The Insane Design Of Messaging

0:00This is the message you sent this morning on WhatsApp, and these two blue ticks means you got left on read, and the other person clearly ignored you. See, this blue tick is the read receipt, and there's an equivalent of this in most of the chat apps you use every day. And when they shipped this feature in 2014, they thought they were just adding something cool and useful. So, they just made it the default. Everyone had a read receipt, and users went crazy. Suddenly, you could immediately see how everyone else is ignoring you, when and for how long. And this caused so much backlash that 10 days later WhatsApp added an

0:32option to hide read receipts. But, this is where things get interesting. On iMessage, you can hide your own, so other people can't see when you read something. But, when you do that, you can still see when others read the messages you sent. [music] And this creates a sort of unfair advantage. So, when the WhatsApp team made the hiding an option, they made it so it actually goes both ways. You're also not going to see when the other person read the message. And this is because if they don't, then everyone would try and get an advantage and turn off their own read receipts with no consequence and just

1:03lead the feature just not being used at all. WeChat, for example, chose something different. They straight-up don't have any read receipts, and that's by design. They even mention it in this internal lecture. If your boss messages you, you read it, and you don't reply, it's trouble. This one seemingly simple design decision, a tiny blue tick, and how an option in the settings worked behind the scenes ends up having a massive effect on the behaviors of billions of users. The anxiety of chatting and seeing who replied. And most importantly, the design of this one little tick might be the one thing that makes millions of people choose your app

1:36versus another. See, chat apps on the surface look all exactly the same. Some chat bubbles, a box to write in, I can send audio, maybe some images. But, in reality, every tiny decision, from how bubbles are positioned to what the enter key does, how delete actually works, is carefully designed to influence your behavior. It's the reason why one chat fails, while another might become the default in a certain country. And once you see what's actually hiding behind the way you communicate every single day, well I'm sorry, but you can't unsee it. We spent the last month diving into this rabbit hole. Somehow even ended up on a call with the guy that invented the

2:11chat interface that everyone on Earth uses every single day. At the time I thought that was a good idea, and then we implemented it. On this show we go behind the scenes of the insane design, engineering, and the secrets behind the tech products we use every day. There's actually three layers we need to understand this, and only the first one is what you can actually see. Take a look at this. This is Facebook Messenger, and they had a problem. The team noticed that many people were sending the same message, "K" to confirm something. So much that "K" is now associated as being something kind of rude to say. But back in 2014, it was

2:44just a way to confirm something. But it was already becoming a problem. Even TechCrunch said so. When you have something typed, the send button is just a regular send button. But when the text box is empty, there is the thumbs up button. They also let you use it to react to things. Today this seems quite normal, and almost all chat apps have this. But reactions like these exist because pure

3:03text communication got messy and failed. And this is just the tip of the iceberg. If you're in WhatsApp, iMessage, or basically any mainstream chat app, on desktop when you press enter, it sends the message. Seems pretty obvious, no? But when you're writing an email, pressing enter creates a new line. And if you want to create a new line in traditional messaging apps, you have to press shift plus enter. This seems like a relatively simple and innocent thing, but it leads to your behavior and the messages you send being wildly different. This one tiny difference in the behavior of the enter key is the reason why emails look like

3:36this, long with many separated paragraphs, salutations. Whereas in chat apps, you can still use bullet points, bold, italic, but writing like this has been designed to be extremely tedious. To create new lines, you have to use shift enter, and by mistake you press enter, so instead the message gets sent. So that's why our emails look like this, but the messages we send every day look like this. But, what happens when your product is exactly in the middle between chat and email? This is exactly the problem that Slack has. It's a work communication app. So, what should the default be? Well, they ran AB test and unsurprisingly enter to send the most

4:11immediate one, but they added a setting to let people choose whether they want one or the other. I'm a fan of shift enter. But, take a look at this. By having all these formatting options on the top always visible at all times, they're also conveying you the idea that this is a formal communication that's meant to be more formal than regular messaging. But, if you have some code block in a Slack message, they built it so the behavior actually flips. Now, inside here when I press enter it goes to a new line because of this is how you navigate code basically everywhere. And that consistency is more important than

4:41sending immediately all the time. Now, I know this might seem silly. I mean, how can the behavior of the enter key or shift make any difference? Well, because chat apps are weird. They're not about the app themselves, it's about the messages that people send in these apps. And as it turns out, you can influence that. And the people building these products realize that all these nuances about how their chat works have a huge impact into what is actually used for by people. And eventually to whether they choose their product versus another. For example, if you're building a mainstream chat app like iMessage, having something like this creates a messaging behavior

5:16that doesn't match. It would make it fail instantly. Yes, these are all things you can do on iMessage like bold, italic, underline, strike through. But, notice that having them always here influences you to write in a much more structured and formal way. Or having a text box that is two lines tall already influences you to typing longer multi-paragraph messages. But, in Slack's case, this is perfect because they actually want people to use Slack to send important and well-formatted messages, but still be closer to a chat than something like email. These subtle differences are what ends up making a difference between a successful billion-dollar product like Slack and

