One of the harder things about making Yearbook is that I know a good clue when I see one.
That is not especially helpful when you're trying to build a system that has to produce four of them every day.
A good clue is factual, obviously. It should point to something that actually happened in the year you're trying to guess. It should be understandable. It shouldn't accidentally contain the answer. It should probably not require a PhD in a very specific branch of chemistry.
But none of those things are really what makes a clue good.
The best clues give you that little “oh yeah… what was that?” feeling.
Something starts moving around in your brain. You know you've heard of this person or movie or event. Maybe you can picture it. Maybe you remember people talking about it. The answer is somewhere in there, just not quite available yet.
Then it clicks.
That feeling rocks.
The clue is a little puzzle of its own
Yearbook is technically a game about guessing a year.
But I don't really think of the four cards as just four pieces of information about that year. Each clue is also a tiny puzzle.
It usually shouldn't just say:
Spider-Man starring Tobey Maguire was released.
That's information. It isn't much of a clue.
The fun part is dancing around the subject just enough that the player has to recognize what you're talking about before they can even start thinking about the year.
Maybe the clue gives you the shape of the thing. Maybe there's a little pun. Maybe it describes a detail that makes the answer feel obvious after you get it.
The goal isn't to make the player decode a riddle written by a wizard.
It's just to add one extra layer between reading the clue and knowing the fact.
And sometimes that layer is where the best part of the game happens.
A movie can take you straight back to fifth grade
One of my favorite Yearbook moments personally was a clue about the original Tobey Maguire Spider-Man movie.
I knew the year almost immediately.
Not because the release of Spider-Man was some defining event in my life.
I just vividly remember being in Mr. Parsons' fifth-grade classroom around the time it came out. I remember people talking about it. For whatever reason, that movie is attached to that little slice of my childhood.
The clue brought the whole thing back.
That's really the feeling I was hoping Yearbook could create when I started making it.
Sometimes a historical clue sends you toward something obviously important. Other times it reminds you that everybody at school was talking about a superhero movie 24 years ago.
Both are useful memories.
Maybe the second kind is even more fun.
Different people have completely different calendars in their heads
I play Yearbook every day in a family group chat with my mom and my sisters.
One of the things I like most about that is that we're not all drawing from the same set of memories.
There are things from the '60s that mean something immediately to my mom and absolutely nothing to me.
On those days, she may have the nostalgic reaction and I get the other version of Yearbook that I really like:
Wait, I've never heard of that.
Then I can click the Wikipedia link afterward and spend five minutes reading about something I wouldn't have encountered otherwise.
We had a clue recently involving the Supreme Court decision that prohibited school-sponsored prayer in public schools. While reading about it afterward, we realized the school district involved was very close to where we grew up.
None of us knew that.
Did learning it materially change my life?
No.
Was I weirdly delighted to know it?
Absolutely.
That is basically my relationship with Wikipedia in one example.
Obscure is not the same thing as difficult
This is probably the biggest challenge in clue selection.
There are millions of true things you could ask about.
That doesn't make them good trivia.
A random scientific discovery might be hugely important within its field. But if almost nobody outside that field has ever heard of it, there may not be much for the player to retrieve from memory.
It becomes less:
Oh yeah, what was that?
and more:
How on Earth would I know that?
Those are very different feelings.
I'm completely comfortable with somebody getting an F on Yearbook.
My family group chat has produced some spectacular Fs.
That's part of the game.
But you probably shouldn't be getting an F every day.
That would be miserable.
Ideally, most players can get the exact year or at least get reasonably close by the final clue. And when the puzzle is hard, I want it to feel like the player just didn't happen to know the right things that day, rather than like the game asked them to identify four facts chosen from the bottom of Wikipedia.
Clues should get more helpful
The basic idea is that the first clue should be the hardest and the last clue should give you the best chance.
That's easier to say than to implement.
How recognizable is a specific event?
How famous is a person?
Will somebody who was 15 when this happened remember it differently from someone who was born 20 years later?
You can use data to help.
The current Yearbook system looks at things like Wikipedia popularity, category fit, article alignment and whether an event can be described without immediately giving away the answer. It also groups candidates into broad recognizability bands so a normal puzzle isn't accidentally made up of four deep cuts.
That makes the system better.
It does not make the problem solved.
Human recognition is weird.
There are things everybody has technically heard of but nobody remembers when they happened. There are things that barely register in general history but are permanently burned into the memory of a specific generation.
So the system can estimate obscurity.
It can't completely understand nostalgia.
Neither can I, honestly.
I mostly know it when I feel it.
The year itself creates another clue
One of the features that changed Yearbook the most actually has nothing to do with clue wording.
