Why does the spreadsheet work … … and how can we improve it? A user interface perspective on the foundations of spreadsheets. Alan Blackwell Computer Laboratory.

Why does the spreadsheet work … … and how can we improve it?

A user interface perspective on the foundations of spreadsheets.

Alan BlackwellComputer LaboratoryCambridge University

Outline

Why the spreadsheet was invented Which features were critical to success HCI research in spreadsheet use Where next?

Grounds for critique of current products Possible product futures Possible futures for spreadsheet users

Timescale of UI Evolution

1940s:Scientific instruments

Timescale of UI Evolution

1940s:Scientific instruments

1950s:Mathematical tools

DIMENSION A(11)

READ A

2 DO 3,8,11 J=1,11

3 I=11-J

Y=SQRT(ABS(A(I+1)))+5*A(I+1)**3

IF (400>=Y) 8,4

4 PRINT I,999.

GOTO 2

8 PRINT I,Y

11 STOP

Timescale of UI Evolution

1940s:Scientific instruments

1950s:Mathematical tools

1960s:Data files & records

Timescale of UI Evolution

1940s:Scientific instruments

1950s:Mathematical tools

1960s:Data files & records

1970s:Command languages

OBEY

YESSIR

Timescale of UI Evolution

1940s:Scientific instruments

1950s:Mathematical tools

1960s:Data files & records

1970s:Command languages

1980s:Bitmapped screens, direct manipulation

“Modern” UI Essentials

Drawing e.g. Sketchpad

1963

“Modern” UI Essentials

Drawing e.g. Sketchpad

1963 Presentation

e.g. NLS 1968 (also hypertext,

online collaboration …)

“Modern” UI Essentials

Drawing e.g. Sketchpad

1963 Presentation

e.g. NLS 1968 (also hypertext,

online collaboration …)

Word Processing e.g. Bravo 1973-77

Inventing the Spreadsheet (1978)

Bricklin’s concept: “electronic

blackboard” closest to NLS

collaboration tools? Frankston’s

optimisation: fast, compact &

responsive Fylstra’s market:

individual ownership and control via PCs

Outline

Why the spreadsheet was invented Which features were critical to success HCI research in spreadsheet use Where next?

Grounds for critique of current products Possible product futures Possible futures for spreadsheet users

From Apple II to IBM PC / MS-DOS

PC Functionally equivalent to the Apple II (no mouse, no bitmapped display)

Business software emphasis simple databases word processing Lotus 1-2-3 replaced VisiCalc, included

charts and plots (as well as some database and text formatting)

Direct manipulation

Original Mac applications were “creative” writing, drawing, painting derived from Kay’s vision of creative machine

Business market developments desktop publishing created business market Apple asked Microsoft to develop a SS

Excel offered direct manipulation benefits of the Mac point and click, menus, windows, prompts

The Spreadsheet “Metaphor”

HCI textbooks propose metaphor as starting point for UI design. But there is little evidence that the paper

SS ever influenced the development of SS software.

Most benefits historically derived from UI features of other software categories.

Outline

Why the spreadsheet was invented Which features were critical to success HCI research in spreadsheet use Where next?

Grounds for critique of current products Possible product futures Possible futures for spreadsheet users

Experimental extensions

UI builder functions (Myers) Navigation aids (Rao & Card) Debugging aids (Burnett et. al.) Typing and inference (Erwig) Gesture interfaces (Wolf, Burnett) Approximation, graphics (Lewis) Multi-user support (various) Tutoring and help systems (various) Voice interfaces (various)

Spreadsheets in end-user context

Bonnie Nardi, A Small Matter of Programming (MIT Press 1993) notes: SSs are immediately useful for real tasks SSs support direct manipulation SSs have limited control constructs SSs lead to collaborative communities

But SSs are: Not automatically easy to learn (Hendry

& Green) Liable to contain errors (Panko, others)

Mental models of spreadsheets

The user’s “mental model” is critical in HCI But hard to say what goes on inside a

programmer’s head! Saarilouma & Sajaniemi (1989) showed SS

users employ visual images Navarro Prieto (1998) found visual images

help SS users understood dataflow Petre & Blackwell (2000) note that many

programmers report experiencing images

Some theoretical design principles

Cognitive Dimensions of Notations Discussion vocabulary for significant

design attributes and tradeoffs (see Green & Petre in JVLC 1996,

Blackwell & Green in Carroll, ed. 2003.) Surprise, Explain, Reward

We’ve built it, but will they come? (see Robertson et. al., CHI 2004) See also Blackwell’s Attention

Investment theory of abstraction use

Outline

Why the spreadsheet was invented Which features were critical to success HCI research in spreadsheet use Where next?

Grounds for critique of current products Possible product futures Possible futures for spreadsheet users

Lessons from history & research

Account for collaboration Be fast and responsive Empower individuals Provide a flat, imageable world Support direct manipulation

Do current products retain these benefits?

Outline

Why the spreadsheet was invented Which features were critical to success HCI research in spreadsheet use Where next?

Grounds for critique of current products Possible product futures Possible futures for spreadsheet users

Features are not the answer

Applications are developed in SSs because of their low entry cost, seldom on engineering grounds. 57% of 5500 SSs had no formulas

Can SS functions be partitioned into developer and end-user sets? How would the transition be managed?

Can we provide engineering benefits to end users? EUSES, e.g. Burnett, Erwig, Blackwell

Outline

Why the spreadsheet was invented Which features were critical to success HCI research in spreadsheet use Where next?

Grounds for critique of current products Possible product futures Possible futures for spreadsheet users

User simplification

If the SS has been moving away from its roots, can users (or organisations) return? Templates Process definitions Reduced-feature standards Adapt open-source products

Any more options? (DISCUSS!)