Guest: Ken Gardner - SOASTA
Host: Robert Scoble - The ScobleShow

Robert Scoble - The ScobleShow
So, who are you?

Ken Gardner - SOASTA
My name is Ken Gardner and I am the Chairman of SOASTA Incorporated.

Robert Scoble - The ScobleShow
And this is like pretty cool place; I was walking in here its right by Google, and right by Mozilla Foundation.

Ken Gardner - SOASTA
We're hoping that just a little bit of that (Inaudible). Yes, yeah this is -- so we are block away from Google, Mozilla is in the building next door. So, and this whole entire complex is startup heaven.

Robert Scoble - The ScobleShow
And you got some cool offices here.

Ken Gardner - SOASTA
Yeah.

Robert Scoble - The ScobleShow
So, I'll just playing around here for a little bit and so there is a whole bunch of startups all over this complex.

Ken Gardner - SOASTA
Yeah, this is called the landings, and this has been here since the late 70s, so it's got that very granule of trees inside atrium kind of feel and this has been startups for, as long as I can remember so lots of small 3,000-5,000 square foot suites which are just ideal for that very beginning stage when you are building something from scratch.

Robert Scoble - The ScobleShow
So, what do you guys do?

Ken Gardner - SOASTA
We are building a product where we're in alpha stage right now, we are building a product that's going to solve the problem of automating the testing of very large Web applications where they're based on SOA or REST or Mashups or any of that stuff. The idea is to provide a set of tools that will allow you to get to a 100% automation of the testing of that. The issue that we are trying to solve is that whether it's salesforce.com or Oracle's applications or SAP or things that I'm doing across Google, Yahoo! and other stuff, the number of permutations gets very large, very fast. Salesforce.com has in their standard product 63 data objects and 31 verbs (ph) and if you thought that you were going to get about a 100 messages to get good testing coverage that yields a 126,000 XML documents that you have to produce.

And the current set of tools in the previous generation, we were users of those tools was, I go into an XML editor and I hand create the stuff and that's very close to being in hell technically. It's tedious and error prone and if you miss a comma or a semicolon, you are dead and so our approach is to take a look at the pieces of that problem, and create text specific Visual Editors that conquer each of those individual problems. Our metaphor for doing this almost every testing tool you'll see in the market today, is script based. Okay? So, whether it's Mercury, or IBM or Borland or any of them, my scripting language is better than your scripting language, but they're script language based, that's not the approach that we are taking. You are a final Cut Pro-User, that's our metaphor. Okay, so instead of...

Robert Scoble - The ScobleShow
What is a Video Editor, you have to deal with?

Ken Gardner - SOASTA
Okay. So, the job and testing message based systems, as I need to create a complex event stream, and I create that complex event stream in our product by instead of creating video clips or audio clips, I create a message clip and that message clip has a series of messages in it, with optional behavior, but it's a series of messages. And I'm going to then play those messages with certain timing and then when I go to assemble the final message, and we'll show you this, it's very cool product. You lay these out on a multi-track timeline just like you would in Final Cut-Pro and you hook it together, and you play it and out of the back-end becomes a repeatable message, a repeatable complex message stream that I can stream at my Web service in very high volume, in order to do functional testing, performance testing, capacity testing, and stability testing.

And so the idea we had was to just completely pivot the problem, and change it. Let's borrow, there is a piece of our product that would remind you a lot of final video composition or audio composition in Final Cut-Pro or Garage Band. There is another part of our -- but everything is designed to be visual and then one of the big things is to then generate messages for you. So, that you don't have to, can you imagine how long it would take somebody to sit down and create a 126,000 XML docs. By the way, Salesforce that's a small problem in this area, if you think about future implementations of SAP, NetWeaver or Oracle Fusion applications as those come out, you get numbers just your regression test is going to be six-seven million messages. So, the complexity of that now....

Robert Scoble - The ScobleShow
So, who's the customer for your product, is it the SAP's and Oracle's or is it people building stuff on?

