Hi Keith,
First, sorry this email is so long. Sometimes I don't realize how much I've written until I'm done.
I can't really speak for other speed tests out there but I can share some information I've gathered about some other so called Internet speed tests. This information may pertain to BT's speed test as well. >> https://testmy.net/ip...la-speed-tests/ -- there are related articles linked at the bottom of the first post of that topic that may also shed some light on the subject.
Most likely the speed is being reported below BT's speed test result because they are testing your connection within their network. These days almost all providers prefer this because it makes your connection appear better than it really is. Also read >> https://testmy.net/legit-speed-test.php
My speed test has many things going for it that lend to it's accuracy.
TestMy.net has been around since 2001... I have more experience than anybody when it comes to Internet speed testing. It started as a hobby and as a tool I built for myself but for the past 8-9 years I've been completely dedicated to TestMy.net. I've actually been developing my tests since 1999 but in the early years I had to shut down for a while because it cost me too much to host the site. Back before 2001 there really wasn't sufficient Internet ad revenue to offset my bandwidth costs. It grew too fast, too quick and too early in the history of the Internet. I had to shut it down until the cost of bandwidth decreased and ad revenue possibilities increased but I never stopped developing it. I even had sponsors that promised me bandwidth only to shut me down when they realized that I wasn't exaggerating with my bandwidth projections. It was pretty frustrating times. Thank god for Google Adsense and my other sponsors... also the cost of bandwidth is more than 100X less than it was.
My tests do not use third party plugins such as Sun Java and Macromedia Flash. Although it might have been easier to design the tests using those... they aren't accurate test platforms. It has always been in my design to utilize ONLY the web browser. Tests that use other protocols are at the mercy of those protocols. They are subject to many factors that my tests simply don't have to deal with. In many cases those tests are making adjustments based on guesses... adjusting for unknown variables. Some of them make adjustments of as much as 30-40% to offset a variable that could be as little as 5%... they don't know the real number of how much they need to offset so they guess! Now last time I checked, my visitors are looking for a test... not a guess. --- I've never seen a referral from search engines using keywords like, "bandwidth speed guess" -- lol
TestMy.net does not host within any providers networks, my host directly peers with many providers but my servers are not on the edge of their networks. All providers are graded based on the same criteria, everyone is given a fair chance to score well. If the provider is weakly peered or has issues with their routing then it will be more apparent using TMN. Providers would like you to test within their networks because when you do you eliminate variables that could make them look bad. These variables that may or may not be in their control. But I feel that's the information you were looking for. If you want to test your Internet speed then you have to test it through the Internet, using a normal routing procedure as if you were visiting any other website. Don't you agree? I provide a real world scenario not a best case scenario.
Overstocked and overpowered... my servers are more powerful than what is needed for my application... WAY overpowered. I have multiple gigabit symmetrical uplinks to one of the most powerful, well connected hosts in the business, hosted on major Internet pipelines. My main server is smack dab on top of the heart of the Internet crossroads, right on an artery where a larger percentage of the worlds popular websites are hosted. Any provider worth anything should peer well to that location.
Distance is not always a factor. Some other sites lead you to believe that you need to test off a server near you. It's a complete fallacy. Okay, let me ask you this, when you visit a website... how often do you get to pick where it's being served from? Sometimes they ask you location on International sites but that's usually more for a language preference. How often is that being served even from the same state that you're in? Trust me, not very often. I have seen people on the other side of the planet pull 200+ Mbps from TMN. A properly configured like on a well connected provider can max out or nearly max out over thousands of miles. I offer east coast and west coast servers only as a comparison to the main server to help diagnose routing issues. I could easily set it up so that people are sent automatically to the closest server but I feel that isn't always an accurate representation of your connection. I personally live over 1400 miles from my home server, I'm able to max my 50Mbps line with no problem... as long as my home connection is working the way it's supposed to.
Randomized, un-cacheable test information and large file sizes. Other speed tests rarely tell you how much information is actually being transferred and I can almost guarantee you that the information used for testing is the same from test to test. Again, I can't really speak for the other guys... but I know that they don't do it like I do. The data that testmy.net uses is created by the server on the fly. Unlike other speed tests I don't use real physical files. While you're testing random, incompressible data is being created by the server and sent to you. No two tests are alike. This makes it impossible for the data to be compressed by your provider, which would lead to a result that's higher than the true speed. (most common among satellite providers)
That's a few reasons why other speed tests out there may report differently. There are more reasons, TMN is unique and proprietary in how it operates. BT as well as just about every provider on the planet USED to send their customers to TMN to test their connections. I've had a lot of call center techs send me copies of training manuals from back in the day where it was standard procedure to test at TestMy.net. It was seriously step one in many cases. What I think happened is they realized that they could cut down on service calls and save money by showing customers that their connection is working fine to THEIR servers. They then blame any problems on the website that the customer is visiting. The problem with that is that the provider is at least partially responsible for the route you take from your computer to the server you're visiting. They choose who they peer through and sometimes these connections going out to the Internet that they have can be overloaded. If the test that you're using bypasses the route that your provider takes to get you outside their network then you won't see those issues. ... how can that be a test. They also have inferior testing technology. I've been developing my speed test over many years, lots of trial and error and the way I see it... I'll never finish this project. Providers who host their own speed tests have stake in the results... if they are misrepresented the company stands to possibly gain something. .... I on the other hand grade everyone the same. I have no ties to any ISP and have no reason to inflate results. If people suspected that I was misrepresenting the information it would hurt my business. The big ISPs could care less about the traffic to their speed tests... the larger companies stand to easily save millions in service calls and call center support if people think that it's the fault of the website they're trying to visit. ... see what I'm laying down?
I've been trying to spread the word about these practices for years. But it's very hard. First, I have to get people to read what I've written. Then I have to convince them that I'm an authority on the subject. It's really hard because I seriously have nearly every Internet provider out to get me right now. I'm telling their customers the truth that they don't want them to know. I've had A LOT of providers contact me wanting me to host my test within their networks, often offering me money and hosting to make it happen. But I feel that I'm working in the interest of the consumer... how can it be in the interest of the consumer if the test isn't really testing the Internet?
Hopefully with help from the community I can get the truth out. I truly feel that they're doing it on purpose in many cases and it makes me sick. Consumers are being blatantly lied to. All I can do is keep programming and writing for what I think is right and hope that people visit and tell their friends.
If you like what I've built, please pass it along... TMN is 100% word-of-mouth.
