Why the New York Times got into multiplayer games

In the span of about five years, The New York Times games team has fielded roughly 120 pitches for prospective new games. Only four have made the cut.

The very latest, Crossplay, formally released on Jan. 21. In the works for nearly a year and a half — and on the back-burner as an idea for even longer — it is the paper’s first multiplayer title, and the first with a built-in social element: players can chat with one another.

I won’t belabor the explanation: Crossplay is the Old Gray Lady’s spin on Words with Friends. It is very nearly digital Scrabble, presented with the polish and quality-of-life refinements Times games are known for, including a super compelling data-driven postgame review dubbed Cross Bot, which shows you your best and worst plays and offers recommendations and better word choices, designed in collaboration with the team that helped birth The Needle.

“We were looking for something that would be interesting to our audience, but also be interesting to other people and that would sidestep where we are in the puzzle space,” Zoe Bell, executive producer of Games at the Times, told me. “And it felt like a two player word game was the place for us to be.”

Let’s zoom out. The smart media take these days is that the Times is no longer just a media company. Instead, it’s a lifestyle product bundler. There was a viral-in-media-circles graphic circulating recently showing the number of Times subscribers who just get the news product. It’s a dwindling bunch. Per the graph (embedded below) the Times has vastly more bundle subscribers than news-only subscribers; news-only even loses out to other non-news single product subs.¹ Those other subscriptions include a cooking vertical, Wirecutter, The Athletic and games.

There’s no public data on how the games section is performing, but there’s some compelling information on which we can base an educated guess. A Times spokesperson noted in a press release that “in 2025, players around the world solved more than 11 billion puzzles across New York Times Games.” And in our conversation, Bell noted that Wordle scores were being shared at an average rate of 46 per second in 2025. (I did the math: that’s just shy of 4 million Wordle scores shared each day.) Anecdotally — based on the bevy of friends I see playing the Times’ games regularly, and my experience seeing The Washington Post’s traffic dashboard, where less well-known games nevertheless dominate — the section is holding its own.

I spoke earlier this year with Bell and Rohit Crasta, lead game designer at the Times, about the risks inherent to launching a multiplayer game, the strategic benefits of playing words like QIS and JO, and the nerdy tech undergirding Crossplay and its postgame helper daemon, Cross Bot.

The following interview has been edited for length and clarity.²

ReaderGrev: Why multiplayer, and why now? What gap does this fill in the Times’ portfolio?

Zoe Bell: We know that our players are social. Not everybody, some people want to be in their own world, but many players are social. They're sharing their Wordle scores every day — 46 times a second last year on average. So we know that people are in these chats, playing together with friends and family.

We also know that social makes games very, very fun, and makes you want to come back and interact more every day. So for us, it felt like a natural segue to add a word game that was more social than our existing word games.

I have no idea what it takes to build a game or even start thinking about what it takes to build a game. Can you take me into the first couple of weeks of working on Crossplay? What did that process look like?

Bell: Normally, for any new game we make a prototype. For Crossplay it was a little bit different, because we needed to build a whole separate app. That was very, very basic and unrelated to gameplay. We had to spend a lot of time just standing up the app.

And then, you know, we start adding in and you get your first game board in and it's gray and black and white, and all the bonus tiles are almost the same color because it's gray and black and white for now. And then you get color and you start playing around with iterating on the board design and iterating on the tile bag, and suddenly it starts feeling like a real game.

But the first few weeks are really boring because literally, the app doesn't work.

In a lot of the Crossplay sessions I've played, there's a late stage where a lot of the spaces are filled and you're trying to squeeze in a three letter word out of the remnants of your tray. I'm curious if either of you have go-to BS words. I played QIS against a friend recently, and he was very upset, but it's a big scorer. What are your favorite words or strategies?

Rohit Crasta: I generally try to go longer, and I like to also play new words because it's fun for me. But yes, sometimes I have to have those as fallbacks — QI being one of them, or QAT, or JO. Knowing all your two letter words is important because that helps you do parallel plays.

Bell: We have this feature [that] really walks you through where you could have played better. Every time I only play a three letter word or something, it's always like: Well, you could have done better that time. The Upshot, which is our team of data visualization journalists, they came in and helped us. They're the same people who did Wordle Bot, but also they're the same people who do the election results and other things. The data and the strategy that goes into that is incredibly interesting.

So every time I put in a three letter word at the end of the game, I'm like: Oh, Cross Bot's going to tell me I could've done something better — which is great, because it's making me get better at playing the game.

Cross Bot is fantastic. Tell me a bit more about how that all came together.

Crasta: It's evolved, and it'll keep evolving, actually. The cool thing about it is that it's very data driven.

