Prior Engagements

Prior to launching E D Reactor our founder and principal enjoyed a distinguished 12-year career as a full-time employee at several of the leading companies in the software industry. Some highlights are included here. See the accompanying resume for all the gory details.

    Propel Logo  

Propel Pattern Language



Kevin's most recent full-time position was Fellow in User Centered Design at Propel, which at the time was creating a turnkey platform for consumer e-commerce sites (think "Amazon-in-a-box"). The company intended to provide e-tailers with a robust server platform, user experience model, and merchandising tools in an easily customizable form that would allow them to focus on their product mix, promotional strategy, and branding instead of the drudgery of building an in-house development team and re-inventing the wheel for the all the back-end implementation work.

Kevin mentored a very large user centered design staff (13 people in a company of 130) and led the development of a set of design patterns for e-commerce (left) that documented the rationale behind a set of packaged components the company was developing to encapsulate current best practices in retail e-commerce design, as determined by the group's extensive research program studying site design and consumer behavior. Unfortunately the bottom fell out of the e-commerce tools market in 2001 before this vision could be fully realized.

    Scoutfire Logo



Holiday Theme toolbars

CBS Marketwatch toolbar


As Employee #4 at Scoutfire, Kevin served as Design Director and chief product visionary for a series of designs that evolved across multiple changes to the underlying product concept in an attempt to come up with a fundable business model. The core technology was a toolbar that , through its various states (top) provided a continuous communication channel to the user and could analyze the current page contents and user clickstream and respond appropriately. The initial product concept was an auction monitoring service that would let users find, track, and bid on items they were seeking across a broad range of auction sites.

To overcome the challenge of distributing a product that required a user download and installation we began to explore a series of viral download motivators. These concepts centered around an expanded area above the toolbar in which graphics and interactive controls could be placed. We explored "widescreen" banner ads with integrated peripheral controls (row 2) that when targeted to the user might provide the same kind of desirable "eye candy" that screensavers exploit. We used the same area to display collections of favorite photos or greeting cards (row 3) for easy access and personalization of the user's work environment.

Eventually, we found it most interesting to make the space available as fully interactive functionality that could be timeshared with whatever else the user was browsing. From a chess game (row 4) provided by and sponsored by IBM (we envisioned "Deep Blue" playing thousands of players around the world simultaneously), to an embedded CD/MP3 player with track info and album sales and cross-promotion links (row 5) provided by, to a market-monitoring portfolio tracker with tickers and alerts and quick trade buttons (row 6) provided by CBS Marketwatch, the possibilities run the full gamut between frivolous and functional. We created a business model in which leading brands like these would distibute the toolbar in return for continuous access to their customers from anywhere on the web.

Finally (bottom) we explored ways to provide the essential filtering and targeting qualities that powered the toolbar to mobile and PDA environments where web browsing was supported but in which standard toolbar designs were not available. We were able to emulate the toolbar functionality fairly well using a dedicated Scoutfire menu.

The company was acquired by Microsoft's WebTV division in 2000.

    Icarian Logo  




Kevin was Employee #7 at Icarian (a Kleiner-Perkins funded startup since acquired by Workstream), where he served as User Interface Architect and Manager of the User Interface Design team. There he was solely responsible for the design of Icarian Workforce user experience through its first release and managed a 3-person design team through the second release. Workforce is a major enterprise class application designed to support the hiring process in large companies with a heavy investment in skilled technical and professional employees. Its design effort followed a very thorough requirements gathering and task analysis phase, with extensive prototyping and informal usability testing throughout the process and a formal testing program prior to the initial release. The resulting was named "Product of the Year" for 1999 by Human Resource Executive Magazine.

The core product (used by line managers and HR personnel) features a pipeline metaphor in which an open position could be chosen from a list of openings (row 2) to view the list of candidates under consideration for that position (row 2), the interview feedback on candidates advancing to that point, and the status of any offers extended thus far. Candidates who accept an offer are added to the list of Employees at the far end of the pipeline to complete the process. The design successfully employs a multi-window design with true pop-up dialogs and command windows for specific local actions (bottom). The result is a true "best of both" experience in which common web idioms were used to enhance the familiar desktop application structure in a way that was simple and natural.

