A Few Words about Lotus

[This post is a response to Chapter 49 of Steven Sinofsky’s wonderful online memoir, in which he talks about competitive pressures facing Office in the mid-90s. Unfortunately you have to pay a subscription fee in order to comment, so I’ll comment here instead.]

Steven,

Excellent and thought-provoking piece – brings back a lot of memories. During the time described, I was a senior developer at Lotus and I’d like to offer a couple of clarifications.

Re components: you (understandably) conflate two separate Lotus products, Lotus Components and Lotus eSuite. The former were a set of ActiveX’s written in C/C++ and were targeted at app construction. My late friend and colleague Alex Morrow used to talk about developers becoming segmented as “builders” and “assemblers.” “Builders” created the ActiveX controls, and “assemblers” stitched them together into applications. The idea persists today in the form of Logic Apps and so-called citizen developers. We at Lotus had some brilliant developers working on Components but I suspect the concept proved premature in the market.

Even more ahead of its time was Lotus eSuite, which was a set of Java applets designed to run in the browser. eSuite got its start when a developer (actually, me) ported Lotus 1-2-3 v1 to Java as an experiment; Lotus and IBM loved it because it was perceived to be disruptive against Office, which while not yet dominant threatened Lotus’s SmartSuite.

Ironically, however, eSuite ran (by far) the best on Windows and IE. I recall attending the first JavaOne, where, at a breakout session, Microsoft demonstrated its rearchitected JVM and Java libraries – vastly better in terms of performance and load time than the Sun-supplied versions. (This was partly due to the fact that where Sun built the Windows libraries on MFC – pretty clunky at the time, Microsoft wrote to Win32, essentially, right to the metal.) And, of course, the IDE, Visual J++, supported the UI nuances and superior debugging experiences that we’d come to expect. It really was, as you quite rightly say, a tour de force.

But it was clear to us at Lotus that Microsoft had mixed feelings about it all. I and several others traveled to Redmond (aboard an IBM private jet no less!) to talk with Microsoft execs about the future of NT and Java (why NT? Because at the time the Lotus Notes server was one of the key – if not the key – driver of NT sales). In a day full of briefings in Building 8 Charles Fitzgerald, then the PM for VJ++, came last, and for that we were joined by BillG, who couldn’t believe we were building “serious apps” on Java. (He told me I was “on drugs.”)

I always thought Microsoft’s abandonment of Java was a bit of a shame: I’d written an email to David Vaskevitch (then the Microsoft CTO) suggesting that Microsoft’s superiority in development tools and frameworks could be used to essentially isolate competitor OS’s – essentially wrapping Solaris in a Microsoft layer. I never heard back.

As it happened, we did ship Lotus eSuite – and it remained the case that neither Macs nor Sun workstations could compete performance-wise with Windows. (To this day I’m stumped why neither Apple or Sun didn’t try harder to make the technology competitive – it was existential.)  And JVM and browser technology were at the time still evolving, so what worked on one platform wasn’t really guaranteed to work on another (belying the “write once run anywhere” slogan).

eSuite also suffered from a particularly stupid design decision in the JVM (which I believe Microsoft was contractually obliged to implement as well). In order to prevent code from jumping out of the sandbox the JVM, on load, analyzed every single code path before launch. For an app like a spreadsheet, which has hundreds of functions, recursive recalculation, directed acyclic graph, etc., the performance hit was murderous. I recall wondering why Sun et.al. couldn’t use digital signature to implement trust but they never quite got the idea.

Anyway, the time for running productivity apps on the browser, unquestionably a great idea, hadn’t hit yet.  (It has now.)

Comments are closed.