We started with some assumptions about, like, what do we think makes a good player, and that was based on our playtesting with small groups of people. But that was kind of limited because it was based on our own experience, and we were making assumptions about spatial reasoning and hooks and resources management and all the elements of strategy we believed were important. But then we started getting lots of data. We did a soft launch in New Zealand and in Australia, and we started getting lots of real players playing, and that made all the difference because what you could see is data from these completed games, and you got to see trends of what tended to work, what tended to make the difference.

In many ways, we're actually now learning from Cross Bot about how the game works on the deepest level, because we’re learning that sometimes a certain move, unintuitively, can help you win, or help you get better at the game. No two moments or movements are exactly alike, but [Cross Bot] can kind of generalize and be like: Okay, we can play out scenarios and actually model predictions based on what [players] can do next and what's left in the bag and actually try to predict what could happen based on this move choice you make. That's when you get those moments that are like: Okay, maybe you're not scoring the most points with this one move, but you're setting yourself up positionally for a much better scenario down the road.

How you are ranking players? What systems are you using to match players against one another?

Crasta: We have a Elo-like system behind the scenes, which is measuring people in terms of their performance, and it's essentially adjusting as you complete games against real people. It's trying to rubber band you to the right skill level. So the more you play, the more you should be guided toward players at your skill level. But it is ultimately an approximation and people have good days and bad days, and there's a little bit of luck in there too.

One thing I do want to call out is that we're transparent. We are matching you with real people. That sounds simple, but actually we're really proud of it: When you go to our matchmaking system, you're always being matched with a real person who's playing back with you. We're trying to foster a connection between real people, so we don't do any kind of faking bots thing.

Have you considered showing people how they rank or what their Elo is?

Crasta: I think it's risky because it can actually really demotivate someone. Not everyone wants to see that kind of thing.

Bell: The games team recently looked up my Elo and it was lower than I thought it should be, and I was offended.

What are your expectations around chat in Crossplay? Chats often require moderation, and sometimes people are cruel to one another online. I'm curious how you thought about that as you implemented and built that system.

Bell: We thought very, very deeply about chat, about the concerns that players might have. There are a few different things that we're doing. One is that any player can report another player, whether from their profile or from the chat in a game itself, and that'll send a report to our trust and safety team. I think that's going to be what we think is the foremost way to catch bad actors, because we do accept that that is a risk when you have chat. It's the internet, unfortunately. We also have some automated moderation as well.

We have a lovely community on the New York Times games right now — the Spelling Bee community, the forums are just absolutely delightful — and I would love to believe that that's what it'll be here, but we also have some backstops as well.

If there anything that you think is essential to know about your team, your work, Crossplay, anything like that that just hasn't come up so far?³

Crasta: This is our first game to really dive deep into strategy. We have elements of that in our other games, but our other games are puzzles in a purer sense. This is puzzle-like thinking but applied in a more abstract way: You need to manage trade-offs, think about spatial reasoning. It's still a puzzley-like thinking in the moment, but it's applied in a more stretchy kind of way across a long game. It's really, really fun. And so you get to stretch your brain as you dive deeper into the strategy. To some degree, you are getting better at the game not to get better at the game, but to kind of better yourself, which feels really good.

The other side is the connection side. You could definitely come to this game and play our computer. You're welcome to do that. But I think the game really sings with people, and it's a chance to play something with someone you love, which I think is a great way to connect with them in a different kind of way. Maybe you're not always talking all the time, or maybe you don't have much to say to them, but through playing, it might generate conversation. I was just telling a friend that we're launching this game and she told me: My 10-year-old is making plans with my mother to play this game, and they're so excited to finally be able to play a game together like this from far away.

Bell: Though I will note, we’re 13 plus, you’ve got to be [age] 13 plus. [laughs] But you can play with your teenagers, it'll be really fun.

Post scriptum: Your great feedback on my plans for ReaderGrev

Last week, I dedicated a segment of my send to thinking out loud about ReaderGrev’s future and outlining some potential pricing options for a paid version of the newsletter, were such a thing to exist. The feedback I got was super helpful — even if, at times, it complicated my thinking.

Let’s start with the actionable feedback, the most important of which was that people wanted to know more about my post frequency and content plans. I’ve taken a first stab at this at the top of the newsletter, where I’ve outlined some likely upcoming sends (barring major breaking news). More broadly, my goal is to strike a roughly 40:60 balance of “take” (meaning essays, reviews, etc.) to “report” (meaning newsletters in which I break news or publish new information) in this newsletter. I don’t think every send will be for every reader; my tastes aren’t exactly esoteric, but they’re a little all over the place, and that’s one quality of this newsletter I don’t want to sand down. Broadly, I think as a reader you can expect searching, in-depth interviews with people I find interesting, scoops about the video games and esports industries, personal-ish essays on goings-on in the media and hopefully more weird unclassifiable stuff all side by side.