Ken Gardner - SOASTA
In anybody so (Inaudible) says that if that IT system can affect the income statement then you have to certify it, and which is bad enough by itself. But it also says that if you make any changes to it, you have to recertify it before you put it in the production. So, now what the SEC the government is really doing here, this is a personal opinion, but what they are really doing is they are imposing the same rigger on corporations that they have imposed on banks in the early 60s. Everything has to balanced to the penny, you have to be able to prove to me that your process will balance it to the penny, because I don't want any surprises.

And there was a little while last year where everybody thought what we were going to get relief from Sarbanes-Oxley and then the Enron Trial started and all the support for that went away and now basically, they are saying, "No, you're actually going to have to do this." Now, we all as engineers, we all know we ought to be doing this anyway. One of things you'll find in the Web services world, because it is so hard to create messages. Okay particularly create messages that show a certain scenario, because of that ...

Robert Scoble - The ScobleShow
(Voice Overlap) messages by the way.

Ken Gardner - SOASTA
Okay. So, the modern applications what we're -- the way that you do work instead of making a procedure call, okay as we would have done in a previous generation. I'm going to formulate an XML message and instead of making a procedure call to create a customer. I'm going to send a message across the Internet to stand the customer. So, if you look at the verbs that are in salesforce.com messaging API, things like login and create, delete, update, query, lots of verbs like that, lots of -- and then all of the various data objects, customer contact, order, all of those kinds of things. So, verb object pair and then a pretty complex data structure, that comes with it. I saw in a video, you did when you were at Channel9 of the Mapper, you look at those messages and it takes your breath away, the complexity -- now imagine, you've been given the task of creating the complete regression test about 2000 data elements repeating groups, in a very, very complex structure, that's a challenging task, if you are in a text editing environment.

Now, and so the idea here is to start to ship away at the complexity of that, and to reduce that complexity by -- we'll take that away from you, you don't need to know XML. We'll capture the Web services description language directly from your Web service. And then at that point, when you want to create an account, we already understand the data structure and we can present that to you as a data form that you can fill in. Now, that's the sort of the foundational stage for us.

The next stage is to then -- because we have those data structures is to do message generation, where you give us a set of rules. I want 10,000 messages and I want to vary them, I want to vary the values and I want to -- and then have us do all of the hard work, and then you can then just come in, just like in video editing, hand do the final transitions in the editing. And so, we want to move away from what we've been doing for 30 years is, we've been writing codes to test our (Inaudible). And so, the idea here in a message environment, is now let's borrow of some of the great user interface work that's been done, playing a song and playing messages at a Web service, and playing a note and sending a message or one for one in the metaphor. It's not more complex than that, it's not less complex than that, but if I have a completely mixed complex piece of music, I might have 16 tracks, they are all contributing to that whole thing, all that various volumes and velocities and beats and all of that and the metaphor halt, what's probably just as important as, we've gone out and made an initial contact with the marketplace, people get it.

Robert Scoble - The ScobleShow
Yeah.

Ken Gardner - SOASTA
Which is visual our computing guy, that's huge.

Robert Scoble - The ScobleShow
Right. Why did you get into it? So, you have told me this was like your sixth startup?

Ken Gardner - SOASTA
Just in we had a company called estonte Software and estonte built a real time dashboard, ran in Internet Explorer using AJAX before we had a name to call it.

Robert Scoble - The ScobleShow
Yeah.

Ken Gardner - SOASTA
And we delivered a real time dash board through the browser, okay. And we got purchased by PeopleSoft, nine hours before Oracle bought them.

Robert Scoble - The ScobleShow
Oh wow!

Ken Gardner - SOASTA
So, that was an interesting experience of its own, but we had a message based architecture, so we had the problem of testing and demonstrating this application ourself, when we got to Oracle, I went searching though -- so Oracle, Microsoft, IBM, SAP, BA all these companies have made this huge bet going forward on message based architecture, as it's the way application are being built on the Web and we can argue about whether it's SOA, or REST and all those sort of normal, technical hu-ha that we always have.

Robert Scoble - The ScobleShow
Academic discussions.

