Choose Gephi if your priority is fast visual exploration of a social or correspondence network with minimal setup; choose Cytoscape if you need rich, queryable attribute tables, reproducible scripting, or a specialist app from its ecosystem. For the typical historian, Gephi wins on speed and ergonomics, while Cytoscape rewards projects that lean heavily on metadata and automation. Both are free, and you can move data between them with GraphML.

This is a troubleshooting-oriented comparison: the decision, plus fixes for the errors that actually stop people.

How do the two tools really differ?

Dimension	Gephi	Cytoscape
Origin	Social/humanities networks	Bioinformatics, now general
Layout speed	Very fast (OpenGL)	Slower on big graphs
Attribute tables	Basic Data Laboratory	Powerful, spreadsheet-like
Scripting	Limited; plugins	Cytoscape.js, automation API
Learning curve	Gentle	Steeper
Best for	Exploration, figures	Attribute-heavy, reproducible work

Neither is "more academic" — they suit different questions.

Why won't Gephi start?

The single most common Gephi failure is a Java conflict. Even though 0.10 bundles a JDK, a stale environment variable can override it. Diagnose and fix:

# Open the config file
gephi/etc/gephi.conf

# Set the bundled or a known-good JDK explicitly
jdkhome="C:\Program Files\Java\jdk-17"

If Gephi opens then crashes mid-layout, you are out of heap. Raise it in the same file: default_options should include -J-Xmx4g for a 4 GB ceiling. After editing, restart fully.

Why is Cytoscape so slow to open?

Cytoscape loads its app manager and a heavyweight table model on startup, so a cold launch on a laptop can take a while. Two fixes:

1. Edit Cytoscape.vmoptions and set -Xmx to half your RAM (e.g. -Xmx8g).
2. In Apps to Manage, disable apps you do not use; each one adds startup cost.

If a layout hangs on a large graph, switch from the organic layouts to Prefuse Force Directed with a node-count cap, or pre-filter the network first.

How do I move a project between them?

Use an interchange format rather than copy-paste. GraphML preserves node and edge attributes most reliably across both tools; GEXF is Gephi-native and round-trips well within Gephi.

Gephi:    File > Export > Graph file > .graphml
Cytoscape: File > Import > Network from File > .graphml

Expect to re-apply layout and styling — visual mappings almost never survive the trip. Treat the exchange as data-only.

Which should I pick for a large historical network?

If you have tens of thousands of nodes and want a readable figure quickly, Gephi's OpenGL canvas handles rendering far better. If you have a medium network but dozens of attributes per node — occupation, parish, life dates, source — and you want to filter and recolour by them repeatedly, Cytoscape's table tools pay off. A common pattern: explore and produce figures in Gephi, then load the same GraphML into Cytoscape only when a specific filtering or scripting need arises.

What breaks when importing CSVs into either tool?

• Encoding — non-UTF-8 letters mangle names; save CSVs as UTF-8 first.
• ID mismatches — edges referencing undeclared IDs create phantom nodes in Gephi or silent drops in Cytoscape; reconcile IDs beforehand.
• Delimiter confusion — European CSVs using ; need the delimiter set on import.
• Header naming — Gephi expects Source/Target; Cytoscape lets you map columns but you must do it explicitly.

Fixing the data once means both tools accept it cleanly.

Key Takeaways

• Gephi suits fast visual exploration; Cytoscape suits attribute-rich, scriptable work.
• Gephi launch failures are almost always Java conflicts — set jdkhome in gephi.conf.
• Raise the JVM heap to handle large graphs in either tool.
• Slow Cytoscape startup is fixed by more heap and fewer enabled apps.
• Use GraphML to move data between them, but re-apply layout and styling by hand.
• Standardise IDs and UTF-8 encoding so both tools import your CSVs cleanly.

Frequently Asked Questions

Is Gephi or Cytoscape better for historians?

Gephi is better for most historians because it is designed for social and humanities networks, with fast exploration and intuitive layouts. Cytoscape excels when you need attribute-rich data tables, scripting via Cytoscape.js, or biological-style integration.

Why won't Gephi launch on my machine?

The usual cause is a Java mismatch. Gephi 0.10 bundles a JDK, but conflicts arise from an old JAVA_HOME pointing to an incompatible version. Edit gephi.conf to set jdkhome explicitly to a 64-bit JDK 11 or 17 path.

Why does Cytoscape feel heavier than Gephi?

Cytoscape loads more apps and a richer table model at startup, which raises memory use. Increase the JVM heap in Cytoscape's properties and disable unused apps to speed it up for medium historical networks.

Can I move a project between Gephi and Cytoscape?

Yes, via interchange formats. Export GraphML or GEXF from one and import into the other; GraphML preserves attributes most reliably. Expect to re-apply layouts and styling, since visual settings rarely transfer.

Which handles very large networks better?

Gephi generally renders large graphs faster thanks to its OpenGL canvas, comfortably handling tens of thousands of nodes. Cytoscape can struggle visually at that scale but offers stronger filtering and table tools for analysis.

Do I need both tools?

Many historians use Gephi for exploration and figures, then Cytoscape only when a specific app or scripting need arises. Standardising on GraphML between them lets you switch without rebuilding your data.

Related reading

• How to Get started with Gephi
• Choose network layout algorithms: A Practical Guide
• Visualise large networks
• Network analysis with NetworkX for historians
• Visualise networks on the web