Early versions were harsher about being wrong.
If the answer was 1987 and you knew the event happened somewhere around 1988, the game didn't do a very good job of recognizing that you were basically in the right neighborhood.
Players found that frustrating.
And they were right.
So now guesses give proximity feedback.
One year off is different from five years off. Five years off is different from ten. Letter grades give some credit for getting close even if you never land exactly on the answer.
That also created another form of deduction.
If I don't know an answer, I'll often guess something round like 1970 or 1980.
Then maybe the game tells me I'm within ten years.
Next guess: 1975.
Now I'm within five.
Suddenly the trivia game has turned into a tiny math problem.
You're using the historical clues and the calendar clues together.
I like that because it makes the puzzle more forgiving without simply making the clues easier.
A clue can be correct and still be bad
This distinction has become increasingly important as I've worked on the game.
A clue can contain nothing factually wrong and still fail for all sorts of reasons.
It can be so vague that it could describe 20 different events.
It can accidentally reveal part of the answer.
It can be repetitive.
It can make a terrible pun.
And I like terrible puns, so that is saying something.
It can use a voice that sounds like an AI was instructed to “make history fun and engaging for today's curious learner.”
That is usually not what I'm looking for.
My preferred Yearbook voice is a mixture of factual and playful.
A pun is good.
A little absurdity can be good.
A clue that feels like somebody is leaning across the table and going, “Oh, this one is kind of funny,” is good.
What I don't want is every clue straining to be clever.
Or grand.
Or profound.
Sometimes a movie came out.
Sometimes somebody won something.
Sometimes a gadget became popular.
We do not need to conclude that the event “changed the way humanity would experience the world forever.”
Serious things need a different voice
This is one place where “make it fun” has a pretty obvious boundary.
History contains wars, disasters, assassinations, violence and tragedies.
Yearbook's visual language is playful. The cards are a little crooked. They move around. The clues can use puns.
That does not mean every subject should be treated the same way.
One of the things the clue pipeline checks now is tone. Sensitive and serious events have stricter wording rules, and certain playful constructions are blocked entirely.
That isn't me trying to turn a two-minute trivia game into a scholarly journal.
It just seems like basic respect.
I would rather have a clue be slightly less entertaining than have somebody read it and think:
Why on Earth did he think that was an appropriate joke?
People are very good at finding the thing you didn't think of
Some of the best improvements to Yearbook came from people playing it in ways I hadn't anticipated.
Early scoring systems made more sense to me than they did to anyone else.
People told me they wanted to replay games they missed, so I built the archive.
People liked comparing results in group chats, so the game added distributions that make that comparison more interesting.
People emailed when clues were tied to the wrong year.
Someone even figured out that the hidden year at the top of the game could be revealed by highlighting the obscured text with a mouse.
Which is, admittedly, an extremely inefficient method of cheating at an internet trivia game.
You could also just Google the clue.
Still, it was dumb, so I fixed it.
That feedback also changed the way I think about clue quality.
I can review a clue and think it is perfectly fair because I already know what it means.
That is a pretty serious disadvantage when evaluating your own trivia.
The player doesn't have the answer key sitting in their brain.
Watching where people struggle, how the result distribution looks and what kinds of questions or complaints come back is useful evidence.
If almost everyone bombs the same day, maybe everybody suddenly forgot history.
Or maybe I made a bad puzzle.
The second possibility deserves consideration.
Fair doesn't mean easy
I don't want Yearbook to become a game where the fourth clue basically says:
This was the year 1987.
Difficulty is part of the fun.
Getting something from the first clue should feel great precisely because it doesn't happen all the time.
And sometimes a clue can be obscure and still belong in the puzzle.
I actually like having the occasional deep cut.
If three other clues give players solid ways into the year, one strange fact can be the thing they learn that day.
That's a good trade.
Four strange facts is not.
So I think about fairness less as “Will everyone know this?” and more as:
“Does the puzzle give a reasonable player enough different ways to find their footing?”
That's also why variety matters.
History, sports, entertainment, politics, science and culture all activate different parts of people's memories.
The clue that does nothing for me might be the one somebody else immediately recognizes.
The best clue leaves a little room for memory to do the work
I wish I had a perfect formula for this.
I don't.
There is now a frankly unreasonable amount of machinery checking Yearbook clues for factual grounding, answer leakage, category fit, recognizability, tone and awkward wording. The system will reject clues rather than automatically publish things that fail those checks.
I'm glad that machinery exists.
I am also fairly certain that no combination of scores and validation rules is ever going to perfectly measure the thing I actually care about.
That tiny moment where you read something and your brain goes:
Wait.
I know this.
What was that?
And then you remember.
That's the clue I'm looking for.