Ken Gardner - SOASTA
Yes, yes very passionate academic discussions, but we are going to do -- well this is going to be a message based going forward. And Oracle, Google, big organizations, everybody is building these tests and hand coded XML documents and Perl scripts or they are writing Java code or they've got an in-house team of four or five people that are running their test harness. And there is a tremendous amount of pain around this issue. And so, after I got out of Oracle, I knew this was a problem, but at that point I didn't know how it's gone solve it. And I got converted to the Mac, I started playing around with iMovie, which then led me to GarageBand and I'm sitting there one day playing with GarageBand and I thought well, "This is it," this video, this digital media user interface, I'm basically creating this composition, where I can control timing, I can control order, I can control placement. I can do things all at once, this will actually work. And I then spent 90 days talking to a lot people, asking them to talk me out of it. One of the things you have to test is, when you come up with a crazy idea is...

Robert Scoble - The ScobleShow
Just how crazy it is.

Ken Gardner - SOASTA
Just how crazy is it and I ask that question all the time still, but what seemed pretty clear at this point is, this works. The product installed at three office. We were probably about four weeks from going to beta and the product will go.

Robert Scoble - The ScobleShow
Ship fairly early stage to startup that right.

Ken Gardner - SOASTA
Yeah, but we are kind of doing it the same the way that a lot of people are doing it now. Where you open it up and let people start to use it very, very early, so that you can get people to tell you what's wrong with it or what -- because you are never going to get it more than about 70 % right the first time. So, you want to have really smart people who have the problem say to you, "Well Ken this is great, but this isn't, you don't have this correct." And so a former partner of mine used to call this, The Collision with the Real World, which all software has to go through. So, we started the company in February 1st.

Robert Scoble - The ScobleShow
Yeah.

Ken Gardner - SOASTA
The, the group of people that are in the company for the most part, we have all worked together across many of those previous startups.

Robert Scoble - The ScobleShow
Okay.

Ken Gardner - SOASTA
And so, we know each other, we got off to a fast start. I would say that something that's remarkably different and better about software development right now, is the amount of cross industry collaboration that you see. And I don't mean just Open Source, we're MySQL, JBoss, we're certainly using the Open Source stake, but that little things like AJAX controls, where some developer somewhere has done a reference implementation of some little piece. And then has shared it and has given a license for everybody to use. And so, you have a much larger amount of Mental Bandwidth being applied to the issues of creating rich applications in the browser and it is cross company. And down at the widget level nobody is competitive at that level and people are sharing in a pretty remarkable and I would say, very spectacular way.

Robert Scoble - The ScobleShow
Where is your (Inaudible) place to go look for a new stuff?

Ken Gardner - SOASTA
One of the blogs I look at everyday as Ajaxian and one of the things that I love about that blog; is that they have -- that's a blog where they have about eight or nine guys, who are contributing articles to that blog and they're deep. They are not scaring (ph) the service they are going down, okay, in Firefox 2.0 with Dojo I have the following issues and or in this guy over here solve this problem and you can go look at that. And it is very, very different. I've been doing this software development since 1972. And this is by afar, the best time that to be doing in it and looking, "Oh, yeah."

Robert Scoble - The ScobleShow
Really? You think has changed that much. We always had news group and that we always had places to go and share information and but certainly here in Silicon Valley we always had a user groups. I remember going to Visual Basic user group.

Ken Gardner - SOASTA
I worked at a computer services company call Timeshare back in the late 70s until the mid 80s. And I worked in the group called, 'Applied Communication Systems,' and all of the things that we are now realizing, people were discussing in the late 70s and early 80s. What's so different is that it now because...

Robert Scoble - The ScobleShow
(Inaudible) was discussing it in 60s (Inaudible) right.

Ken Gardner - SOASTA
Doug Engelbart was at Timeshare when I was there. Okay, Augment Group ended up being acquired by Timeshare. They set in a group of cubes next to me. He is one of the world's smartest people. Yeah, he was thinking about it in 1963. Yes.

