On IBM. Happy Birthday alphaWorks!

I see by Stephen O’Grady’s blog that this week marks the 10th anniversary of IBM’s alphaworks site (aW). [1]

aW was the inspiration of Irving Wladawsky-Berger (known to everyone within IBM as just “Irving” or IWB). It was created not to provide puffery about work in progress but to actuallly provide working implementations of new ideas — a way to throw stuff over the firewall so people could give it a sniff and decide whether to eat it up or to walk away holding their nose. Hence the name: “alpha” is the step before “beta”.

alphaWorks set up shop two months after Jikes had reached the “hello world” stage [2], and when we learned about aW soon after its opening our goal became to get our code out, to see what the world thought.

We had Jikes ready by early April, and just before its release I sent an email to Peter Coffee, a PC Week reporter whose work I had admired. I received my weekly copy of PC Week in paper form those days [3], and just after I got the next issue I got an email from someone saying there was an article about Jikes.

Lucky me! I poured a fresh cup of coffee, skipped past the front page, and went over the rest of the issue page by page, looking for the little, teensy-weensy, box that said “IBM alphaWorks today announced the release of Jikes …”

And you know what? I couldn’t find it. So I turned back to the front page. There was indeed an article by Peter Coffee, something along the lines of “Super Nova announces new Java technology…” Which meant that Peter had found better things to write about. So I tried again and still couldn’t find that itsy-bitsy article about Jikes.

So I went back to that front page one last time, to see what Peter had written,. What had SuperNova done that made Peter decide Jikes was just a hill of beans? [4]

And you know what? Buried in that article was a throw-away section with a couple of sentences, “IBM’s alphaWorks division today released a new Java compiler. It’s called Jikes and it sets new standards for conforming to the Java language specification…”

Now of course I wasn’t the only person reading that front page. I was the only one who read it to learn about Jikes; the others were reading it to see what influential writers thought.

So it should come as no surprise that within a few days Philippe and I were making presentations to senior management to tell them about Jikes. These were folks we had seen before, but had never spoken to. In short we were now “playing with the big boys.” [5]

Those big boys let us continue working on Jikes. Although Jikes was well-known for its speed [6], our focus was on language correctness. We found that we got about one good bug report per thousand downloads. We first heard about Linux in 1997 but weren’t able to give it serious concern until a year later [7]. Linux offered the promise of more downloads — and hence more bug reports — so one of us spent a week reading the GPL [8]. and then wrote a memo to management requesting approval to make Jikes available in binary form for Linux.

Jikes for Linux (binary) was released in mid-July 1998; it was the first program written by IBM release for Linux. It resulted in the biggest splash then yet seen at aW. — true exponential growth [9]. We were Slashdotted. [10]

We knew that releasing a binary would inevitably result in a request to release the source code. This effort was approved just over eight years ago — two years after aW opened.

Jikes in source form was released in early December ’98, and set new aW records for downloads within hours after its release. [11]

Jikes was the first code written by IBM released as open source under a license written by IBM (later approved by the Open Source Initiative). It was later run as an OSS project from a website with an address ending in ‘ibm.com’, using the open source “rules”.

IBM followed up alphaWorks (aW) with developerWorks (dW) three years after aW’s opening. Jikes was the initial offering from dW’s “open source zone.” From the day it was first available, Jikes enjoyed the same success as Apache http server — it enjoyed market share was greater than the combined share of all the others.

As is evident by all of the above, I would not be writing this post now if aW hadn’t been created …

• aW put Jikes on the map;
• aW directly led to the best success I — and I think it fair to say Philippe as well – have yet enjoyed at IBM;
• aW directly resulted in Jikes becoming the first IBM-written program to be included in a mainstream Linux distribution (Red Hat);
• aW led to all the the other successes, such as on dW, inclusion in SUSE, and on and on, and
• aW is an ongoing demonstration that the internet is a meritocracy, one that separates the “dogs” from the show-stopping champs [12];
• and in particular, aW resulted in a project that continued even after we had to stop working it back at the end of 1999; and
• As a final example, it was aW that directly caused my son’s recently telling me that someone wanted to meet me. It turns out that my son — a fourth-year rabbinical student at Hebrew Union College / Jewish Institute of Religion– was at a wedding of one of his fellow students. The groom happened to be a developer. He uses Jikes every day, and on learning that I — me, moi, myself , my son’s dad –was a co-author of Jikes, said he would like to meet me.

So, as we in the Shields family say, “bust my buttons!”[13] and so therefore let it be said:

Happy Birthday alphaWorks! Keep up the good work.

