Why The First Computers Were Made Out Of Light Bulbs
- [Derek] The modern era of electronics began with the light bulb but not in the way you might think. Early light bulbs consisted of a carbon filament sealed inside a glass bulb with a vacuum inside. When a potential difference was applied across the filament, current flowed through it, heating it up to over 2000 Kelvin, so hot that it glowed. If there had been much oxygen in the bulb, the filament would've burned immediately. That was the reason for the vacuum.
But from the perspective of electronics, the real breakthrough came from a curious observation made by Thomas Edison. He saw that over a bulb's lifetime, the glass became discolored, turning yellow, and then brown, but only on one side. So what was going on? Well, the heated filament emits not only light and heat but also electrons. You can think of them being boiled off the surface of the carbon. This phenomenon, known as thermionic emission, had twice been discovered independently by other scientists up to 27 years earlier. But after Edison, it became widely known.
In fact, for a time, the emission of electrons off a hot filament was called the Edison effect. Now, these electrons floating around were unobstructed because they're in a vacuum. But since there was a potential difference across the wires that led to the filament, the electrons were attracted to the positive wire. So they accelerated towards it, and most would whizz straight past it and crash into the glass, over time discoloring it only on the positive side. I should note that Edison was using DC electricity. If he had been using AC, then both sides would've been discolored.
But it was this observation that set the scene for an electronics revolution, and eventually the first digital computers. In 1904, John Ambrose Fleming patented a device that was very similar to Edison's light bulb, but with one important addition: a second electrode in the bulb. By charging this plate positively with respect to the filament, electrons could be accelerated across the gap, completing the circuit. But if the plate were slightly negative relative to the filament, then it would repel electrons and no current would flow.
Fleming called his device a one-way street for electricity. Since only one of the electrodes was heated, electrons could only flow from there to the plate and not the other way around. The device was called a thermionic diode, and it was used initially for detecting radio signals, but it could also convert alternating current to direct current. Scientists quickly realized that a more efficient design had the filament in the center and the other electrode to the plate or anode as a cylinder surrounding it. This geometry captured more of the electrons coming off the filament and allowed for larger currents to flow.
Oh, there she goes.
[Derek] With just one of these diodes, you could convert AC into a bumpy kind of DC. But combining a few diodes and a capacitor led to a fairly steady direct current, and this was a big deal. It was the first practical vacuum tube device and the model for all vacuum tubes that would dominate the industry for the next half century.
In the early 1900s, the big problem in electronics was amplification. Radio had just been invented but its range was limited by the lack of reliable equipment that could boost weak signals. Similarly, telephone calls were limited to at most 1300 kilometers because by that point, the signal was too faint to hear. A rudimentary form of amplification had been built for telegraph operation called the relay. In a relay, there is an electromagnet. And when current flows through that electromagnet, it attracts a switch, turning on a second circuit.
But when the current through the electromagnet stops, the switch is released and the second circuit is open again. This device works well for amplifying the dots and dashes of Morse code along a telegraph line, but its binary output means it's incapable of amplifying the complex and analog signals of phone calls and radio waves. And that's why it was such a breakthrough in 1906 when Lee de Forest took the diode and added another electrode into the bulb.
This electrode wasn't a solid piece of metal but rather a sparse wire mesh, and it was positioned in between the filament or cathode and the anode. With three electrodes, it was called a triode. Now, a large potential difference could be applied across the anode and cathode, but the number of electrons that actually flowed between them was controlled by the voltage on the grid, as this new electrode was known. If the grid were slightly negatively charged, it repelled electrons from the filament so that none could flow through to the anode.
But if the grid were slightly positive, then electrons were attracted away from the filament, and most of them would pass through the holes in the grid, and then accelerate to the anode. So in this way, a small change in voltage on the grid can control a huge voltage at the anode and the response is rapid. So you can get high frequency amplification. I like to think of it as standing at the top of a high cliff, opening and closing a valve on a big water pipe. I mean, it doesn't take much energy to turn the valve, but that small input is converted into a huge output of water falling down the cliff.
[Derek] You're powering up this track here.
[David] Warming up, you can see it warming up there.
[Derek] So the yellow is the input?
[David] Yellow is the input. Purple is the output. We have essentially a two-volt change on the input giving us, what is this? Five volts, so five, 10, 15-volt change on the output.
[Derek] For this demonstration, we were only using 24 volts on the anode. If we had used a higher voltage, we could have got a lot more amplification and people did. This was the device that allowed us to call long distance for the first time. Using vacuum tubes, the first transcontinental call from New York to San Francisco took place on the 25th of January, 1915.
[Derek] Whoa.
[David] Yeah, there we go, should be 10 volts.