Robert Scoble - The ScobleShow
I had much for them earlier this year and the guy is just so smart.

Ken Gardner - SOASTA
He is a remarkable.

Robert Scoble - The ScobleShow
How did he get so far, out of the rest of us?

Ken Gardner - SOASTA
I don't know the answer of that question, but he certainly has always been way ahead of everybody. And it's got to be really gratifying for him to see this. I've always thought of technology has being wish fulfillment. It's like iChat. iChat is very much like the old wizards things where you could look in and see if somebody far. When I am on the road, I iChat with my kids, which make that works a lot better than in previous generation to this. So, the resulting systems can be substantially better.

Robert Scoble - The ScobleShow
Now, is your system a browser based system or (Inaudible)?

Ken Gardner - SOASTA
Yeah. We're -- or let me just create a bunch of messages for you. So, that's how you create messages in our systems. So, we run in Firefox and IE 7.0 at this point. By the time we release we'll support Safari and Opera, okay.

One of the things that you wanted to do in a testing environment, as everybody has moved to geographic collaboration whether it's to India or China or just multiple sites across the United States, you want, lots of people to be able to contribute to the development effort and the testing effort as part of that. And so, the idea here is, we want to provide a rich experience that's based in the browser, so that everybody can jump in. All I need is a user name and credentials and I'm in. And I might need to be trained, but hopefully the UI will give you pretty substantial amount of help to that.

Let me just really, quickly jump to -- I'll show you a composition and you'll get I think really quickly, the video editing part of the metaphor here. So, I have -- these are message clips. So, and then I can by dragging up from the bottom, I can add something else on a separate track. And those are then; those messages are going to be played in parallel. So, the server has built to do this from the beginning. I can change the timing if I want to send these messages more quickly. I give it less time and so all of a sudden the number of messages, I am sending per second.

Now, the idea here is that each one of those message clips, corresponds to unit test. Okay, and so if you set that up and you create a taxonomy that maps to the modules in your system then all of our, all the automated test validation, the reporting will come back to you by unit test, okay. And so the idea here is to do -- so the alternative to this is to write code, to do everything that I just did. And if you are going to -- I want to send 40,000 messages a second against that Web service in order to put it under a stress and to find out how its going to behave and if it will fall over -- if it will stand up to it, that's pretty challenging to do in code. At that point, you are not writing a script, you are writing something that's threaded and paralleled. And we'll do that for you.

The other part of the innovation here, we're going to deliver this as a service, over the web. So, if you are a small startup, our size startup. You're not going to go to a big testing tool that cost hundreds and thousand dollars. You just can't just afford that, but we'll let you use our tool for a couple of hundred dollars per month, per tester. We'll sell it as -- so we're going to use this software as service model, we'll also sell it there some organizations who believe it their data is so important that it could never be outside their Firewall. Yeah, so fine, we'll sell it to them as a subscription, but so we want to innovate on the business model aside of this. So that it not so expensive to get started. This is that whole clue train idea where, okay you have to actually make it work and you have to actually provide value and then if do you'll get something back from that. And I'm perfectly fine with that, I think we're on our way. We seem to found a wide space in this market, for reasons that I don't understand -- nobody got funded to solve this problem.

Robert Scoble - The ScobleShow
Why do you think that is?

Ken Gardner - SOASTA
Well, I think there was this I've heard venture guys that I know talk about, there was this pause from about the time the bubble burst to about last year, where there were people who believed that the venture opportunity in enterprise software was over. I don't believe that and I don't believe that's going to be true, as long as, (Inaudible) holds up, right. As long as we get doubles every 18 months that means every seven or eight years I have 10x, 20x, the power and I can do stuffs that I couldn't do in the previous. I couldn't do things that Doug Engelbart wanted to do in 1963. I can do them now. And that will continue to be true and when that happens, that's happening now, we switch architectures. We went from Mainframes to client server. We're now going from client server to composite applications based on Web services. And every time we do this, it shouldn't be a big surprise this is the third it's happened.