5:51being another one of the thousands failed chat apps out there. Or take Instagram DMs. It's kind of weird because it's technically a chat app. It has everything iMessage has, messaging, emojis, images. So, why do you only ever use it to send Reels back and forth to your friends? Well, because while it looks like a chat app, it was actually designed to be something completely different. See, under every single post and every Reel, there's this, the paper plane share icon. Tap it and you don't go to chat, you get all your friends' faces. Tap one, two, three of them and just send it off. When you are in the

6:22actual chat, there is a permanent button visible next to every single Reel to share it with someone else. The Instagram chat has been designed from the ground up to actually not be a chat, but a sharing machine to feed valuable data to the algorithm. And the head of Instagram said himself, One of the most important signals we use

6:39in ranking is sends per reach. way, this is why they moved the messages button from the top right, where traditionally chats into other products exist, to the main navbar. And people were pissed about [music] this. You used to just swipe over on Instagram and that'll be your inbox. Now you swipe over, it's just Reels. But the reason they did it is that for Instagram, chat is not a chat meant to message. It's a way to share their main product, which is Reels, and gather data for their algorithm. But now, WhatsApp has the same button, forward. But take this message and try to forward it to 20

7:12people. You can't. You can only share something up to five people at a time. When a message has been forwarded many times, there is a badge that highlights that specific thing. And there's a limit to the amount of times an individual message can be forwarded. And this is because WhatsApp is built to be an actual everyday messaging app. In their case, sharing is often associated with spam, scams, and those message chains that boomers send to each other. They built their product to optimize for something completely different. But things can go even deeper. Take these two. These look extremely similar. Have channels, group messages that look like

7:48this, threads, DMs, sidebar organization, huddles. But while these two products on the surface look the same, 16-year-olds playing Minecraft will never download Slack to coordinate their redstone builds. A large company will almost never use Discord to communicate at work. And when you start to look really close, that's when you see why. Reply to a message on Discord and it lands right here in the chat with a little quote on top. Reply on Slack and your message disappears. It goes into a thread off the main screen. And Slack asks you a question every single time. Also send to channel. So on Slack, every reply is a decision. So the main

8:22channel stays clean for whoever reads it tomorrow. On Discord, conversation is one messy stream and the mess is the point. Discord does have threads, but you have to make one on purpose and if nobody types there for 3 days, it closes itself. Whereas a Slack thread never closes. It sits there, searchable forever. Now, click here on names. Slack gives you two states, active or away, and it decides for you. 10 minutes without touching the keyboard and you're away. Discord gives you four and the last one is invisible. You'll appear offline entirely. And by default, Discord tells your friends what game you're playing or what you're listening

8:55to on Spotify. Slack would never ship that. At work, what is this person doing right now is not a fun feature. It's seen as surveillance. So it's kept at a minimum so users don't feel looked at. But it's also not something the user decides. It's based on activity because in the end, that's what matters when selling to companies. The green dot on Discord is there to tell your friends what you're playing and the green dot in Slack is there for your boss to know who's there. See, Slack made every choice so the message can be found later. Can be read by agents that act as a sort of company vault. Discord is

9:25instead designed for a present moment to give the feeling that there's just a room that you walked into. But what's interesting is not just how every single chat messaging app is designed to influence and drive your behavior. But as we saw, on the surface, they all look basically the same. But how we got to

9:42that is probably even more interesting. And in this rabbit hole, we even ended up on a chat with the guy from Apple that invented this whole chat UI itself. But before we get to that, let me show you what me and my team are doing with Granola, the sponsor of this video. See, we are building a harness, a complex set of skills and instructions that help us doing the long and complex process of

10:01researching for these very videos. Problem is that this is a very complex process with multiple people. It's always been hard for an AI to understand. I mean, what are we really doing here? It takes like a lot of hours to do this. And this is because it's something that happens on a call between me and my producers. We are on a call for hours every single week. But take a look at this. We're using Granola to record our calls. So, not only you can just take some random [music] notes during the call, and it enhances them

10:25with the real context of the meeting. But through Granola's MCP, for example, Claude, our agent, can read all the calls. So, we're just feeding it hours and hours of meetings automatically recorded in Granola to have it understand how our process works and just learn. This was the foundation for building complex research, fact-checking, drafting, and ideation skills that works exactly for how our team works, and they are completely custom. The Granola MCP plugs into Claude or Codex or any agent you use right where you already work, so you can ask things like, "What did we decide about the intro last week?" and get the answer from the actual call. Granola's

11:00giving new users 100% off their first month with the link in the description. So, you can try it in your next call. And now, back to the video. Because take a look at these. WhatsApp, iMessage, Messenger, Telegram, they all look the same. Bubbles on the left for other people, yours on the right, a box at the bottom, everything scrolls up. Different companies, different countries, and they

11:18all landed somehow on this exact design. This is Jens Alfke, and he's the inventor of the chat, the UI. This UI that me, you, billions of people have been using every single day. And diving into this rabbit hole, we actually ended up learning how the chat UI we use everyday was actually born. That goes back to comic books, right? If you've got a panel with two people facing each other, then one of them's going to have their chat balloon going to the left, and the other one's going to have their chat balloon going to the right. I mean, if you've used kind of a a terminal-based IRC client, that's