Another more complicated piece of feedback I got was that a number of you read this newsletter for the viewpoint and voice. That’s extremely flattering! (In my mind, I view the comps for this newsletter as one part Read Max and one part Game File.) But breaking news really is the primary driver of new subscriptions; I can trace every major jump in my subscriber count to a reg-walled scoop. My instinct is to pay- or reg-wall all reporting (you may have noticed that this newsletter, which includes reporting, is for subscribers only) or to put up a paywall any time I run more than one newsletter per week. But these are all just early thoughts. Truthfully, my gut instinct wasn’t strategic at all: I sort of pictured the paywall as a way of generating revenue from dedicated subscribers who just want to throw me a few bucks now and again. I’d welcome folks’ feedback on all this.

I’m reproducing the poll I ran in last week’s newsletter below for people who may have missed it. I believe it’ll retain the same answers from before, and any new responses will be additive. If you’ve already responded and want to see updates to the poll, just choose again whatever you previously selected.

Also: Last week, I wrote about two big stories I was working on, and the challenges of figuring out where those stories might live. I’m closer to an answer now — and I’m going to be fully transparent about how things are looking in the interest of giving you all a peek behind the “business of journalism” curtain. One of those pitches has been picked up by a major outlet. I can’t speak to when it’ll run or what it’ll end up looking like, but I’m very excited to have institutional backing for this one, and I’ll be sharing a link whenever that piece goes live. (I also don’t want to overstate things: an accepted pitch is not a total guarantee that a piece will run, so read this all in the conditional mood.) The second story will likely run in this newsletter, and will almost certainly cost in the low four figure range to publish. The expected freelance rate for the first story should help offset costs, but probably not even by half. This will all make more sense when that second story hits your inboxes, but in this case, beyond just being accurate and fair, I’d also like to be legally buttoned up.

One last thing…

I wanted to link to some good stuff I read recently in an effort to make this feel more like a full-service newsletter — but also to maybe start rebuilding something of a cross-linking community of newsletters and outlets. Maybe, just maybe, the games writing space can start to feel like, well, a space again. (I’ll also be including some non-games nods).

• The Secretive Company Filling Video Game Sites With Gambling And AI — Aftermath | Probably the most essential thing I read this week, a sneak peek into the nightmare of games publishing and low-tier freelance labor.

• Perfect Tides at Four — Meredith Gran | Chris Plante shared this short blog post on Bluesky. It’s delightful: “If a game is slow to find its audience in the weeks following release, there is the sense that it’s dead and buried, a casualty of the landscape. Why do we allow ourselves to think and focus our discussions in these terms? If and when (and I do think it’s ‘when’) a game of any uniqueness finds its audience, it will be loved. Explored voraciously. Obsessed over.”

• Critical Distance, the games writing roundup site, announced on social media that it was seeking a boost on Patreon to be “truly sustainable.” They’ve already cleared that goal, but I want to boost their post just to point people in their direction. They do great work and are worth supporting. I’m a long-time patron!

• One key chart showing the game industry's struggle — Game File | A heaping of bad news about the video game industry, and a sprinkle of good.

• Reporters Seek Comment. What Happens Next May Surprise You. — The New York Times | Love these behind the scenes looks!

• How Anonymous Bettors Cashed In on the Iran Strike, Just Hours Before It Happened — The New York Times | Yeah, yeah, a lot of the Times in this newsletter. I just can’t get over this report, which feels like it would have been a bombshell under any other administration.

• I also just finished Very Good, Jeeves, a collection of short stories about the titular valet and his master Bertie Wooster. Probably the funniest thing I’ve read in ages. P. G. Wodehouse drops some unbelievable turns of phrase, such as this one, used to describe a character caught in an embarrassing moment: “I saw my Aunt Agatha for what she was — not, as I had long imagined, a sort of man-eating fish at the very mention of whose name strong men quivered like aspens, but a poor goop who had just dropped a very serious brick.” It will have a permanent effect on my vocabulary.

That’s all from me today. As always, appreciate you reading. Cheers.

Thanks for reading ReaderGrev! Consider sharing it with a friend, on Discord, Twitter, LinkedIn, or even a subreddit where folks might appreciate it. Word of mouth helps this newsletter grow!

If you have a tip, I can be reached on Bluesky, on Twitter or via email at mikhail (at) readergrev (dot) com.

1  I don’t want to overstate things here. My guess is that the bulk of bundled subscriptions include the news product. The chart is not necessarily indicative of a decline in news readership.

2  For transparency’s sake, my changes are largely focused on cleaning up “ums,” “likes” and “ahs,” whittling down questions to let you get to the answers faster, and cutting certain parts of answers (or entire exchanges) that are redundant or irrelevant or which make sense over audio but not over text. My goal is never to change the meaning of what’s been said to me, and I outline my approach to editing these transcripts to interview subjects upfront.

3  I typically don’t include this question in the transcripts to my interviews, but I really liked Crasta’s answer.