This update is about making the simulation line up with the game you thought you were playing: your platform means what you wrote, the calendar means the date on the screen, the next election is the real next election, and the AI should stop holding the whole game hostage when it is late.

It also adds two big new ways to play: custom platforms on every career path, and Story Mode for players who want the consequences without every chart, margin and model readout on screen at once.

The political compass and write-your-own-platform editor used to exist only in the presidential campaign setup. If you started as a state legislator, mayor, governor, senator, cabinet secretary or vice president, you never saw either of them.

They are now part of building your politician, wherever you start:

• Pin yourself on a two-axis compass instead of a left-right slider, then name your own ideology. "Agrarian Distributist", "Prairie Populist", "Market Green", whatever you are actually running as. That name replaces the old quadrant label in the game and in AI-written text.

• Write your real position, in your own words, on eleven policy areas: economy, healthcare, immigration, climate, diplomacy, education, crime, guns, trade, civil rights and a final area for what the parties will not touch.

• The game scores those planks live as you type, using a deterministic offline keyword system. It then routes the result through the same policy machinery as every preset stance, so state effects, enthusiasm, electability and debate bonuses all work normally.

The preset stances are untouched and still the default. If you never open the compass or custom platform editor, the old bar and old buttons behave exactly as before. Two calibration bugs were fixed while testing it:

• Domestic phrases like "let Medicare negotiate drug prices" no longer accidentally score as dovish foreign policy.

• Ordinary left positions no longer saturate the economic axis as hard as outright nationalisation.
  

On first launch, the game now asks one question: how much math do you want on screen?

Neither answer changes the rules, the difficulty, the events, the outcomes or the simulation. The same decision plays out identically in both modes. Story Mode only changes presentation.

Story Mode - relaxed. You still see the people, the choices and the consequences, plus a plain-English read on how you are doing. Charts, analysis panels and redundant figures step back.

• Approval history charts, favourability sparklines and win-probability breakdowns are hidden.

• Political compass panels, demographics, voter blocs and campaign-finance analysis are hidden.

• Stat bars lose trailing numbers where the bar already communicates the result. Funds become words like "Comfortable"; win chance becomes words like "Toss-up".

• Momentum disappears entirely, because a bare signed float was not helping anyone make a better decision.

The presidency sidebar goes from thirteen panels to seven. It keeps what comes to you and what tells you how you are doing: approval, policy performance, relationships, agenda, the bill docket, the diplomatic bag, executive orders, treaties and the Supreme Court.

The deeper machinery is folded, not deleted. A More tools button in the corner nav brings every panel back unchanged, so scenarios, milestones and advanced play can still reach everything they need.

Approval keeps its percentage in both modes. It is the one number the whole game is about, and "54%" and "41%" are not the same feeling. Policy letter grades stay too, because a grade is already a readable summary.

Analyst Mode - the full simulation. Every poll, margin, trend line, budget line and seat count stays on screen. This is the game as it has always displayed itself.

You can switch any time in Settings -> Play Style, and you can show the intro question again from there too.

The game still runs on its internal day counter. Saves, milestones and challenges are keyed to that day number.

What changed is that every day now maps to one real calendar date, pinned to the historical beats that have to be right:

That means the badge and the milestone beside it can no longer disagree. Status badges now read like this:

The same date appears in event cards, close-of-business cards, election night and AI prompts.

Under the hood, date maths is clock-free. Nothing new reads your computer's current date. The game derives the political year from the calendar date, so the year now turns over on January 1 rather than on the anniversary of the oath. The old rule disagreed with the displayed calendar on 22 days of a governor's term.

Every office now asks one shared election-calendar system whether a seat is up.

• House seats run every even year.

• Senate classes use the true six-year rotation.

• The presidency runs every four years.

• Governors follow their actual cycles: midterm states, presidential-year states, New Jersey and Virginia odd years, Kentucky/Louisiana/Mississippi odd years, plus two-year New Hampshire and Vermont.

• State legislatures use real seat counts and the appropriate odd-year states.