[Derek] It's hard to see the grid here because just like with the cylindrical diode, the best configuration for a triode is to have a cylindrical configuration. The anode is on the outside, the grid is cylindrically inside that, and the cathode or filament is in the center.
The invention of the triode was incredibly important. Radios, TVs, whatever electronics people had were powered by vacuum tubes. You would've had so many in your house even up until the 1960s and '70s. But vacuum tubes weren't done revolutionizing electronics. In his 1937 thesis, Claude Shannon found a connection between electric circuits and a branch of mathematics known as Boolean algebra.
Working in the mid-1800s, George Boole was trying to find a mathematical foundation for logic. Under his system, a true statement was represented as a 1 and a false statement as a 0. And Boole also developed a few operations like AND. If both statements A and B were true, then the output would also be true. What Shannon realized is that Boole's operations could be represented as electronic circuits, that there was an equivalence between mathematical statements and electric circuits.
All you needed to realize these circuits in the real world were a couple switches. That same year, 1937, George Stibitz built the first digital calculator. It could add two 1-bit binary numbers. That is, it could add two numbers so long as they were either 0 or 1. The calculator worked using a relay, that electromechanical switch from telegraphy. There were two inputs. If they were left open, the input was 0. If closed, it was a 1. The output was shown with two light bulbs. If no lights were on, the answer was 0. If the output light was on, the answer was 1. And if the carry light was on, the answer was a 2.
The circuit diagram works like this. If neither switch A or B is closed, so adding 0 + 0, then no current flows through the circuit and no light bulb would light up. But if input A was closed, the current would flow through the solenoid, creating a magnetic field that pulls the switch inside it closed, and this connects the output light bulb to power and disconnects the carry light bulb. So the output light bulb lights up, meaning that the answer is 1. And the same thing would happen when input B was closed and A was open.
But if you closed both A and B simultaneously, then there is no current flowing through the solenoid, but there would be a current flowing from the battery connected to A, which is connected to the carry light bulb. So it lights up indicating 1 + 1 equals 2. This is the beginning of the digital age, and no, it was not glamorous. Stibitz built his device out of a few batteries, light bulbs, and relays he had lying around. And for the inputs, he cut up a tin of tobacco. He built it in one night at his kitchen table, which is why it became known as the Model K.
The circuit that Stibitz built is now called a half adder. But if you look at the circuit through the eyes of Claude Shannon, you realize it's actually a pair of logic gates. The output light bulb should turn on when either A or B, but not both, are closed, so this is known as an exclusive OR gate. Whereas the carry light bulb should only turn on when both A and B are closed, so this is an AND gate.
This circuit uses electrical versions of Boolean operators, XOR and AND. And it's possible to build other Boolean operators as electronic gates for things like OR, NOR, and NAND. And you might say, what's the big deal? I mean, the big deal is that you just tricked a bunch of electrons into doing math for you. Sure, it's very simple math, but you could connect a bunch of these half adders together and build more and more complicated circuits that could do more complex math, which is exactly what Stibitz and his colleagues at Bell Labs did.
Two years later, they built the model I, which had more than 400 relays and could add two eight-digit numbers together in a tenth of a second. It could also multiply eight-digit numbers and do multiplication of complex numbers, though these more complicated operations took longer, about a minute per calculation.
So you put a voltage through a coil and it'll turn that switch on or off. So you've got your two operands here. And if you wanna add two numbers together, so 2 is this, 3 is this, right? So 1, 0 is 2.
[Derek] Yeah.
[David] And 1, 1 is 3. And so then when you want to calculate it, you just hit the go button down here. We get 101.
[Derek] I love the sound of that. That is amazing.
It's magical, so if you wanna do like say, 8 plus, that would be 4 so 8 + 8, I think, right?
[Derek] Okay.
[David] Yeah, 8 + 8, that would be, 16, no, it clears it on its own.
[Derek] Okay.
[David] There you go. Eight plus eight is 16 in binary, which would be 1 0 0 0 0. (machine clacking) This is essentially a 1-bit arithmetic unit. There's no logic functions. It only ever does add.
Now, let's say we wanna do 5 - 2, the answer's gonna be 3. So we flip this little switch here that lets me know that I'm doing a subtraction operation, and we do subtraction by doing 2's complement. So essentially what we're doing is we're inverting one of the operands and we're adding one. So now when I run it, (machine clacking) we can see 5 - 2 is equal to 3, so 2 and 1 is 3. Now, because of the way that we do this, the final carry flag ends up getting illuminated here on the end. But if we know that we're doing a subtraction operation, we know that this final carry flag is never going to be on otherwise.