Footnotes:

1. I knew aW’s 10th was coming up, but I had forgetten it as I wasn’t involved. Seeing Steve’s blog reminded me it was happening. This was an instance of someone outside IBM telling me something about IBM that I had forgotten; more usually it’s that someone outside IBM knows how to help me find someone inside IBM. For example, in early August ’98 I wondered if a single individual (me,moi,myself) could run a project like Jikes, so asked Bruce Perens if he knew anyone who might have some insight. He said I should call someone called Brian Behlendorf. I did, learned that Brian was part of “apache” and said I had IBM was involved with Apache but that I didn’t know how to find these people. He gave me a couple of names.

2. “hello world.” is the most commonly written program in the world. All serious programmers write it at least once each day, and over their careers will realize they have written the same damn simple program for an extranordinarily large number of reasons. For example, in unix, it is just echo hello world, which is easy as pie; in Ada it is about as bad as it can get, since you need to get generics working to handle this simple sucker.

3. I get almost no paper mail. It has reached the same obscure state as most bloggers: knowing that no one will read it, then why write it? For example, I recently changed buildings in Somers. Friday I got a call from the mail-room asking where they should forward my e-mail. I said I only got about one piece a month, so there was no rush — just send it when they had the time.

4. It might have been the dot in dot net folks, or it might not. But anyone like myself who has ever worked in the Java space that is not on the Far Side of the World will take every opportunity to have some fun with the bean-counters.

5. “playing with the big boys.” See Jikes Coupon.

6. Most foks came to Jikes because it was fast. We only cared about getting it right, but knew that speed would help sell Jikes to mgmt and bring us new users. See the work of Clayton Christensen for the implications of all this…

7. Jikes came out in April. While getting Jikes ready for release we learned in February that Sun had announced a new language feature, “inner classes.” An initial review suggested it would be hard to implement. We put that aside while we got Jikes ready for aW. Re visited the matter after the aW April rush and realized that it really was hard. So we shut the project down, and delivered no new release between July and the following March or so. It was this language extension that separated the men from the big boys. Before inner classes, anyone with good “dragon knowledge” could get the job done; after it, the number of serious players shrank drastically. While Jikes has labored on, another language changed called “generic types” (or such, as my Java skills are rusty), may prove sufficiently difficult to implement that it will finally kill the Jikes project. It could be argued that keeping the language simple — avoiding the descent into DUDness — might have been the wiser course.

8. It’s a difficult document to sort out.

9. I have charts of this. It was phenomenal. I bought a scanner a few months ago. When time permits I will plug it in and scan those charts.

10. I’m the IBM “rep” for Slashdot. I love these guys. That’s why I write about them so often.

11. Jikes was announced at 1PM EST on 7 Dec 1998. I was at the phone so the following events would happen with the next minute: IBM keynote speech by Pat Sueltz would include mention of Jikes, alphaWorks would open up the ftp server, Slashdot would post a story, LinuxToday would post a story, and PC Week would release a story about the announcement. (I had given Peter Coffee the “inside skinnie” about the Jikes announce in gratitude for his help in the initial aW announce; this caused our press folks some grief but Peter’s putting it yet again on the front page helped assuage their anguish.)

12. In a prior post I wrote about the neutrality of the net: you are judged by what you do, not what you look like or what you say. I’ve done many of the OSS proposals for aW (most come from Research, and as an ex-IBM-Research RSM I handled a lot of them for a few years, usually working with Jim Chao of the aW team.

A couple of years ago there was a proposal from someone in Japan, a technology for helping disadvantaged folks read a screen. I worked with the proposer over several days. Due to time zone differences we never spoke by phone. We used e-mail and instant messaging. The proposer’s language was always correct English and very courteous, and the instant messaging proceeded at a normal rate. What was unusual was the hours during which this happened. I work odd hours myself — it’s 1:30AM as I write this — and Japan is on the other side of the world. And I realized after a while the proposer was “always on” in that their hours were as unusual as mine.
Work went on at a normal pace, however. I mentioned this to Jim Chao a couple of days later, and he informed me that the proposer was Chieko Asakawa, a blind Japanese woman who was one of the world experts in this area and was also a member of the IBM Academy of Technology, an august group of IBM’s 300 or so of IBM’s best technical experts. This explained her unusual hours: to her night and day were the same.

13. The welcoming entry given to those who ask to see Oz. Not the creator of Notes, but the one behind the curtain.

PS: I have followed Stephen’s blog for some time. I’ve also been running an experiment in the “future is now” issue that has given me new insight, and has also made me appreciate that Stephen is almost up in the Tom Friedman league. I’ll try to write about in … whenever