Category
Decision Makers for Everyday Choices
Dinner, films, weekends, conversation — filtered first, then decided.
Six tools for the choices that stall a group. What they share is not a mechanic — it is an order of operations: narrow the options by real constraints first, then draw from whatever survives. Every other category on this site draws from a set you supply or a set that is fixed; here the set is computed from what you say is possible.
That ordering is the whole design and it is worth understanding why it matters. A random answer you were always going to reject has not decided anything, and the failure mode of every naive decision tool is that it produces a suggestion which violates a condition somebody had not mentioned. Filtering first means the suggestion that comes back cannot be rejected on a technicality.
Each tool shows how many options survived your filters before it draws, which is the number that tells you what the draw is actually doing. Seventeen survivors is a genuinely open decision; three is a d3 dressed up as one, and both are legitimate — you just want to know which you have.
The other thing these tools have in common is that they refuse rather than improvise. If nothing matches, they say nothing matches. None of them will quietly relax the strictest condition you set in order to produce an answer, because an answer that violates a stated constraint is worse than no answer at all: you may not notice, and the whole point was that the tool was answering rather than you.
Dinner Decider
Filter by diet, effort, and budget first — then let the draw settle it.
What to Watch Picker
Paste your watchlist, draw a shortlist of three, then commit to one.
Weekend Activity Picker
Weather, budget, energy and time in — one thing to actually do out.
Conversation Starter Picker
Questions that get deeper as you go — a stratified draw, not a flat one.
Icebreaker Question Picker
Tell it your group size and minutes — it does the time maths, then draws.
Would You Rather Generator
Draw a dilemma, tally the room's split, and see where the group actually lands.
Constraint-first: dinner, weekends and what to watch
Three tools where the interesting work happens before the draw. The dinner decider filters a curated corpus on four independent tags — diet, hands-on time, cost band, and whether ordering in counts — and the weekend activity picker filters on five, adding weather and energy. In both cases the draw over the survivors is the simplest uniform index pick on the site, and the filters are what make it usable.
The filters compound faster than people expect, and that is the single most useful thing to know about this group. Each condition on its own looks mild; four mild conditions applied together routinely cut a twenty-option corpus to two. A bad-weather, low-energy, free, two-hour Saturday genuinely has very few options, and seeing the count fall as you set them is information rather than a fault.
The what-to-watch picker works differently and is grouped here because it solves the same social problem: it draws a shortlist of three from your own pasted watchlist, then draws one of the three. The arithmetic composes exactly — a 3-of-n shortlist followed by a 1-of-3 pick gives every title precisely a 1-in-n chance — so nothing has been distorted. What changes is acceptance, because choosing between three options is a thing people will do and accepting a verdict is not.
Conversation: where a uniform draw is actively wrong
The conversation starter picker is the only tool on this site that is deliberately stratified rather than uniform, and the reason is concrete: drawing uniformly from a corpus that ranges from small talk to genuinely personal will open with the personal question about a fifth of the time, which ends the conversation it was supposed to start.
So the corpus is tiered by depth and the draw only ever samples the tier you have selected. Questions in tiers you have not selected have a zero chance rather than a small one, which is a stratified draw and a genuinely different thing from a weighted one — weighting changes relative odds while leaving everything reachable, and stratifying removes options from the sample space entirely until you move the boundary.
The icebreaker picker solves an arithmetic problem instead of a social one. It prices each question format against your group size and your available minutes, and the pricing is where the tool earns its place: a whole-room question costs its duration multiplied by attendance, while a paired activity costs one turn regardless of room size. That difference is why a format list that works at twelve people is unusable at fifty, and why the available options change character rather than degrading gradually somewhere around fifteen.
Games that produce a discussion rather than a decision
The would-you-rather generator sits in this category because it is shaped like a decision tool and is not one. It draws a dilemma from a no-repeat cycle and then tallies how a room votes, and the tally is the output — the dilemma itself is only the prompt.
It deliberately does not balance the two options. Several of the dilemmas here split heavily in most groups, and a nine-to-one result is a fact about your room rather than a fault in the question. A generator that selected for evenly-split dilemmas would be selecting for the dullest ones, which is a real trade worth naming.
It also keeps the side order fixed rather than randomising it, which is a small, real and deliberate asymmetry. People demonstrably favour the first option they read, and randomising the order would produce a cleaner tally and a worse game — the same dilemma reading differently on two occasions is confusing in a room where people are already shouting at each other.
What none of these tools can do
None of them know anything about your preferences. Two options you feel completely differently about have identical odds within the surviving pool, because the tool has no representation of how you feel about either — and building one would require asking you, at which point you have made the decision yourself.
None of them have memory. The same suggestion can come up two nights running, which is what independence means and is occasionally infuriating. That is a deliberate omission rather than an oversight: a dinner suggestion that repeats is a legitimate suggestion, whereas a student called on twice running is a problem, which is why the no-repeat cycle lives in the classroom tools and not here.
And the corpora are curated and finite. Anything not on the list has a zero percent chance however much you would have enjoyed it, and the blind spots in a hand-picked list of twenty-five dinner options are real and invisible from the inside. Where you want a draw over options you supply yourself, the spinner wheel and the name picker do exactly that.
Frequently asked questions
Why do these tools filter before drawing rather than after?
Because a random answer you were always going to reject has not decided anything. Filtering first means the suggestion cannot be rejected on a technicality nobody mentioned.
Why does the surviving option count matter?
Because it tells you what the draw is doing. Seventeen survivors is a genuinely open decision; three is a d3, and both are fine as long as you know which you have.
What happens when nothing matches my filters?
The tool says so and refuses to draw. It will never relax a condition you set in order to produce an answer, because a result that violates a stated constraint is worse than no result.
Why is the conversation picker not a uniform draw?
Because drawing uniformly across all depths opens with a deeply personal question about a fifth of the time. The tiers exist so the fifth question is not the same weight as the first.
Why does group size change the icebreaker suggestions so much?
Because whole-room formats cost their duration times attendance, and paired formats cost one turn regardless. Above about fifteen people the whole-room options stop fitting entirely.
Can I add my own options to these tools?
Not to the curated corpora. The spinner wheel, name picker and what-to-watch picker all draw from a list you paste in.
Do these tools remember what they suggested last time?
Only the would-you-rather generator, which cycles through its dilemmas. The rest draw independently, so a repeat two nights running is ordinary.
Why does the would-you-rather generator not balance its dilemmas?
Because a lopsided split is a real result about your group, and selecting for evenly-split dilemmas would mean selecting for the least interesting ones.
Are these tools any use for a two-person decision?
Partly. At two people a veto costs almost nothing, so a single answer gets rejected — which is why the what-to-watch picker draws a shortlist rather than a verdict.