Speedtest.net is a popular online service that lets you test the speed and performance of your internet connection. It's been on my mind for a while to see if these speed tests are accurate! I even considered creating my own speed test app for a project. While there are technical aspects involved in measuring speed tests, I want to keep things simple and get straight to the point in this post.

I've recently discovered that third-party speed test companies like Speedtest by Ookla can be biased. They seem to favor certain networks (I won't name names, but let's just say they're mostly associated with wealthier individuals) and manipulate the results by tampering with the bits and bytes.

The manipulation happens by injecting fake bits into your browser's local storage, causing the results to fluctuate. During my research, I found that many of the bits used in the speed tests are pre-calculated and randomized. It's a common procedure for speed test services. Websites inject a JavaScript code that generates random data bits or streams, sends them to the server, and then calculates metrics like latency, round trip time, and average waiting time. This process mostly relies on TCP communication.

One important reason to avoid using speedtest.net is that they utilize compressible data for their tests. This can lead to highly unrealistic results, especially when it comes to connections using DOCSIS technology.

The issue with relying on stream compression is that it doesn't reflect the reality of internet content, which is already sent over the wire in a compressed form. Therefore, any stream compression used in the speed test doesn't accurately represent the actual performance of your internet connection.

In personal experience, speedtest.net has often overestimated my connection speed by 2-3 times its actual throughput. This significant discrepancy can be misleading and doesn't provide an accurate assessment of the true speed of your internet connection.

Also ookla has lots of trackers that sells your online footprints and bombard with ads, but lets be honest here, everyone does these days, and I won't judge it based on that.

In conclusion, I've noticed a pattern in the biased results, and it seems to change over time. It appears to be a way for them to hide these discrepancies and maintain their reputation.

If you're concerned about this, I recommend trying alternative speed test options like LibreSpeed, Cloudflare speed test, or giving Sparkyfish a shot. These alternatives may provide a fairer and more unbiased assessment of your internet connection's speed and performance.

P.S: If you want to know more about my research email me at contact@isalman.dev

This article reflects my personal opinion and should not be considered as legal proof for any study