project developers get the sack after converting to Open Source says ewic

billwg

Keep in mind that anyone who has accepted the terms of the EULA, CULA, CELA, OEMLA, GLA (End user license agreement, Corporate User License Agreement, Corporate-Enterprise License Agreement, Original Equipment License Agreement, Governement License Agreement) has agreed not to publish any benchmarks without Microsoft's prior written permission. Legally speaking, a benchmark is ANY direct comparison between two competing products. In this case, between Windows and Linux.

So what does it take to get that written approval? First, you have to sign a nondisclosure agreement which prevents you from telling anyone, including federal investigators, what you had to do to get this approval. Then, you send a copy of your benchmarks, including all of the test parameters, data collected, and the conclusions you intend to publish. At this point, Microsoft will reccomend a few revisions - possibly even providing you with a revised draft as a "guideline".

In some cases, Microsoft may simply have you change the filters and conclusions. For example, if the TCO for Windows is lower in the first year, but the TCO for Linux is better in the second year, then Microsoft will have you limit your data to the first year, and have you put a caveat "Linux costs were somewhat lower in the second year", usually in your final paragraph, almost as an afterthought.

Keep in mind that the final approved report must be legally accurate, but can also be deliberately and wilfully misleading as long as it is legally accurate.

Another example is the MindCraft benchmarks. In this case, the test originally published by Mindcraft, showing that Linux was 50% faster than Windows NT couldn't be challenged or filtered, so Microsoft had MindCraft run a new set of tests using two machines carefully configured to make Linux deadlock as much as possible while Windows continued to run smoothely if slowly. There was a caveat at the very end of the report indicating that this was an unusual configuration, but most readers didn't get to that paragraph and didn't follow the link to the details.

What is the penalty for publishing a benchmark without Microsoft's permission? First, all of your licenses to all Microsoft products can be revoked. Once they are revoked, you have the option of removing them from all computers, or repurchasing them. As a result, it's quite possible that the big-mouth who published an benchmark without his employer's permission would be sacked immediately. This is much preferable to having to spend $millions repurchasing licenses you have already purchased - only this time at full retail price.

There have been a few cases where a company has taken the second option, simply removing Windows from every PC and replacing it with Linux. These customers are quite happy to tell their story, but it's only good for about 3-4 weeks of coverage, and then Microsoft has some 10 person small business tell their horror story of how they failed at switching to Linux because they didn't have any Linux people.

There are two things to note here. First, they are on NT4, they didn't need to upgrade to Windows 2000, or Windows XP. Second, under the NT4 license, they can run their software under WINE or Xen or Bochs. However, they still can't talk about how they have made the migration, or any other benchmark related material.

This silence could be an indication that things are going BETTER than expected, and that Linux IS working. A total defeat would be if Munich decided to kill Linux entirely and replace all of their Windows NT 4.0 machines with Windows XP.