Human Diversity and the Choice of Interface: A Design Challenge
Starr Roxanne Hiltz, Murray Turoff · 1981 · Proceedings of the Joint Conference on Easier and More Productive Use of Computer Systems (CHI '81), Part II: Human Interface and the User Interface · doi:10.1145/800276.810974
Summary
Hiltz and Turoff report findings from a multi-year National Science Foundation-funded field trial of the Electronic Information Exchange System (EIES), a pioneering computer-mediated communication platform built at the New Jersey Institute of Technology that by 1981 had hosted more than 2,000 users and 200,000 hours of online activity. The paper is framed by an analogy the authors push throughout: the relationship between a user and an interactive system resembles a marriage — complicated, adaptive, and with no guarantee of success. Because humans are highly variable in skill, temperament, and task, and because real users rarely have the time or money to try alternatives, systems that offer only one interface inevitably fail some of their users. EIES was deliberately designed as a 'polygamist' system, offering four coexisting interaction styles — long-form menus, short menus, answer-ahead command streams, and direct commands — alongside self-defined command macros built with INTERACT, an embedded scripting language. User assistance was equally diverse: a printed loose-leaf 'How to Use EIES' binder, the on-line 'CHIMO' newsletter, a comprehensive explanation file indexed by `?word` and `??` help triggers, and a team of volunteer 'user consultants' who earned free account time and TELENET access in exchange for mentoring, testing, and writing documentation. Three mailed questionnaires (pre-use, 3-6 month follow-up, and 18-month post-use) plus system monitor logs and participant observation formed the empirical base.
Key findings
No single interface or help source satisfied all users or even the same user over time. Follow-up data (N=96) show a loose progression — new users relied on the long menu (45% frequent use at 0-20 hours), then migrated toward commands (75% frequent use at 100+ hours) — but the pattern was never clean. Even heavy users kept menus available (41% of 50+ hour users still 'sometimes' used the long menu to refresh their memory or explore unfamiliar areas). 70% of surveyed users endorsed teaching menus first while making equivalent commands available, and support for menu-first onboarding grew stronger with experience (88% among 100+ hour users). On help sources, human user consultants were ranked top at every experience level and drew overwhelmingly enthusiastic open-ended comments ('fantastic,' 'essential,' '9 times out of 10 prompt, helpful, courteous, and friendly'); the CHIMO newsletter was second; the `?word`/`??` on-line help was third; and the full explanation file was least used (only 2% rated it extremely valuable, 30% never used it). The authors also note wide variance in help-source combinations, suggesting underlying personality differences between users who prefer human assistance versus automated help.
Relevance
Although EIES predates the modern web by over a decade, this paper articulates a principle that is now central to accessibility and inclusive design: there is no universal interface, so systems should offer multiple coexisting modalities and let users pick — and switch — as their abilities, contexts, and expertise evolve. The argument anticipates later frameworks such as ability-based design, progressive disclosure, and the WCAG principle of 'Adaptable' content. The finding that human helpers outperform all forms of documentation is also a useful counter to the assumption that better docs alone will solve onboarding problems; for accessibility practice, it supports investing in live accessibility support channels, peer mentors, and community knowledge rather than relying solely on written guidelines. Limitations: the sample is professional scientists on a niche research system, the questionnaire on user consultants was leading, and concepts like 'diversity' here refer to general human variation rather than disability specifically — though the paper does note that one of the EIES user groups studied 'devices for the disabled.'
Tags: user interface · computer-mediated communication · adaptive interface · personalization · novice vs expert · user support · historical · interface design · human diversity