Two old bugs were behind a lot of wrong career-ladder offers:

• Presidential-year Senate races were drawing from the wrong third of the country.

• The ladder was handing every race the year your current term ended, even when the office you were running for had a different cycle.

The menu now resolves the race before it offers it, including the wait if your next legal run is out of office:

The AI gets the same authoritative table, including your own seat and, for a second-term president, the 22nd Amendment instead of an invitation to run again.

Both chambers are now playable where a state has two chambers. The setup screen lets you choose House or Senate, with each state's real seat count and term length.

Term length now drives the length of the run. A four-year state senate term runs 1460 days, with the twelve beats stretched across it.

Nebraska now collapses correctly into the Unicameral:

• 49 members

• all titled Senator

• four-year staggered terms

• nonpartisan ballot

• LD- seat prefix

• "Caucus Leadership" becomes The Executive Board, because a nonpartisan chamber has no caucus

The presidency header also had some cleanup: the old "How to Play" and "Menu" labels are gone from the header, while the corner nav still carries Guide and Menu.

Annexation now opens a full nine-stage integration track:

There are three governing postures, each with a real trade:

• Garrison keeps order but cannot build the legitimacy needed for statehood.

• Development builds capacity and integration.

• Devolution lowers resistance but can stall the hard institutional work if you move too early.

Measured timelines now look like the system was intended to work: best play can admit a territory in roughly 11-12 years, ordinary play is closer to 22 years, garrison-only never gets there, and devolving too early can also strand the process.

The continuity requirement is simulated rather than scripted. The panel projects admission against the last year you are eligible to serve, and links to the Article V panel when the dates do not fit. Repeal the 22nd Amendment and that warning correctly becomes: there is no longer a constitutional date on which you stop being president.

Admission now lands in the same new-state pipeline used by DC and Puerto Rico, so a new state immediately gets senators, House seats, electoral votes and updated majority thresholds.

Two simulation bugs were fixed:

• Monthly progress was being rounded every tick, which erased any rate below 0.5 per month and froze some paths forever.

• The projection was dividing distance by rate in a coupled system, so fresh territories could read "never" even when the forward simulation admitted them in 12 years.

Independent and custom-party candidates were still getting punished by the major-party escalation ladder on repeated risky actions.

That is now fixed in the shared risk function. Independents and custom parties stay flat at zero risk fatigue for the risky-action escalation, while Democrats and Republicans still ramp as before:

The same fix covers the effect roll, the summary line and the button badge. Policy pivot had a separate escalation ladder, so independents are now flattened there too, at the intended flat 28% base.

Major-party behaviour is unchanged.

This update fixes the root cause behind several reports that looked like separate bugs: freezes, lag after generation, mouse slowdown, missing briefings and "Preparing your briefing" hanging forever.

The renderer and the model process disagreed about how long a generation was allowed to take. The screen gave up after roughly 20 seconds, while the main process could keep generating for 75, 100 or 150 seconds depending on the job. The queue is first-in, first-out, and local inference can use every CPU core, so dead jobs were delaying live ones long after the player had moved on.

That has been rebuilt:

• Every AI job now carries the deadline of the screen waiting for it.

• A job whose deadline passed while queued never starts.

• A running job aborts at its deadline.

• Token budgets are sized to replies that can actually arrive in time.

• The bridge no longer silently floors every wait to 60 seconds.

The "Preparing your briefing" freeze had another layer: the procedural situation builder could make two extra unbudgeted AI calls after the first briefing call had already spent its time. One day could hold the spinner for more than 150 seconds.

The turn now has one shared AI budget. A scheduler error can no longer skip the blank-turn guard. A 38-second watchdog publishes the written briefing if the briefing area is still a spinner. The Don't Wait button is gone, because you should not have to rescue the game from its own loading state.

The live scheduler had lost the periodic AI-retry reset. Once six briefings failed, the model path could stay switched off for the rest of the presidency, which is why some players kept seeing curated fallback headlines like "a key NATO ally..." instead of the intended AI-written event.

That reset is now in the scheduler that actually runs.

