Not trained as a programmer.
The software behind ACARSd was written by a self-taught developer whose formal trade was baking — and whose route into C went through operational computers, radio, Perl, a web counter and a printed copy of the Apache source code.
From bakery computers to Internet systems.
In the final years of East Germany, KjM maintained the computers of the bakery where he worked. In 1995 he joined the young SWF3 Online team, where he helped build and operate the ELCHBOX mailbox system and began programming in Perl.
His boss there was Karl Ewald. Karl was not only a boss but also a friend. One project, however, produced a memorable argument: a Perl web counter occasionally fell back to zero when concurrent accesses collided with the locking behaviour available at the time.
Karl’s verdict was characteristically direct: “You’re too stupid.” KjM was furious. The dispute eventually contributed to his decision to leave SWF3 — but it also triggered one of the most consequential acts of revenge in this story.
The revenge was C.
In 1998, KjM printed the complete Apache source code and used it to teach himself C. Within roughly three weeks the Perl web counter had been replaced by a C implementation. The resulting free service at webcounter.goweb.de became a widely used web counter and continued for many years before eventually being retired.
With hindsight, the argument with Karl looks very different. Karl later died unexpectedly, and an old technical quarrel became a very small thing compared with the friendship behind it. Yet that quarrel changed the direction of KjM’s life: without the determination to learn C, there might never have been an ACARSd.
Thank you, Karl. Even our argument changed the direction of my life.
That learning path matters when reading ACARSd. The code was not written from a university DSP curriculum. It came from studying production source, solving concrete systems problems and measuring what worked.
Armin also brought Tcl/Tk into the story.
Years before ACARSClient, Dr. Armin Zundel had another idea that initially made KjM groan: build a community system not in Perl or PHP, but in Tcl/Tk because it would be fast. KjM learned it reluctantly, then discovered that it was actually enjoyable. That detour mattered later. When ACARSd needed a portable graphical client, Tcl/Tk was already a familiar tool — and it became the basis for maps, server lists, databases, chat, aircraft images and a few well-hidden games.
Source note: Armin's Tcl/Tk influence is a 2026 recollection by KjM. The surviving ACARSClient source independently confirms the extensive Tcl/Tk implementation that followed.
Why that matters for libacarsd
The decoder reflects the same mentality. Rather than forcing the radio signal into an elegant theoretical model, it used captured sound, empirical comparisons and multiple cheap decoding strategies. Failed messages became test inputs; successful tables were counted and prioritized; CRC and message structure decided whether the result was credible.
The code is less an academic exercise than a record of an engineer teaching the machine to survive real-world radio.
ACARSd as a technical time capsule
The project sits at an unusual intersection: Linux desktop/server software, analog radio, early web services, Tcl/Tk clients, distributed Internet receivers and aviation databases. Seen from 2026, many of its ideas feel familiar. Seen from 2003, they were being assembled with very different hardware and a far less mature Internet ecosystem.
This museum exists to preserve that context.
From hobby project to FlightGlobal ATLAS
By 2010 the same systems-minded approach led into a commercial aviation project. Recovered CVS sources show that ATLAS was built for FlightGlobal as an Adobe Flex/Flash application on Google Maps, fed by community SBS stations. The source contains type-specific SVG aircraft and an airline-colour engine that could colour body, wings, tail and engines separately.
The project also marks the difficult end of ACARSd development. According to KjM's recollection, the commercial contract tied continued work on ACARSd and related software to the continuation of that collaboration. When it ended, so did active ACARSd development.
Read the ATLAS source analysis →
And then I simply walked away.
For years aviation, ACARS and aircraft tracking occupied an enormous part of KjM's spare time. There was also a quiet uncertainty in the background: would an airline or another organisation one day object to what the network was receiving, decoding and archiving? That letter never came. The aviation years remained surprisingly peaceful.
According to KjM's recollection, when the FlightGlobal/ATLAS collaboration was not renewed, he honoured the agreement and stopped active work on ACARSd and related projects. He did not rebuild the tracker under another name. He simply walked away from aircraft tracking.
His aircraft-photo database, www.fysb.de, also disappeared later in the privacy-policy/GDPR era. During the ACARSd years it had been much more than a photo gallery: clients used it as a source for aircraft images and icons, helping turn a registration received over radio into a recognisable aircraft.
This museum is not an attempt to restart ACARSd. It is here to remember those years.
The people around the project
ACARSd depended on a remarkable informal network of testers, aviation-data specialists and public server operators. A dedicated community page now preserves those acknowledgements and invites former users to contribute their own memories.