11:50pretty much the state of the art of the way things worked before. I was really interested in MUDs. Uh I don't know if you know what those are, the multi-user dungeons. They're like a a cross between a a you know, text adventure like Zork and uh and a and a RPG. This is like '97 that I started really thinking about this stuff. Because I was thinking about MUDs, I was like, "How do you deal with

12:16chats with many people in them?" But, when you start to put 10 people in a terminal, watch what happens. It starts to become a complicated, garbled mess. It's like a Twitch chat from today, which is built to be minimal over the stream, but not really readable and understandable. So, in 1996 at Apple, when they were building the first real chat app called iChat, that's when this guy, Jens Alfke, who was spending his evenings in chat rooms and message boards, was having a really hard time keeping track of many-way chats. And

12:43that's when he drew this. One of the most difficult things at the time was getting the chat bubbles to draw in behind the text. The comic strips, obviously, were the inspiration for that. And then the tail on it, the direction of the tail sort of can indicate who is talking. You know, it's basically like drawing a round wreck

13:02with a triangle sticking out. He thought, "Well, what if I keep the vertical scrolling, but instead of isolating your messages with color or something else, I just align them to the right?" And not only that, why don't I take inspiration from comics and give each message a comic-style balloon around it to visually break up messages. And then I can use color to differentiate who's typing. And this simple idea is where basically all chat we use today was born, including basically all the AI tools and agents we

13:28use everyday, it all came from here. But they were much more literal about it. Like they literally had some comic book artists draw panels with like blank balloons. You know, it didn't correspond to who the people were at all. It was just kind of a visual gimmick. novelty. So, it was a novelty, right? But not all the experiment really landed. For example, they even tried this. Basically, whenever you started typing something, not only did the other person had a typing indicator, but they also got a real-time preview of

13:59everything you were typing in real time. And what this is representing is an a message in progress. Yeah, everybody sees that. You know, and at the time I thought that was a good idea. And then we implemented it. We had that as a as a preference. Um, and people didn't like it. It gives you this feeling like you're being watched. And everybody gets to see every

14:20typo that you make. Or you can do tricks like, you know, write something inflammatory and then backspace over it so it never happened. Most users found it annoying to have other people seeing every typo they made. So, they just turned it off. But not all went to waste because what we have left today is this, the typing indicator we use and we see every single day. And by the way, Apple still has a patent on this. Now, all of this was still on desktop because before 2007, most of SMS worked not like a conversation with someone, more like

14:51this, a pile of last received messages. Kind of like email but with no threads. And then, with the first iPhone came the SMS app, which when they had to build it, where do they take inspiration from? Of course, our balloons in iChat. Two years later, WhatsApp launches and look, it's the same thing. Then Telegram,

15:09Messenger, WeChat, Kakao. How does it feel to build this evergreen building block that everyone else is doing it? Yeah, it took me a while to realize that, you know, this was the way that everyone was doing it now. And it's it's nice, you know? It's like I've always wanted to be able to change

15:27the world in some way. This is one of the most important principles in building products, Jacob's law. People actually spend most of their time not on your product, but likely on other products. So, they prefer your product to work the same way as all the others they already know. See, out there there's probably a better design for representing chats than this, but people are so used to it that it will be almost impossible to change their behavior unless something completely new and groundbreaking comes along. This is why most product categories end up converging on being slight variations of the same thing. From this to all becoming black rectangles. This is why

16:02all streaming apps looks the same. But, even though we converge to this, we've just realized how powerful and universal this is. And we start to morph it in weird new ways. For example, Figma is a design tool where the core idea is that you see other people's cursor. You can move stuff around together. The next logical step would be to build a chat into it. Now, the most obvious way of doing it would be this. I mean, it's obvious. It's literally what we just saw. But, instead, they did this. Since the cursor is the main element that represents a person, they're using it to let you just type, and the text appears

16:35next to your cursor. There's no send. There's no multiple messages. There's no history. And that's because of this chat, it's proximity-based and ephemeral. Only the people that are there at the time close your cursor see it. But, if you are already working together, that's all you need. It's a digital version of just being in the same room and telling something to the person next to you. Or maybe it's a different version of that old Apple pattern that we saw. Same thing with Uber. Of course, you want to chat with your driver, and the obvious way to build it is a normal chat. But, their problem is that drivers were driving,

17:07and they couldn't really chat all the time. So, instead, they built this. Your message gets read out loud, and the only way they can answer is one tap, a thumbs up. And you still get delivered and read because the only thing riders kept asking was, "Did the driver read my message?" But there's something else

17:23hidden behind a tech you use every day. Because the best products are secretly built to be exactly like games. They actually use the same design principles to make you learn and keep playing. This is already happening in the product you use every day. And you can learn all about it in this video right here. I'm Américo and I'll see you in the next

17:41episode of Behind the Product. See you.