You look around, you open up the bag of tools and the bag is empty. Oh! I want to test complex composite applications where there's potentially trillions of permutations and I want to be able to very, very quickly build a set of test that proves that this thing is actually work when I put it out in the real world or mostly work. When I put it down in the real world and nobody has been working on that problem. And so, this is the third -- one of the architectural switches, I have been through, they're kind of fun, they're kind of fun. And it represents an opportunity almost never, I wouldn't say never, but almost never the existing companies in the space, the testing companies, the application development companies, almost never provide the tools for the new generation. So, going back in my history I was I had a...

Robert Scoble - The ScobleShow
That's the innovator (Inaudible).

Ken Gardner - SOASTA
Oh! Yeah.

Robert Scoble - The ScobleShow
So, you're serving an existing customer base and you can't, you can't serve a small thing, and the approach that you have taken to the marketplace. If I've got a big company that depends on an enterprise license model and I'm selling software for a $0.25 million ASPs then the idea that I am going to come offer this to single developers for a couple of hundred dollars a month, isn't going to be amazingly interesting to me. For me that's a great way to build the business. If you go back to when Windows beat OS2, because I was there. Windows beat OS2, because IBM picked a dozen partners and it was called the AD/Cycle partners.

And if you are running startup at that point and you weren't one of those dozen companies that got picked to be an IBM partner, you partnered with Microsoft. Microsoft got 8,000 developers and got 12 and I think you are going to see that same thing here. Out of those 12,000 companies doing Mashup and social communities and all that stuff, all of them by the way are potential prospects for us. There are going to be some very important, very large companies come out of that and certainly a tremendous amount.

Robert Scoble - The ScobleShow
You already have.

Ken Gardner - SOASTA
Yes, yeah absolutely. (Voice Overlap) No kidding. Well that happened really quickly. Didn't it?

Robert Scoble - The ScobleShow
(Inaudible) six billion and they didn't demonstrate they had business model, but they there are more around the Google's business.

Ken Gardner - SOASTA
Yeah. There is some code I can't remember or should I tribute somebody, but I can't. I would just like to have one more bubble in my lifetime, only this time I want to sell. So, I don't know that this a bubble, I think this is the beginning of a completely new wave this wave it is going to -- becoming upon the beach for the next six, seven, eight years as everybody makes the transition. Certainly, at the level of integrating things, I can get access to your back office system and put data into it, from your customer or your partner or a third party like FedEx or UPS. All of a sudden, it gets easier to make sure that everything is updated the way that it ought to be. The issue that gets to you is essentially 100% of the time, doing integration testing.

We have a prospect that we've talked to where he collects leads for people on the Web. He becomes your registration page and then he cleans it up and eliminates all the dark daters and Mickey Mouse and all that. And he then integrates to peoples back office system, to hand those off to them in an automated way. He is doing continuous integration testing everyday of the year. He has got a code line, he's got 40-50 other companies who could be changing their code line. He needs to be able to very, very quickly direct go through all of those messages and find out the one that broke. And get it fixed and get it fixed very quickly. So, it's easier in a certain sense, but it's we are adding, we keep adding layers of complexity. So, fun stuff.

Robert Scoble - The ScobleShow
Tell me around your company a little bit.

Ken Gardner - SOASTA
Okay. Yeah this is our Galactic Headquarters.

Robert Scoble - The ScobleShow
One thing I noticed, when I walked in, everybody has these very beautiful 30 inch Apple monitors.

Ken Gardner - SOASTA
Yeah, one of the things that I am -- what Apple has brilliant done, is become the platform of choice for developers, because on the Intel MACs, I can run Windows XP, Windows Vista, MAC OS and all forms in Linux, all at the same time. Moving back and forth.

Robert Scoble - The ScobleShow
You're using parallels?

Ken Gardner - SOASTA
We are using parallels, yeah. And we were in that jumped on that initial bay. They came along exactly at the point that we needed them. It was really great and so all of a sudden instead of -- we started of buying, we are buying Linux boxes and these boxes and probably we are going to end up with each developer having three or four machines. And we're now what we have is -- the developers have MAC Pros and 30 inch monitors here. And then everybody has a MACBook Pro for at home. So, the developers have these screens both at home and here.

