How does open data and open software relate?
At a recent event I was told that open data is independent from open software. While that at the first moment seems to be true, on a second thought it gets far more complicated.
Yes, you can have open data in an open data format and load it into a proprietary software. You can work with the data. And then you may or may not be able to get your results out of it again. It is not up to you, it is up to the software. It may decide that export is only possible to closed data formats. Which means that you entered a dead end, or cul-de-sac like they say here in Ireland.
Using open software, there should not be such a dead end. Open software depends on open data formats (because closed data formats are not free, may have to be reversed-engineered from documents which means that they are not complete, can be changed from one day to another by the owner, etc). And even if a piece of open software does not offer data export possibilities, no one can restrict you from having a look at the source code and write an exporter himself.
So, yes, open data is possible without open source, but it is a dead end. And what use is open data if it can not be used to publish new insights?

Just had a funny Ubuntu problem:
I could not log in with my (valid) password if the screen was locked. But I could log in if I then switched to the „switch user“ screen and selected the same user. Reason: /etc/shadow had no rights granted, instead of -rw-r------…

I have added a simple DCM model renderer script called dcm2dot to my SPM Hacks page. With this function DCM model files (DCMxxx.mat) will be analysed and their network written out as Graphviz .dot files. You then can use graphviz small programs like dot to render the network to PNG, PDF, SVG, etc.
Here a simple example of a V1-MT+ DCM:

I have added a short script that will estimate all available DCM models within a directory to my SPM Hacks page. This is a simple way to get parallel DCM Model estimation without a dedicated PC cluster and works also with multiple PCs if you put the files on a network file system. Note however that matlab uses one licence per user per PC, so you may run out of licences fast. But if you have a 8 core Linux PC you may be able to estimate 8 models at the same time in 8 matlab sessions using just one licence.

There was a time when Safari was the fastest and the best webbrowser on the mac. It was my default browser for more than 3 years. But one thing has changed it all:

Flash.

It started small, like only one annoying flash applet on webpages that I used to read. It wasn't doing much that could not be achieved with animated gif or png. Apart from eating like 100 times the CPU power.
It did not worry me that much, I could set most of them to "hidden" with my user stylesheet in Safari. I did not notice that people all around me started cutting down the number of open pages in their browsers to only one, so that their computer did not slow down to a crawl.
But now the times have changed. Webdesigner startet using not only more and more flash-applets per page, they also started hiding them in the code, making it hard to filter them by CSS. The last drop was the switch to javascript based flash inclusion to bypass the filtering on the side of the user.
Within the last three months using Safari on my Mac has turned into a nightmare of slowliness. I could not switch of Javascript because I need it, and I could not switch of Flash because I also need it. Safari does not let me filter flash. What could I do?

Switch to Firefox. Install the Flashblock-addon. And there is no step three.