Because early soundfonts were often hacked together by enthusiasts (ripping waveforms from forgotten synths, sampling toys, or recording a single piano note and stretching it across the keyboard), they accumulated strange quirks. A flute might have a stray click. A bass drum might include a second of room tone. A strings patch might have an unintended vibrato baked in.
In modern production, we call these “happy accidents.” In a soundfont, they’re features. That slightly off-pitch violin? That’s emotion. That percussion hit that loops into infinity? That’s a rhythmic bed no synth can replicate.
The statement "old SoundFonts work" is factually accurate. The ecosystem has successfully migrated from hardware dependency to software emulation.
End of Report
To understand why old Soundfonts work, you must understand the spec. Developed by E-mu Systems and Creative Technology (Sound Blaster), the SoundFont (SF2) format is essentially a sample-based synthesizer in a single file.
Unlike a modern VST that requires installation, a SoundFont is a map. It tells a sampler where to put the "Cello hit," how to loop the "Pad swell," and what filter to use on the "Bass drop." The genius of the format was its portability. In 1996, if you downloaded a 10MB SoundFont, you had a playable instrument. Today, that same 10MB file opens instantly in dozens of players.