One of the significant design innovations in the product was a flexible workflow and policy management system that allowed the HR organization to establish procedures and impose constrains on user actions where appropriate (e.g, you can't make an offer before a req has been approved) while providing the flexibility needed to handle unique cases and work around any process breakdowns that might arise. The right-most column in each display lists the next action or decision in the "normal" sequence for that line item. Alternative actions are accessed through an action menu attached to the icon immediately to the left of the line item. This solution preserves relatively efficient access to all of the commands the current user is eligible to perform while streamlining interaction by optimizing visibility and access for the most common case.


Icarian Analytics



A key selling point of Icarian Workforce was the immediate visibility that tracking data collected throughout the hiring process could provide to senior executive management. The data allowed us to project when a given position would be filled based on candidate flow and historical fill rates for similar positions companywide. This was critical knowledge as it shed light on future payroll costs and product delivery plans. Kevin developed and sold the product concept (in his initial position reporting to co-founder and VP of Marketing Rani Merrit he was deeply involved in all of the product management work), developed the detail design and HTML prototype (shown here), and oversaw the implementation of Icarian's first and most important analytics application.

    Netscape Logo  
Motorola CBT Overview  

Kevin worked briefly with co-author Darrell Sano as a Managing Principal Consultant and Engagement Manager in the Information Architecture and Design group of Netscape's Professional Services organization. Before being lured away by Icarian, he led the design and delivery of a major Computer Based Training project designed to ease the transition from a variety of legacy email clients to Netsape Messenger throughout Motorola's Semiconductor Products Group.

The diagram shown here illustrates the available navigation paths through a clearly organized information space. After an Overview segment (purple pages) seen by everyone, users could proceed through standard sequential tutorial material covering Basic and Advanced Topics (the long linear axis in the visualization), with the ability to skip ahead to any topic, or they could view QuickChange summaries that covered only the key deltas for people switching from a particular legacy platform (the short linear axis), or they could check screen level Reference pages (gray elements "below" the short axis).

    Macromedia Logo

Macromedia UI Logo

Authorware Dialog w/ Tabs


Kevin was hand picked by CEO Bud Colligan to lead a much-needed cross-product design effort resulting in what came to be known as the Macromedia User Interface.In five years the company had, through numerous mergers and acquisitions, more than tripled the number of products it offered. These products had all been developed separately, so they had very little in common at the level of the user experience. Kevin assembled a team of approximately 25 stakeholders, with represntatives from each product team, and led them through a series of weekly meetings during which the topics to be addressed and the general solution strategy to be pursued were outlined and agreed to. By the end of the quarter-long process, the team had produced a thick design document that was agreed to by each of the development teams.

Kevin then made the rounds to the major product teams and assisted with the detail design work needed to help realize MUI (more or less) in their application. He branded the program with the logo shown here (top) and had t-shirts made as an incentive for product teams that completed their migration to the new MUI standard. By the end of the year even applications as fundamentally different as Director and Authorware were sharing the same dialog style and layout conventions (bottom). Tools that all had their own 3D look and playback controller design converged to a coherent, companywide (well, except for Dreamweaver!) standard (bottom). The program was considered a huge success by management, which was a mixed blessing, since it removed the pressure to complete work needed to fully realize the vision.


D6 Score Design

Director 6 Overview


Kevin's second Director release (Direcor 6.0) was a huge success that finally began to fulfill the promise of a simpler, more straightforward authoring environment for non-expert users. The major achievement was the creation of an entirely new Score design (top) in which sprites became true objects spanning multiple frames with exposed keyframe control points. We conducted extensive usability testing on the detail design and interaction dynamics of the new score throughout the developmetn cycle and were very happy with the results.

The other major breakthrough feature was encapsulated behaviors, which were simply normal score scripts that could be managed by external tools such as the Behavior Inspector shown here (bottom, upper right). To everyone's surprise, not only novice users, but even skilled programmers found the Behavior Inspector much easier to use than the existing Script Editor for simple repetitive programming tasks in which parameter changes could be input directly once the affected object was selected.