[Derek] Over the next 10 years, they built six more computers based on relays which were used by the US military and the National Advisory Committee for Astronautics or NACA, which later would become NASA. But even by the early 1940s, it was clear that the mechanical nature of the relay, the physical closing and opening of the switches, was too slow to be the future of computers, and they were also prone to breaking.
Anytime you have something that's mechanical, it's gonna wear down. Every time that relay switches, there's a little bit of friction on the rotation point inside of there and there's contacts that are making and breaking electrical connections, and those are gonna wear out.
[Derek] And all the relays opening and closing meant that the computers were incredibly loud. (machine clacking)
So it doesn't really work in a business environment all that well. You can't really stuff it into your office that you're gonna drive people insane.
[Derek] So what computer scientists really needed was an electronic switch and that is where the vacuum tube triode comes in. Whoa! I mean, sure it can work as an amplifier if you put slightly positive or negative voltages on the grid, but it can also work as a switch. If the grid voltage is very negative, then no current flows. And if the grid voltage is very positive, then maximum current flows.
So a triode can be controlled using no moving parts. Just a voltage will set it to be either a 0 or a 1. And best of all, switching between the two can be done rapidly with no noise since you're just controlling electrons zipping around in a vacuum. This is the invention that took computing to the next level. The world's first electronic programmable computer was called ENIAC and it came online for the first time on December 10th, 1945.
It took up a whole room, weighed 30 tons, and used 175 kilowatts of power. So much that it led to a rumor that every time it turned on, the lights in Philadelphia, where ENIAC was located, would dim. Now, that was just a rumor, but mainly because ENIAC had its own dedicated electrical generator to keep up with the enormous power draw. Unlike previous computers, ENIAC wasn't limited to just solving one type of mathematical problem.
It could be programmed, and it was fast, completing 500 operations per second. At the time, the word computers still referred to people doing calculations with pen and paper. So 500 operations per second was really fast. ENIAC's flexibility and power was immediately useful for the development of the hydrogen bomb. The computations needed were so complex that the director of Los Alamos at the time said that, "It would've been impossible to arrive at any solution without the aid of ENIAC."
This is a hilarious part about having a processor that is 1 meter tall and 70 centimeters wide, is that you can point at actual parts of the processor.
[Derek] This is what a 1-bit vacuum tube computer looks like.
[David] Can you feel the heat coming off of it?
[Derek] I certainly can.
[David] I can feel the heat coming off.
[Derek] It's getting warm.
Well, I mean, 190 vacuum tubes is a lot. I think we figured it out. This is pulling like 350 or 400 watts of power or something like that, which is absurd. At night, it's awesome. It looks like a city.
- [Derek] But there were also major flaws with vacuum tubes. The filaments always needed to be heated, so they used a lot of power even when idle, and they were big. It was hard to make a glass vacuum tube with complex electrodes inside arbitrarily small. They were also unreliable. On average, a vacuum tube in ENIAC broke down every few days. And then, it needed to be found and replaced. The longest that ENIAC operated for without failure was just 116 hours.
The first digital computers ran on glorified light bulbs. That is why they were so big, power-hungry, and unreliable. The miracle, and what has made our modern lives possible, is that someone figured out how to perform the same trick with electrons inside a solid piece of material, in silicon. But that's a story for another day.
If you wanna learn about how modern computing devices store and access information, I highly recommend you check out this video's sponsor brilliant.org. They have courses on everything from computer science fundamentals to neural networks to quantum computing. Their latest interactive course is called How Technology Works. It includes a lesson on computer memory where you can get hands-on with the kinds of computational logic gates we discussed in this video.
It also has a lesson that delves into the ingenious ways video compression makes modern-day streaming possible, so you can watch videos like this one. Ever wondered how your phone knows where you are? There's a lesson on that too. There's even a lesson on password theft where you can walk in the footsteps of a hacker to discover why some passwords are stronger than others.
You can try all these out for free by going to brilliant.org/veritasium. You know, whether it's getting a working knowledge of modern-day technology or building skills in math, computer science or data science, Brilliant is the best way to learn. They have thousands of bite-sized interactive lessons that not only help master key concepts but connect what you learn to real-world examples.
And because each lesson is hands-on, you'll build real intuition so you can put what you've learned to use. Now, as we've seen, the development of technology is an ongoing process that constantly reshapes the way we interact with the world around us. Today, it's more important than ever to keep up with these changes. Who knows what revolutionary technology will emerge next? But whatever it is, you can rest assured that you'll be able to learn about it on Brilliant.
And you can try everything Brilliant has to offer for free for a full 30 days. Just go to brilliant.org/veritasium. I will put that link down in the description and the first 200 of you to sign up will get 20% off Brilliant's annual premium subscription. So I wanna thank Brilliant for sponsoring this video and I wanna thank you for watching.