# Arena: decisions beyond painting the map > Current status reviewed September 29, 2026. See [the release index](README.md) for shipped versus planned features, network, wallet signing and deployment boundaries. Older dated experiments below are retained as history. **October 1 update:** new practice matches use **Control v4**. Cores now score time held, every objective is equidistant from both bases, isolated pockets cannot act and repairing costs 3 energy with a four-second delay after each hit. Separate onchain v4 contracts passed local EVM and two-wallet browser tests; deployed league v2 remains unchanged. See [current rules and migration](orb-arena.md). The v3 prototype and design rationale below are a historical record. Roles, coordinated bonuses and pings remain proposals. Legacy practice saves retain their own version. ## Playable practice prototype - Orthogonally connected territory scores one point per sector. The connection must reach the team's original base (A1 or L8); bases cannot be attacked. - Cutting a bridge isolates the pocket. Ownership remains; a dashed border and desaturated portrait identify it. Reconnecting immediately restores its territory points. Islands may still attack/expand/reinforce to rescue themselves. - A core also needs a path to its base at the award boundary. The first five awards are 8 points; the final core at G6 is worth 16. Its location and value are visible throughout the round. Earned bonuses are never revoked or awarded retroactively. - Server rules settle elapsed awards before a move changes the map. A retry is actor-bound and idempotent; the existing version check serializes concurrent moves. - No extra polling, background scoring timer or transaction is introduced. Supply is a bounded flood fill over 96 cells; the onchain version still needs separate gas benchmarking and multiplayer balance tests before adopting it. Validated with deterministic rule tests (cuts, rescue, award-boundary moves, final award, protected bases, deadlines, team limits and legacy compatibility), local API tests, and a browser fixture using the real practice component and rules in EN/PT. The recorded local bot match finished 96–6. This is one scripted match, not player research, but it illustrates a risk: an early territorial lead gives the winning team more adjacent targets, and isolated tiles remain productive forever. More clicks and an ever-larger map would amplify that problem. ## Design rationale and next experiments **Supply lines.** Territory must connect to the team's starting base to produce territorial points. Cutting a narrow bridge isolates a pocket: the portrait stays, but its border becomes dashed and it stops scoring until reconnected. An isolated pocket is not automatically handed to the attacker. This introduces flanking, rescue and defensive choices without stealing wallet assets or changing ownership outside the match. Scores can be recomputed for a bounded 96-cell board; benchmark the contract cost before choosing implementation. Never assume a full-board scan every transaction is cheap merely because the chain has high throughput. **Coordinated attacks.** A reinforced sector becomes a short team objective: one player pressures it, another lands the capture while a teammate maintains the supply route. Start by making existing shields and pressure legible; only add a combo bonus after multi-player testing. A solo player must retain a slower path to the same objective. No advantage from paid boosts or sending duplicate moves. **A meaningful finish.** Announce the last core early and increase its final award relative to territory. This creates an obvious place to contest near the end, without randomly choosing a winner or secretly buffing the losing team. Tune the value against typical score gaps: the final objective should matter, but should not erase the first two minutes of play. ## Later options, if the core loop is fun - **Roles chosen before the round:** scout (recon/route markers), defender (temporary shielding), disruptor (temporary supply interruption). One distinct cooldown per role, the same total power budget, and clear counterplay. Prototype this separately from supply lines to avoid changing everything at once. - **Team pings:** mark “attack here”, “protect bridge” or “next core”. These are ephemeral communication, not transactions, and complement voice chat. - **Short rematch loop:** winning community portrait, MVP by useful contributions rather than clicks, then a one-button rematch with fresh consent where required. Trophies express achievement; they never strengthen a player in the next game. ## Visual change implemented now Owned sectors display a square community portrait in place of the diamond. The background is neutral; a subtle team-colored border, coordinates, shield marks and a gold objective outline preserve readability. Scoreboards use the same portrait. Missing/inaccessible images use initials. Each available community image is fetched once per mounted board and reused across cells via an object URL. Changing the icon refreshes it; unmount aborts requests and revokes object URLs. Images remain permission-checked offchain branding. Public chain rounds only provide community IDs and names, not permission to view a private community image. The current UI resolves pictures from communities visible to the signed-in user; an inaccessible opponent uses initials. Open test rounds have arbitrary team names and no community binding, so their square marks are initials. Do not infer identity from a matching name or loosen the avatar API's access controls. ## Validation before a rules release Prototype rules in practice first. Test 1v1 and coordinated teams, deliberately fall behind early, try cutting/reconnecting a bridge, and confirm both sides can explain why they won or lost. Measure comeback opportunities, useful team actions, uncontested clicking and rematch interest. Use the results to decide the smallest onchain v3 change. A contract update requires a new version and explicit treatment of existing rounds and sponsored pools; do not replace deployed bytecode in place. Fast, independent confirmed actions across many matches are the legitimate reason to use Monad. Extra clicks, chat messages onchain and synthetic TPS are not gameplay.