Director 5 Dialog Programme

Director 6 Interactivator


Kevin was hired by Macromedia to redesign Director, their flagship product and one that he had been using regularly since the days when it was known as VideoWorks. Director was famous for its capability and flexibility as a multimedia authoring tool, but it was also notorious as one of the most complex applications on the market. The challenge (as so often is the case) was to make Director more approachable for people who were not expert programmers, animators, or multimedia producers without undermining the product's efficiency for advanced users who spend large chunks of every day with the Director Score (top) as the focal point of their mission critical application.

Major enhancements for Director 5.0 included a delightfully simple Align and Distribute dialog (2nd row) that provided nine cells and eight border areas mapped onto the various spatial relationships between objects. Users simply clicked on the cell corresponding to the desired alignment policy, noted the dynamic graphical preview to confirm it was the alignment they wanted, and clicked the Align button to apply the policy to the selected obects.Or they could just double-click the appropriate cell to apply the policy immediately. The result was probably the fastest general alignment dialog that will ever be created, yet it remained simple enough to appeal to novice users as well as seasoned professionals.

The menu system was completely redesigned and the entire dialog program received a new layout (rows 3 & 4) based on a grid designed to more effectively isolate transaction controls and bulky previews and read-only status information from the controls reflecting the content to be manipulated. David Cortright collaborated on the layout and took on the bulk of the production work required to generate the necessary Windows and Mac resources as a summer intern. The layout was subsequently adopted as the standard for all MUI applications.

Included in the Director program was a new set of dialogs (bottom) created to automate various common animation and authoring tasks. Unlike the rest of the product these dialogs were implemented as Director movies that used the new score authoring API's to write content directly to the score based on the user's choices. Kevin handled the design, production, and Lingo programming for all of these tools so when he needs to assert credentials as a coder - generally in debates with development teams asserting credentials as designers - his claims will not be entirely without merit.


Helix GUI

Helix Widget Palettes


Kevin's final project at Sun took place in Sun Technology Enterprises where he was brought in to serve as the design lead for Helix, a rapid application development environment that allowed developers to quickly string together a collection of pre-packed UNIX libraries and data storage components and assemble a GUI using the OSF/Motif toolkit (Sun having finally thrown in the towel on the GUI wars and joining the rest of the industry to make Motif the standard). Helix used an embedded Scheme interpreter to emulate C++ while the application was being constructed, so users got the benefits of a very powerful scripting environment at design time and the performance of real compiled C++ code at run time. The back end never really worked as advertised and the project was eventually cancelled, but the user interface was a tremendous success.

The process began with extensive prototyping (in SuperCard, on a Mac- we had to use a Sun monitor and hide the Mac CPU to get around Scott McNealy's "Run Sun on Sun" dictate!) to validate product organization and navigation model ideas and to explore various interaction designs for the crucial editing tasks (top and 2nd row). We next wrote tight specifications for the appearance and behavior of the Application Browser (row 3)and Library Manager, with particular attention to the complex selection, filtering and sorting, and multi-object editing protocols required for efficient editing of large scale applications. We abandoned structured dialog-style layouts for component editing at a fairly early stage in favor of the more generic lightweight table editor (row 2), figuring we would want that eventually and its greater simplicity would get us up and running as quickly as possible. What we quickly learned was that the flexibility permitted by the generic editing mechanism easily outweighed the "comfort" of a GUI form.

The final implementation (row 4) was fairly true to the evolving design. A major issue with the development team had always been the heavy distinction drawn between the Application Browser and the Library Manager. These initially shared the same design, and could even share the same window. In fact, they shared the same source code since the structure of the Helix application mirrored the structure of a UNIX library (the better to support a uniform component model). This was a classic example of how it is sometimes necessary to introduce a purely artificial distinction (from an implementation standpoint) in order to support the correct user model. Here we see how effective our strategy of differentiation at the presentation level was in organizing the environment and helping the user remain oriented. By applying the same color scheme as the Motif palette (bottom) and a more "closed" default display mode to the Library Manager we virtually eliminate the chance for confusion between the two. This neatly handles our most important focal task, namely, "Make sure you don't edit the library when you think you're editing the application!"


