- Spain’s deadly rail collision raises urgent questions about what happens after a safety system stops the trains
- Rail and telecom infrastructure need automated guardrails that constrain both human and machine decisions
- Performance can evolve rapidly, but safety must remain independently enforced
The horrific train disaster in Urda, Spain, which claimed at least 68 lives and injured more than 100, is a sobering reminder of the compounding risks in critical infrastructure. When modern demands push systems to their breaking points, legacy architectures can collapse beneath the strain. This is true of both rail networks and telecom networks.
In the railway sector, “high speed” means moving hundreds of passengers at velocities that leave little room for error. In telecommunications, the challenge is different but structurally similar: handling vastly more traffic than a generation ago, where data is no longer for fun but a lifeline for healthcare, industrial automation, emergency responses and financial markets.
The Madrid-Seville line relied on LZB, a longstanding German-designed train-protection system that continuously exchanges information through cables laid along the track and onboard antennas. According to initial reporting, a fault interrupted that protection and the trains stopped — the safe response.
The disaster happened after the stop. Human controllers authorized traffic to resume, but the train ahead, run by operator Iryo, remained stationary. The Renfe AVE train behind it proceeded and struck the first train. Initial reports point to a human error in authorizing that movement; investigators must establish exactly what instructions were given and why.
The distinction is crucial. The initial failure brought the trains to a halt. The suspected human error came when traffic was restarted.
What might have prevented this tragedy — the fourth deadly train crash in Spain this year? An upgrade to the underlying safety system and, ironically, a reversal in the role of the human operators.
A modern architecture like European Train Control System (ETCS) Level 2 uses 2G-based GSM-R wireless technology to send movement instructions to the train. Onboard equipment checks the train’s movement against permitted limits and can automatically apply the brakes.
If normal signaling fails, ETCS allows trains to proceed under an enforced maximum speed of 18 mph. Humans still decide how to respond, but automated protection limits what the train can do.
The driver operates the train. The protection system enforces the limits.
In vast communications networks, the exact same principle applies. Historically, when a core router failed or traffic surged, human engineers logged into command lines to manually reroute data packets — frequently introducing catastrophic configuration errors. Modern software-defined networks (SDNs) can eliminate this risk through automated circuit breakers and policy-as-code guardrails. If an engineer pushes a faulty configuration or a malicious script, these automated safeguards instantly reject the change or contain its blast radius.
Crucially, an SDN does not provide these protections simply because it is programmable; engineers must intentionally architect them into the system. Most importantly, the person or software changing the network must never have the authority to casually switch off the autonomous safeguards governing those changes.
Yet best practices preclude throwing either system — trains or networks — into unrestricted autonomous operation. Train drivers still operate high-speed trains, even where automation handles parts of the journey. Humans remain responsible for operational decisions, recovery and responding to unusual trackside anomalies like debris. True automation also introduces massive cybersecurity vulnerabilities; a single malicious hack or rogue software loop could hijack or crash an entire transit grid. Both people and machines need limits.
In telecommunications, Level 5 describes full autonomous network operations. Pursuing that without effective oversight, safety limits and human recovery mechanisms is a dangerous idea for the exact same reasons. A network stripped entirely of human supervision becomes a black box vulnerable to systemic cyber exploits and cascading software errors spreading instantly across the system.
Full autonomy does not have to mean abandoning human responsibility. TM Forum’s work with UAE-based operator e& retains human governance and policy controls within a roadmap toward Level 4 and Level 5. Machines can manage routine operations while humans set the objectives, define the limits and remain accountable for the outcome.
On the back of an appalling loss of life, we must extract a vital lesson for critical infrastructure design: decouple safety from performance. Safety should not depend on a human making the right decision in the seconds after something fails. Conversely, tie performance to the latest technology — like upgrading telecom connectivity to 5G or introducing AI-driven routing — but keep human engineers accountable for the driving loop.
Applying this “Automated Safety Loop, Human Driving Loop” model to virtual networks is a frontier dividing resilient infrastructure from vulnerable systems. Industrial automation companies, network equipment suppliers, operators and hyperscalers all have a role in building these protections into physical infrastructure and software-defined networks.
Fragmented ownership and legacy systems can make that work harder. But the dividing line does not fall neatly between countries. Germany’s dependence on aging 2G railway communications illustrates how advanced infrastructure can still depend on vulnerable foundations.
For the builders of vast communications networks, the mandate is clear: give your engineers the throttle to drive performance with the latest technology, but let independently enforced autonomous code maintain the guardrails.
Related reading by Stephen M. Saunders
The ghost in the machines — Why the world can’t quit 2G
China Mobile takes the express train to 5G revenue
The Lobito Corridor: China enters the chat
Stephen M. Saunders MBE is a communications analyst and USPTO-registered inventor examining how digital infrastructure — 5G, cloud and AI — is reshaping industry, power and society, as well as underpinning the emerging, ubiquitous global digital economy. As anchor of FNTV and a longtime industry insider, he focuses less on growth narratives and more on execution, risk and how hyperscale technology is distorting markets, governance and society at scale.
Opinion pieces from industry experts, analysts or our editorial staff do not represent the opinions of Fierce Network.