Behind Video Call Glitches

unraveling the technology behind Video Calls

I was catching up with an old friend from college on a video call the other night. We were laughing about something when the screen did that familiar glitch. His face froze mid-smile, the audio turned into robotic garble, and then, a heartbeat later, everything snapped back to normal as if nothing had happened.

We usually just shrug it off. Blame the WiFi or say “My internet is bad uwu.” and move on with the conversation.

But this time, I found myself staring at my camera, wondering, what magic is happening that the friend who is miles away from me, his face is being projected in front of me, exactly my phone and not to the millions of other devices nearby? How does any of this actually work?

To understand how that frozen glitch happens and why it doesn’t last forever, I went for a side quest and understood that magic. Here are my notes for how the internet has to solve three fundamental problems every time you make a call, send a message, or load a webpage.

Problem One: Speed vs Reliability

When you send anything across the internet, whether it’s a heartfelt email to a friend or a silly selfie to your sibling or your partner/spouse, whatever, it doesn’t travel as one solid piece. Instead, the internet breaks it into thousands of tiny fragments called packets, each one a digital envelope carrying a small piece of your message. 📦

But here’s where it gets interesting. Before those packets leave your device, the internet has to make a fundamental choice about how to send them. It’s a bit like choosing between two very different shipping companies, each with its own philosophy about what matters most.

The first option is TCP (Transmission Control Protocol). Imagine TCP as that meticulous mail courier who treats every delivery like it’s carrying the Crown Jewels. This courier checks off every package on his list, ensures they arrive in perfect numerical order, and if even one goes missing, he stops everything, backtracks, hunts down the lost package, and redelivers it before continuing with the rest.

This makes TCP perfect for things like downloading files, loading web pages, or sending emails. When you’re downloading a recipe, you definitely don’t want step five to arrive before step one, and you certainly don’t want the ingredient list to go missing entirely.

Now, the second option is UDP (User Datagram Protocol). Think of UDP as a courier racing through the city on a motorcycle, running late for a dozen different deliveries. As he speeds past your house, he hurls packages onto your porch without slowing down. He doesn’t check if you caught them. He doesn’t verify that they arrived in order. He just keeps moving.

This sounds reckless, but it’s actually brilliant for video calls. Here’s why: If we used TCP for our video chat, every time a single frame of video got lost in transmission, your entire conversation would freeze while computers exchanged messages to resend that frame. By the time it arrived, we’d be seconds behind in the conversation.

With UDP, when a packet goes missing, the system shrugs and moves on. That brief glitch I saw on my video call, my friend’s face freezing for a split second, that was UDP in action. A few packets got lost somewhere between his device and mine. But instead of grinding to a halt, my computer simply skipped those frames and played the next ones that did arrive. In real-time communication, “now” will always matter more than “perfect.”

Problem Two: Running Out of Addresses

Now that we understand how packets get sent quickly, we need to tackle an even more fundamental question: where exactly do those packets go? How does the internet know which device is yours among the billions online at any given moment?

This is where IP addresses come in. Every single device connected to the internet needs its own unique identifier, just like every house on a street needs its own address for the mail carrier to find it. For most of the internet’s history, we used a system called IPv4, which generates addresses that look like this: 192.168.1.1. You’ve probably seen numbers like this before in your router settings or network preferences.

The problem is that IPv4 uses 32-bit numbers, which means it can only create about 4.3 billion unique addresses. When computer scientists designed this system back in the 1970s, 4.3 billion seemed impossibly large. The idea that we’d ever run out felt laughable.

🎈Fun fact: Your device doesn’t have a permanent IP address

They couldn’t have predicted that by 2025, the average household would have a dozen internet-connected devices. Your laptop, your phone, your tablet, your smart TV, your gaming console, your doorbell camera, your thermostat, your refrigerator, each one needs its own unique address. Without a workaround, half these devices simply wouldn’t be able to connect to the internet at all.

The long-term solution is IPv6, which uses 128-bit addresses. This gives us such an astronomically large number of possible addresses that we could assign one to every grain of sand on Earth and still have plenty left over. But upgrading the entire internet to IPv6 is like trying to replace every road in the world while cars are still driving. It takes decades.

In the meantime, we still have billions of devices that need to get online right now. This shortage forced engineers to invent a clever workaround, one that you use every single day without knowing it.

Problem Three: Getting Through Your Router

This is the part that usually trips people up, but once you see the metaphor, it all clicks into place. It’s the reason you can connect your phone, your laptop, and your smart TV to the same WiFi network, even though your internet service provider only gave your house one public IP address.

Picture your home WiFi router as the front desk of a large apartment building. The entire building has one official street address that’s visible to the outside world, let’s say 742 Evergreen Terrace. This is your public IP address, the one your internet service provider assigned to you, and it’s the only address the broader internet can see.

Inside the building, however, there are dozens of individual apartments, each with its own unit number: Apartment 1, Apartment 2, Apartment 3, and so on. These apartment numbers are your private IP addresses, the numbers starting with 192.168 that your router assigns to each device in your home. Your phone might be Apartment 1, your laptop might be Apartment 2, and your smart TV might be Apartment 3.

This system is called Network Address Translation, or NAT for short, and it’s like having a very organized receptionist at the front desk who keeps track of who ordered what.

When you decide to browse the web on your laptop, here’s what actually happens: Your laptop (Apartment 2) wants to visit Google, so it sends a request to the router at the front desk. The router, being meticulous, makes a note in its logbook: “Apartment 2 asked for Google.com at 3:42 PM.” Then the router steps outside, uses the building’s official street address to contact Google, receives the response, checks its logbook to remember which apartment asked for this, and delivers the data back to your laptop in Apartment 2.

This works beautifully for normal web browsing because you’re the one initiating the conversation. The router always knows you’re expecting a response because you asked for it in the first place.

But here’s where video calls get tricky. In a video call, someone from the outside world your friend across the country, is trying to reach you directly. They’re trying to call Apartment 2 specifically, but they only know the building’s street address. Their packets arrive at your router and essentially say, “Hi! I have a video stream here! Who should I give this to?”

Your router looks at these incoming packets with suspicion. Nobody inside the building requested this video stream. The router checks its logbook and finds no record of anyone asking for this data. As far as the router knows, this could be spam, or worse, a malicious attack trying to break into the network. So the router’s firewall, its security system, makes a safe decision and blocks the connection entirely.

This is the paradox we’ve created. We have a fast delivery system with UDP, we have a working address system with IP addresses and NAT, but we’ve accidentally built a wall around ourselves. Two people, both sitting behind their own protective routers in their own apartment buildings, can’t figure out how to find each other to start a conversation.

So how do video calling apps like Zoom or FaceTime actually work? How do they pierce through these walls and connect two devices that are both hidden behind NAT routers?

That’s where the story gets even more interesting. The solution involves something engineers call “hole punching,” a technique that’s both elegant and slightly devious in how it exploits the very system designed to keep us safe. And it’s the reason your frozen glitch moment doesn’t last forever……

And that’s a wrap for today, and before I say goodbye for today, here’s a quote by Jeanette Winterson I’ve been pondering,

"What you risk reveals what you value."

Please don’t forget to share it with your friends, family, and strangers.

Have a Great Day 💖