OPEN LOOK Magnifier Patent


An interesting skunkworks project developed with Darrell Sano and patented successfully by Sun is shown here in its earliest form as a Director prototype. The design is for a moving magnifier that shows a different view inside the "glass" than the background over which it is moved (top). Typically this would be a "zoom" to a higher resolution but it could also reveal a different type of structure or level of absraction or class of annotation. The "MagicLens" technology announced years later by Xerox PARC did most of the same things and would probably require a license from Sun if the matter ever went to court.

One insight they didn't uncover was that you need a "spotting scope" mode when changing magnification so that you can see what you're aiming at when positioning the lens. Our design uses the heurstic that the lens goes into "spotting mode" as soon as it is being moved(bottom). In spotting mode, the glass shows the normal resolution with the portion that will be visible in "magnify mode" indicated by an aiming recticle of the proper size. As the zoom factor is increased, the reticle gets progressively smaller. The result is a simple dynamic "Focus + Context" viewing system that has a wide range of applications (most of which have yet to be exploited!).


DOE Workspace


In the early days of the Object Management Group (OMG) SunSoft and HP decided to collaborate on a standard desktop for UNIX that would unite the various company-specific OS layers under a Common User Environment. While the CUE product that eventually shipped (years later) was little more than a Frankenstein's monster combining tools from three different vendors (IBM having joind the mix by that point) with little or no design coherence under an OSF/Motif interface, the project started out as an ambitious venture led by object-oriented techology experts and distributed computing pioneers out to change the world by creating an environmet that supported "Desktop Objects Everywhere."

Sun's Human Interface organization ended up with the bulk of the responsibility for designing the next generation user experience. Kevin worked closely with Desktop Architect Kathy Hemenway to lay out the conceptual design foundations for the various desktop components and served as the design lead for the filesystem level components. He later represented the HI organization on the DOE Architects' Council and led the creation of a number of Director prototypes with Darrell Sano to illustrate the high points of the proposed design and build momentum behind a vison of what the user experience should be like in a world where applications largely disappear and documents are replaced by "Information Objects." (top and middle).

In addition to high level architectural repsonsibilities Kevin served as the design lead for the Desktop Foundation components that provide filesystem support for organizing information. In addition to the usual foldering representing a pure containment hierarchy, the design called for a new kind of container (bottom) that would hold only references to other objects. The new type was called an Index. It was distinguished from regular folders by some default layout characteristics and the selection feedback, which highlights entire items instead of just the icon. A given index window could set to be be static or computed based on a query-like rule determining the current contents each time the window was opened or refreshed.





The OPEN LOOK Graphical User Interface was a constant focus of attention and a formative influence on Kevin's philosophy of GUI design. Although the design was a joint venture by Sun and AT&T, Sun's Human Interface Technology organization was responsible for owning and maintaining the specification. A great deal of time as spent helping Sun's application developers (and a few ISV's) maximize compliance with the various usage policies and application style guidelines. We also worked with 3 or 4 toolkit teams, each providing a slightly different realization of the OPEN LOOK standard on top of their particular platform technology.

The HIT squad was also responsible for extending the existing spec where necessary. The work shown here represents a proposal for extending the OPEN LOOK visual language to provide the true "3d" representation that was being sought so eagerly in those heady days following the release of NextStep. These examples make a clear distinction between data and control areas and are designed to give a special prominence to Sun's patented pushpin mechanism.

Sun Bugtool  

Kevin was delighted to join the Human Interface Technology group at Sun Microsystems in 1990 after a year in Berkeley working as an Associate of Aaron Marcus. The group was made up entirely of psychologists and, as a learning experience, was like another two years of grad school. One his most influential designs at Sun was a simple refactoring and relayout of Bugtool, the company's vastly complicated bug tracking application (until broken down into the pieces seen here, the property window filled an entire 19-inch workstation screen), which he completed with Jarret Rosenberg. The Bugtool team was impressed enough to come looking for us (this was in the days when progammers never went looking for UI designers!) two years later when they were ready for a major revision and re-architecting of the product.