von Julia Kleinschmidt, TU Berlin
You founded and represent the C.A.S.E. (Center for Applied Statistics and Economics) computer museum. Could you please explain the inventory of the collection and what is special about it?
Petukhina: With this collection we can show the development of calculating machines, starting from the beginning of the 20th century. When we started collecting these computers our idea was to show how statistics and computer sciences are coming together and how this results in the development of computer power and computing machines. The implementation of some techniques and statistical methods, such as Big Data analysis, bootstrapping, deep learning or supervised machine learning would be impossible without existing computing capacities, and that is why knowledge of the computer’s history is quite important.
Härdle: Special on our collection is that we are not just collecting hardware. We also try to keep the software, the user interface on the computer as well so we can show how many data points they handled with. Among other things, it’s interesting to see which GUI (Graphical User Interface), which DOS (Disk Operating System) or which programming language (as UCD based Pascal) earlier computing machines used and how they developed immense branching processes. All these things you can run on our machines. It is a “lively” museum. You can also see machines that became extinct like the Commodore Amiga 500 Plus (1991).
Another unique feature of our collection is that we have a seminar on deep learning, called “Computational Statistics”. Computers and statistics are interwoven and for that we inevitably have to look at computers. Since the computers came partly from our chair and the Humboldt University is connected to GDR history, many of our oldest exhibits are objects from that time, for example the Robotron A 5120 (1982), as the first version of the office computer in the GDR. We not only want to show the differences in bulkiness in that time, but also aspects like the lack of IT-development in the GDR.
Why did you start to collect computers and how did the collection develop?
Härdle: The computer museum started when I took the Ladislaus von Bortkiewicz Chair of Statistics in 1992. From then on I collected computers from my GDR colleagues in that time. After the German reunification the chair staff thought these computers would be valuable to keep.
Pethukina: I joined the team of the computer museum two years ago to support a seminar for master students, “History of statistics”, where we work with the exhibits of the museum. Later I also was part of a team that worked on the catalogue “Computing Machines”. My part was to describe the laptops we have and I also was responsible for collecting all technical details of the objects. One of the key principles at our Institute is that we are trying to do reproducible transparent research, so all data and statistical analysis, which we did for the computer museum , can be found on our website www.quantlet.de.
Is there an object or a group of objects that you especially like? Which one is it and why?
Härdle: : In general really interesting for me is the demonstration of the branching processes when you see computers as Amiga or our other computers. They all look like PCs but they have completely different philosophies.
The PC ZX Spectrum Clone (1987) is one of my favorite objects and, as the name says, a GDR clone version of ZX Spectrum. In the GDR industrially manufactured computers were only available for public bodies. This is a privately built computer by electronic engineers only with parts that were available. Several simpler chips were used instead of one processor.
Petukhina: My favorite object is the iMac G4 “Sunflower” from Apple.
Härdle: This computer was given to me by a PhD student and it was his personal computer. Produced from the beginning to the mid of 2004, it had a completely new design compared to older Macs. With this object we want to show the design evolution, for example there is a trick to make its screen seem bigger due to its frame. You can still see this phenomena at the so called “thunderbolt displays”, so it looks more prominent. But for us it’s much more important to look at the statistical computing power that can be produced by these machines and what data they can handle.
What do you think about the significance of scientific collections like the computer museum for research and teaching?
Härdle: First of all, collections play a role in the identity of people. Someone who sees an old computer, for example, remembers how it was used by the brother or an uncle.
The computer changed statistics. For example, the Apple software Siri is statistics based on a deep learning network. It works by being optimized to the user’s voice, its tone, frequency and vibrations. It is a result of bringing a statistical technology, in this case a classification result of calibrating data, together with speech. So if you ask Siri how the weather will be, it has to understand that you want it to go to a webpage called “weather”. It seems trivial, but it is based on things that we are actually teaching in a regular seminar, like the functioning of neural networks. The seminar “History of Statistics” with Annette Vogt from the Max-Planck-Institut für Wissenschaftsgeschichte uses the computer museum to show the development of the computer and to understand methods like bootstrapping (which is called after an English version of Baron Munchhausen, who pulled himself out of a swamp by using his pigtail or in this case his own bootstrap). It is a statistical technique that uses essential data twice by resimulating.
What do you think would be the difference between object-based and text-based work?
Petukhina: I can see differences from the teacher’s point of view. Objects can support telling stories in seminars. When students can touch hardware and try software, see how it works and what it looks like, they can run code on our computers which are more than 10 years old, they get a better understanding, also about how far computer technology is developed today.
From the researcher’s point of view, the visualization of the computing machines is a good reminder of their historical development.
Härdle: We could easily build up a complete stock of computers that have ever been developed. But this would be more interesting for cultural or technical disciplines. But for now our students have a practical claim on the collection. They want to be top-notch, get deep learning and use statistics to build startups or make money and to change the world.