# Network Speed Test gives curious pattern of results.

**URL:** <https://boards.straightdope.com/t/network-speed-test-gives-curious-pattern-of-results/782190>\
**Category:** Factual Questions\
**Created:** [March 14, 2017, 8:46pm UTC](https://boards.straightdope.com/t/network-speed-test-gives-curious-pattern-of-results/782190 "2017-03-14T20:46:51Z")\
**Posts on this page:** 7\
**Page:** 1

<div class="post-metadata">

**Author:** ![janeslogin](https://avatars.discourse-cdn.com/v4/letter/j/df788c/32.png) [@janeslogin](https://boards.straightdope.com/u/janeslogin)\
**Post date:** [March 14, 2017, 8:46pm UTC](https://boards.straightdope.com/t/network-speed-test-gives-curious-pattern-of-results/782190/1 "2017-03-14T20:46:51Z")

</div>

My laptop came with a ‘‘Network Speed Test’’. When I run it I first get a download speed of a bit less than 10. Re-run it and the download speed is near 20. Re-run again and again and and again the download speed is almost always near 20.

Close ‘‘Network Speed Test’’ and re-open and repeat gives similar results. First run near 10 and subsequent runs near 20. Just curious. Is there an explanation?  
The USP is ATT DSL. Windows 10. WalMarts cheapest HP15.

---

<div class="post-metadata">

**Author:** ![igor\_frankensteen](https://avatars.discourse-cdn.com/v4/letter/i/839c29/32.png) [@igor\_frankensteen](https://boards.straightdope.com/u/igor_frankensteen)\
**Post date:** [March 14, 2017, 8:59pm UTC](https://boards.straightdope.com/t/network-speed-test-gives-curious-pattern-of-results/782190/2 "2017-03-14T20:59:17Z")

</div>

The fact that the pattern repeats as described, strongly suggests that you are witnessing a …I can’t remember the word for it right now, but the word means “a phantom symptom which is a quirk of the measuring procedure, and not a real reading.”

it could be that the act of initiating the measurement, causes the apparent slowdown of the network, while the later measurements do not.

---

<div class="post-metadata">

**Author:** ![doubleminus](https://avatars.discourse-cdn.com/v4/letter/d/5daacb/32.png) [@doubleminus](https://boards.straightdope.com/u/doubleminus)\
**Post date:** [March 15, 2017, 2:09am UTC](https://boards.straightdope.com/t/network-speed-test-gives-curious-pattern-of-results/782190/3 "2017-03-15T02:09:52Z")

</div>

> [@igor\_frankensteen](#):
>
> The fact that the pattern repeats as described, strongly suggests that you are witnessing a …I can’t remember the word for it right now, but the word means “a phantom symptom which is a quirk of the measuring procedure, and not a real reading.”
> 
> it could be that the act of initiating the measurement, causes the apparent slowdown of the network, while the later measurements do not.

_Artifact_ ?

---

<div class="post-metadata">

**Author:** ![Rysto](https://avatars.discourse-cdn.com/v4/letter/r/ecccb3/32.png) [@Rysto](https://boards.straightdope.com/u/Rysto)\
**Post date:** [March 15, 2017, 3:42am UTC](https://boards.straightdope.com/t/network-speed-test-gives-curious-pattern-of-results/782190/4 "2017-03-15T03:42:32Z")

</div>

I agree that the symptoms described here sound like an issue with the test and not your network connection. I would guess that the test has to do some initialization on the first iteration that is slowing the test down on your machine, leaving it unable to keep up with the network. Subsequent iterations are able to reuse the resources allocated by the first and aren’t affected by the slowdown, and thus the test more accurately measures your connection’s bandwidth.

---

<div class="post-metadata">

**Author:** ![igor\_frankensteen](https://avatars.discourse-cdn.com/v4/letter/i/839c29/32.png) [@igor\_frankensteen](https://boards.straightdope.com/u/igor_frankensteen)\
**Post date:** [March 16, 2017, 1:20am UTC](https://boards.straightdope.com/t/network-speed-test-gives-curious-pattern-of-results/782190/5 "2017-03-16T01:20:22Z")

</div>

> [@doubleminus](#):
>
> _Artifact_ ?

Yup, that was the word!

My guess, is that the measuring program is designed to FIND the network, test how to connect to it, and only then run the tests. The first pass sets the parameters into the test memory, and so on the ensuing test runs, it JUST tests the now known net.

---

<div class="post-metadata">

**Author:** ![Shalmanese](https://avatars.discourse-cdn.com/v4/letter/s/45deac/32.png) [@Shalmanese](https://boards.straightdope.com/u/Shalmanese)\
**Post date:** [March 16, 2017, 1:32am UTC](https://boards.straightdope.com/t/network-speed-test-gives-curious-pattern-of-results/782190/6 "2017-03-16T01:32:20Z")

</div>

Some ISPs detect when you’re on a speed testing site and temporarily up your bandwidth for the next few minutes. Check with a site like [fast.com](http://fast.com) to see if you’re getting the same speeds. [Fast.com](http://Fast.com) is run by Netflix off the Netflix servers so it’s much harder to game.

---

<div class="post-metadata">

**Author:** ![JimNightshade](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/jimnightshade/32/3354_2.png) [@JimNightshade](https://boards.straightdope.com/u/JimNightshade)\
**Post date:** [March 16, 2017, 1:44pm UTC](https://boards.straightdope.com/t/network-speed-test-gives-curious-pattern-of-results/782190/7 "2017-03-16T13:44:52Z")

</div>

Sounds like TCP windowing.
