Why traditional speed tests don't work with Winston

Many users' first reaction when using Winston is to run a speed test. Sometimes this indicates impossibly fast speeds but often the results indicate a slower connection. The reality is that traditional speed tests are misleading in the context of a cloaked network. Here's why.

Speed test measures a single fat pipe, not parallel skinny pipes

The traditional speed test starts by detecting a server near you, then transferring batches of fixed length files back and forth to your computer. The speed at which the files are transmitted indicates the total bandwidth of your connection.

However when running behind a Winston, most of your internet activity will be routed in parallel through different locations simultaneously. Selection of these pipes varies based on local networking conditions so the fastest ones at any given time are generally chosen. This parallelism is not detected in most speed tests because speedtest measurements run through a single pipe in serial.

Winston fools speed test's location detection

Furthermore, speedtest's accuracy is heavily dependent on the proximity of the test server. However while on Winston, location cloaking will kick in and the server the speedtest picks will likely be far away from your true physical location. The requests have to travel farther so this further penalizes the result.

Winston's network is optimized for small, fast requests

The fat pipe technique is ideal for measuring the performance of streaming and other large binary transfers. It doesn't reflect what you'll typically experience while browsing the web. In reality, the difference between 50Mbps and 100Mbps download speed is largely imperceptible.

What really matters with browsing and gaming is latency. Specifically, any specific site will make dozens (if not hundreds) of small requests which need to load as quickly as possible. Winston's privacy mesh network is optimized for this. This parallelism inherent in the system is what makes it vastly faster than a VPN.

So does that mean large downloads and streaming will be slower? No.. the reason for this is that when Winston detects these large transfers, it routes them locally to make best use of your native ISP speed. These files don't make use of Javascript, cookies, or fingerprinting and are typically hosted on CDNs so they present minimal privacy risk.

Speedtest measures raw bandwidth, not perceived speed

A big part of what makes Winston so fast is that it blocks so many requests from ever leaving your browser. This means lower CPU utilization, less demand on browser background processes and less saturation of your comparatively slower ISP uplink speed. This is not a trivial optimization! Our recent benchmarks show reduction in data usage approaching 70% for Quantcast Top 500 sites.

Working towards a real speed test for Winston

We've been asked a lot to provide a speed test that can accurately measure how much Winston accelerates your browsing activity. This is now in active development and we expect to have it available for general release in January, 2020.