns green and lush while enemy's land turns Mordor-brown.
John 1997-02-11	Suggestions for programming assignment. Determining what continent a given city resides in.
Paul 1997-02-11	Will use Alexander continent generation algorithm for programming assignment.
Paul 1997-02-12	ID tags for each land cell?
John 1997-02-12	Possible metadata for land cells. Could players build a Panama Canal?
Paul 1997-02-20	Possible modifications to generation algorithm: stacks and queues.
John 1997-02-20	Skepticism about stacks, uncertainty about queues - need to run experiments.
Paul 1997-02-21	Expect queue test files soon...
Paul 1997-03-16	First queue-based world - "a curious diagonal stringiness."
John 1997-03-17	First color map of queue-based world.
Paul 1997-03-17a	Another attempt using modified queue.
Paul 1997-03-17b	Dynamic array algorithm runs faster. A million cell world in 30 seconds.
Paul 1997-03-17c	More experiments. Tracking the continent ID of each land cell. "Seeds" instead of "Sparks".
John 1997-03-18a	Maps for three more test worlds.
Paul 1997-03-18a	Continent IDs might work as Country IDs.
John 1997-03-18b	What could we do with countries?
Paul 1997-03-18b	Need description of how color map GIFs are made.
Paul 1997-03-18c	Two more test files: Island World and Continent World.
John 1997-03-19	Maps of Island World vs. Continent World.
Paul 1997-03-27	First public presentation of Alexander technology.
John 1997-03-28	Next tasks: nailing down continent IDs, adding cities, adding mountains and rivers. Rivers could be both narrow and wide.
Paul 1997-03-31	First worlds with continent assignments.
Paul 1997-04-03	Two functions for copying worlds.
John 1997-04-04	Possible use for copied worlds.
Paul 1997-04-08	Using fractals for smoothing.
John 1997-04-08	Need more information about fractals. Adding a third dimension to algorithm.
John 1997-04-09	Placing cities. Twin cities would be interesting.

Sources

The first written record appeared on a Hypercard-based journal I created called Archipelago, which was also sent through the mail and received by about a dozen subscribers. Since none of the other Archipelagoans shared our obsession, Paul and I took the conversation offline using yet another Hypercard-based journal I created using my own HyperCard Essay Construction Kit (HECK). A facsimile of that journal was created as part of this exhibit.

In the early 90s, Paul and I both acquired email accounts; the remainder of our dialogue was taken from email records and preserved transcripts of early AOL chat rooms. We both eventually outgrew AOl, and over the next five years used a variety of email accounts, including work as well as home email. Paul and I both changed jobs several times. Fortunately, I preserved copies of most of my work email, and was able to recreate most of Paul's work email from copies captured within the body of my replies.

It was surprisingly challenging to salvage the text from even rather recent email records. The underlying formats kept changing and many files were all but unreadable by the time I located them. Some reconstructions took pain-staking detective work; in some cases I had blocks of text but no date. Paper notes of our taped conversations supplied additional clues. I was also able to recreate almost all of the various charts, sketches and maps we sent back and forth over the years. Tables and lists within the text were all converted to HTML by hand.

The final dialog consists of 170 web pages supplemented by dozens of specially-created graphics and took a full year of stolen moments to create. I suspect that few will ever read these pages, so this exhibit was truly a Project Of No Apparent Redeeming Value. It was a labor of love, but I have labored in the faith that even the most unlikely PONARVs can provide unexpected dividends.