Bitcoin: the dial-up modem


If you think about digital transactions, you think about information. Special kind of information of course, the information must not be forged, otherwise bad things would happen. It must have autheniticy and hard to tamper.

This does bitcoin well. Many computer who check transactions(called "miners") are distributed and have their best interest in verifying transactions and keeping it right. Authenticy is guaranteed as it uses cryptography public/private key cryptography.

But if you think about digital transactions, you also think about bandwidth. The higher the bandwidth, the greater amount of transactions: it's simply information. In this aspect, bitcoin is not so good.

At home, I have a full duplex 200 Mbps connection. The bitcoin network, would be the equivalent speed of ~13.65 kbps. Yes, that's worse than your 28.8 kbps dial-up modem from the 90s.

Frankly, terrible. Bitcoin transactions are on average small; a quarter of a kB. But still, that leaves it at best at 6.8 transactions/second(note), for the entire planet.

How did I get this figure? Well, on average every 10 minutes a 1 MB block is "mined". 1 MB = 1024 kB = 8192 kb(actually, block size is a little more complicated than 1 MB, but I use 1 MB for simplification). 10 minutes = 600 seconds. 8192 / 600 = ~13.65 kbps.

Can we just increase the block size? Yes! That's possible if everyone agrees. However, it also means you have to store more data. Basically every transaction that ever was transacted must be stored at every miner/node. Problematic. Same is for time; you can decrease the amount of time per block of course. At 10 minutes, it's reasonably fast, but far from instant; not practical to get a cup of coffee.

Worse, is that not only is it slower than a 28.8 kbps modem, the modem uses multiple gigawatts of electricity.1, 2

Let's do something crazy. Assume 25% of the electricity which powers bitcoin comes from coal. This is a reasonable but safe guess as the majority of bitcoin miners is in China, and China uses a lot of coal.previous sources, 3 Now, let's say bitcoin uses 40 TWh/year, based on previous assumptions that means 10 TWh from coal. Multiply by 24.62 deaths/TWh from coal4, 5 and tada! The crappy modem kills 246 humans a year. This is a very safe guess, I did not include actual estimates and other sources(gas etc).

The higher the price of bitcoin, the more incentive there is to mine, the more energy is used.

Back to transactions: If there are only very limited transactions possible, how do the miners get to decide which transaction first? Well, the basically hold an auction. Who pays the most fee, gets in first. You can imagine that sometimes the fees get really high. I remember during the december 2017 bubble, I suddenly had more than 100 EUR of BTC, yet I could not sell my btc, because the fee was like 25 USD equivalent or something, which would lower it under 100 EUR which was the selling threshold at the broker at the time. 25 USD is a lot of money for a few bytes of bandwidth. Fees since then have dropped, but can still be expensive; it can make a paypal fee look like a joke for smaller transactions.

The transactions themselves however, do work, and work quite well.

I could go on. There's much more to talk about. But I wanted to leave it just here, otherwise this article would become "and this, and that, and..."


Bitcoin is an interesting proof-of-concept of a cryptographic currency. I think bitcoin is a great invention in computer science and practical uses of cryptography. However, I think much like IPv4, it's not designed to be used for broad, intensive, global use, and I doubt it was ever intended this way.

After all, it's a multi-gigawatt dial-up modem.

Update 2020-12-17: I fixed a humiliating mistake at calculating fatalities, I forgot to multiply the power usage by 25% to get the coal percentage, which quadrupled the death estimate. I am not sure how I missed this. How embarassing!