I wanted the game to be easier to understand. At one point, that became a screen with too much explanation. Then simplifying the presentation took away parts of the game I actually liked.

The September 5 interface corrections are a useful record of both mistakes.

The first version repeated information, ran beyond the viewport and did too little with the illustrated cards. We contained the layout and moved detail into inspectors. But the correction also lost the two-sided replies and the illustrated wordmark. I asked for those back, along with a hand that felt held and a selected card I could identify immediately.

“Make it clearer” had become an incomplete design instruction. We needed to name what clarity was supposed to preserve.

More explanation can create more work

Showing the full rules and every consequence beside every option sounds helpful in a source review. In the browser, those explanations compete for the same limited space and attention.

The reader still has to locate the current choice, compare it with an alternative and find the action that commits it. More visible text can make that sequence harder.

We kept the immediate decision on the table: the response, the selected ability, the expected consequences and Play. A bounded inspector carries the longer rules and larger artwork. On very short screens, the response comparison opens separately while the current response remains visible.

The information still has a place. It does not all need the same place.

The conversation is part of the game

Restoring both speech bubbles did more than fill empty space. This is a game about an exchange, so seeing both sides helps explain what just happened.

Before the first post, the player's line is marked as a draft. After posting, both bubbles show the actual last exchange. Preparing another card or response does not rewrite that history. The rival's upcoming intent has its own place.

That separation lets the screen hold three different moments: what happened, what may happen next, and what the player is preparing to do. Collapsing them into one changing block would save space at the cost of meaning.

Selection needs to survive the cursor moving away

The three cards now fan across the bottom of the table. Hover or focus brings a card forward. Selection keeps it raised with a frame and an explicit selected stamp.

That persistent state matters because choosing a card does not spend a post. The player can move away to inspect consequences and return to the same prepared action. Pressing Play is the commitment.

The frames, titles, artwork and flavor text also restore the feeling of handling cards. They do not silently add a rarity system or change the rules. Visual personality has to remain honest about the mechanics it represents.

Look at the space a person actually has

The September 5 checks covered 66 recorded state and viewport combinations, including short landscape screens, duplicate cards, pending bonuses and held Auto play. The later v5 campaign had its own 42 phase and viewport checks. These are different verification records, not numbers to combine into a grand total.

The browser checks established that the measured controls fit and behaved as expected. Looking at the captures helped catch compression and hierarchy problems. Neither step established that new players understood everything or enjoyed the result.

For another complex interface, I would start with one real decision on the smallest useful screen. Identify the current state, the prepared action, its consequence and its commit control. Then open the detail. Finally, check whether the screen still feels like the product people came to use.

I want less searching and less repetition. I also want the absurd argument, the illustrated cards and the little bit of theater that makes this particular game worth opening.

Record and method

Written September 6, 2026, about the September 5 contained-desk and card-table corrections, with the later v5 check scope identified separately. AI assisted with drafting from those records. The current demo includes later campaign work; the hero image is artwork, not a before-and-after screenshot.