DEV BLOG / 14
Joe becomes a grounded conversational general manager
More useful advice, real league context and some firm limits on what the assistant can do.
Joe started by explaining buttons. Now he needs to understand more of the franchise behind them.
That gives him a better job than reading the manual aloud. He can help me interpret the team, find a coach, examine recorded opponent habits and make sense of a performance trend. He still needs to stay attached to things the game actually knows.
What changed since the aging and personality work
- Performance advice uses completed results. It states the period and sample, removes duplicate matches and stays in preparation rather than pretending to call every play.
- Opponent advice uses historical choices. Joe can suggest a counter to recorded tendencies without revealing a future selection.
- Team, rival and coach briefings have links. Name and ID matching, ambiguous-city choices, profiles and strategy histories make the directory usable.
- Live reactions use actual events. Named participants and final results feed bounded messages, with priority, expiry and duplicate suppression.
- Conversation has continuity. Related follow-ups, recent or cleared chat and three short saved conversations have defined limits.
- Speaking has visible feedback. Typing and mouth animation accompany answers; work can be canceled and an outdated answer discarded.
- Desktop conversation and mobile help differ. Desktop has the larger local conversation integration; compact and mobile views retain authored help and fallbacks.
- The world has eighteen recorded seasons before the new coach arrives. Historical players, teams, coaches and matches populate the archive.
- All-Star selections and Joe’s own history are browsable. Joe Mamatania has a retired-quarterback story in that frozen league.
- The football itself has more detail. Return touchdowns have their own statistics, with contextual onside kicks, blocked field goals and rare year-long injuries in the expansion.
- Coach progression has two tracks. Permanent trait choices arrive every five seasons; seasonal Momentum uses separate offers, limits and loss-reset rules.
- AutoSim can use bounded assistance. Preparation, strategies, earned traits, Momentum and eligible injury cover still go through ordinary validation and fallback rules.
Useful advice needs a little restraint
An opponent who preferred Ground Control in recorded games gives me something to consider. It doesn’t prove what he’ll choose next. A team with a short run of poor results needs that sample identified before Joe treats it as a lifetime diagnosis.
Correspondence follows the same rule. Rival-coach responses use the recorded result and tone, stay suitable for the game and remain stable when reread. Live highlights should react to the current event, not replay an old game as fresh news.
The help library has grown too, reaching 235 topics and 201 lessons in this part of the work. Joe can greet the manager, respond to thanks, explain scoring formulas or clear the recent conversation. There’s an authored fallback when an answer can’t be produced usefully. I’d rather have a straightforward explanation than a confident guess.
The computer clubs have a separate decision system from Joe’s help. Keeping those roles distinct avoids turning the assistant into an unexplained authority over everything.
More conversation, still ordinary rules
Joe can explain ratings, scoring, training, injury cover, rosters and drafts. He can point to financial planning, contract guarantees and options, department effects, payment timing and bank quotes. Those answers need to lead to the right screen and use the information actually available.
He cannot make an unaffordable decision valid just by sounding confident. I’m quite capable of that mistake on my own.
Next the league gets better ways to recognize careers and manage the players moving between clubs: the Hall of Fame and waiver market.