Site updates

Humphrey and text adventure parsers

Two substantial new sections broadened the site beyond the original ELIZA demonstration.

• Humphrey is a second working rule-based chatbot. Instead of imitating a psychotherapist, it turns straightforward requests into evasive, procedural and magnificently overcomplicated bureaucratic advice. The accompanying page explains why this style is particularly well suited to simple pattern matching.
• ELIZA and text adventure parsers compares two early forms of typed computer interaction. It examines how ELIZA used words to continue a conversation, while adventure parsers interpreted commands against objects, locations and a changing world model.

More precise handling of emotional statements

The ELIZA conversation rules were refined so that phrases such as “I’m sad” and “I feel worried” receive more appropriate responses without treating every sentence beginning “I am…” as an emotional disclosure. The homepage examples and guidance were updated to make this behaviour easier to test.

A new FAQ entry also began documenting how the browser emulator has changed since its first version.

PARRY and ELIZA: when early chatbots talked

A dedicated page was added about the famous exchange between PARRY and ELIZA. It looks at how two very different rule-based personas produced a striking machine-to-machine conversation, and explains the significance of the surviving 1972 transcript and its later publication as RFC 439.

Classic and Enhanced modes, response details and expanded rules

The browser demonstration received its first major expansion. A Classic / Enhanced mode choice was added, along with a response-details panel that exposes some of the rule matching behind each answer. The rule set, reflection handling and light topic recall were expanded, and clickable example prompts were added to the homepage.

The supporting material was revised at the same time: How ELIZA Works was expanded to explain the newer behaviour, while the About page gained an About the Site Creator section.

What is the ELIZA effect?

The ELIZA Effect became a full standalone article. It explains how pattern matching, reflected language, the therapist framing and the user’s own expectations can make a simple program appear to understand more than it does. It also connects this effect to modern conversational AI and large language models.

Search, sharing and site structure

The site received a broad technical and editorial pass: clearer page titles and descriptions, canonical URLs, social-sharing metadata, structured data, a sitemap and an updated robots file. Homepage wording was also sharpened so visitors and search engines could identify the site immediately as a browser-based ELIZA emulator and educational resource.

The ELIZA Emulator launches

The first version established the core shape of the site: a working browser-based ELIZA-style console, supported by pages explaining how the program works, its influence, the ethical questions around conversational software, and the background to the project. The clean cream-and-white design, shared stylesheet and retro terminal presentation introduced here remain the basis of the site today.