There is also prefetch. Tomorrow's briefing is written in the background while you read today's, in a lane that is served last and aborted instantly if you are waiting on something interactive. On a healthy machine, the wait is often zero; when the model has room, it gets longer to write well instead of being rushed into the visible turn.

The four AI-written options already existed. The event just often did not arrive in time. Now it should.

Several stacked bugs made custom legislative orders grade badly even when they were obviously legislative.

• The instrument detector required the literal phrase "a bill", so "a Farmer Protection Bill", "this law" and "in this bill" could all fail to register.

• An order with no briefing context was structurally capped at a D because relevance was measured against an empty string.

• The narrator shared by multiple career rungs still said "president" and used presidential crawl text.

• The spell checker could rewrite "foot" to "food".

That path now recognises the bill, grades it against the actual order, uses the correct office language and leaves words alone when it should.

The President can now draft a bill and veto one.

Draft a Bill. Write a title, say what it does, pick a scope and decide whether you are going to insist it is paid for. Legislative Counsel drafts it, a member introduces it, and it lands on your docket like anything else, with a Congressional Budget Office score, a whip count and a floor vote.

The veto. Congress passes things you did not ask for, and the more of it the opposition holds, the more of them arrive on your desk. Sign it, veto it, or let it become law unsigned. A veto triggers a real override attempt at two thirds of both chambers. A veto that holds is one of the strongest things the office can do. One that gets overridden is the opposite.

Bills used to pass or fail almost entirely on party arithmetic. Real chambers do not work like that: trade, surveillance, war powers, farm bills and tech regulation all split the room on axes that are not always party lines.

Roughly a fifth of the legislative docket now carries a different kind of coalition:

• Cross-cutting - both conferences are divided and nobody's whip list is any use.

• Inverted - the other party carries your bill while your own conference revolts. You can pass it. You will pay for it.

• Theirs more than yours - a bipartisan bill the opposition wants more than your own members do.

The roll call now shows the split by party, so you can see exactly whose votes carried it.

Party-line bills were also rebalanced. Party unity is now derived from how much slack the majority actually has: a leadership with four votes to spare whips like it, and a leadership with forty lets members vote their districts.

Some of what arrives on your desk is not a crisis. It is one of your own appointees moving first and leaving you to explain it.

When that happens, the briefing now says who did it, by name and by title. And you can do something about it:

• Haul them in - costs nothing, fixes little, and everyone hears about it by lunchtime.

• Move them off the file - they keep the job and lose the portfolio.

• Fire them - the seat goes empty, the department loses a step, and it leads the bulletins for two days. One firing reads as decisive. A fifth reads as chaos.

Not everyone is yours to fire. A President cannot dismiss an Inspector General over an investigation or a Fed chair over a policy disagreement. A Governor does not employ an elected Attorney General. You can demand a resignation in public and find out whether they go.

Firing works, too: a disciplined operation freelances less. It is a lever, not a mood.

The old presidency tutorial is gone: the six-card overlay, the auto-open on first entry and a vestigial no-op startup wrapper. The guide system covers the presidency now, and it does not cover the briefing you are trying to read.

The full guide still covers eleven screens: the menu, building a politician, the campaign, the presidency, the governor's mansion, city hall, Congress, the statehouse, a cabinet department, the vice presidency and election night.

Quit Game has been added to the Your Game card. In the desktop app it opens a confirmation dialog and then quits the app.

• The Menu button no longer sends you to the campaign announcement page. It takes you to the menu.

• Loading a save after restarting the game no longer breaks the UI, and the World Diplomacy Map button no longer disappears.

• Outcomes are no longer graded F when they succeeded.

• The career ladder no longer lists Quick Play and Challenge saves as careers to resume. They are still in the normal load menu.

• Unrequested events cost less money. A disaster you never asked for still costs something, but it can no longer run a treasury into the ground on its own.

As always: thank you for the reports. This update is full of systems that got better because someone noticed the exact place where the game was lying, stalling or flattening something interesting. Keep them coming.

- Dackers Studios