Robert Scoble - The ScobleShow
Wow, that's pretty cool.

Matt Solnit - SOASTA
Absolutely.

Ken Gardner - SOASTA
Now this is Matt Solnit. Matt is ...

Robert Scoble - The ScobleShow
(Voice Overlap) can't hear you so. You can take my mic.

Matt Solnit - SOASTA
Thanks.

Robert Scoble - The ScobleShow
Sorry for the wild camera moves.

Ken Gardner - SOASTA
Matt is one of our server engineers and he is a -- we are a Web services based application ourself.

Robert Scoble - The ScobleShow
Okay.

Ken Gardner - SOASTA
And so Matt is SOASTA Concerto, which is the product name, is composed of a repository service, which is what you used to build tests and then there's a maestro service, which plays the test and Matt is evolved in that site. He was also at estante software. Matt just recently got married as well.

Robert Scoble - The ScobleShow
How do you build this stuff? What tools do you use on your Mac?

Matt Solnit - SOASTA
We're building using J2EE the Mac is a really friendly Java environment, so we have Eclipse and JBoss and MySQL and all the other Open Source friendly stuff that you can just get going with. It's been really easy, there was a slight learning curve of about ten minutes figuring out how to do this Mac stuff an ...

Robert Scoble - The ScobleShow
What platform were you on before?

Matt Solnit - SOASTA
Windows.

Ken Gardner SOASTA
We were on windows, the estonte product was built on top of the windows platform. For 15, 20 years I drank the Microsoft Kool-Aid.

Robert Scoble - The ScobleShow
Oh! Yeah, me too.

Ken Gardner - SOASTA
Yeah, yeah. (Voice Overlap). And nobody should count them out because its -- that would be a huge mistake, but I think if you are doing things on the Web and you want to the largest possible audience, you are going to go to this Open Source stack. It gets you the ability. We had an alpha running five months after we started and I have never been able to do things in that cycle and of course, you have to just like going to the junkyard you got to sort through lots of parts, but every once in a while you get a clean carburetor for a 57 Chevy and that fits right into what it is that you are tying to do and you just get this tremendous ability to build a complex system very quickly. So, and it's been, I think it's been a lot of fun what you think?

Matt Solnit - SOASTA
Yeah definitely.

Ken Gardner - SOASTA
Yeah.

Robert Scoble - The ScobleShow
What are the challenges of this new system that you are working on?

Matt Solnit - SOASTA
Well J2EE was a challenge at first, we were all .NET guys so we took a lot -- well we bought a lot of big, thick books.

Robert Scoble - The ScobleShow
You miss anything from .NET?

Matt Solnit - SOASTA
I won't really go there, yeah there is some things.

Ken Gardner - SOASTA
Now come on, come on, come on.

Matt Solnit - SOASTA
So, I can definitely appreciate the J2EE means we are coming on any platform. We're never going to have a customer tell us, Well you don't do Solaris so I'm not interested. There are some things that are a lot harder in J2EE, so the new Java EE 5there's a long way towards correcting a lot these issues and we are not ready for that yet, because we want to be able to run on some of the older application servers, but that's Java EE 5 goes a long way to (Inaudible).

Robert Scoble - The ScobleShow
What do you (Inaudible) stuff in your code? If another geek locked in, what would show him or her?

Matt Solnit - SOASTA
I would show him, probably the way that we are using hibernate, to make all the database stuff really easy. I was a really big SQL guy in the past, I've always like things very close to database and I would never touch an ORM layer, but hibernate lets us drop...

Robert Scoble - The ScobleShow
(Inaudible) ORM is?

Matt Solnit - SOASTA
ORMis Object Relational Mapping, which I never found it really took off in the .NET world, but it basically lets you create your class hierarchy in Java or in .NET and have this intermediate layer, that mapped on to relational database tables, without you're really doing any of the work. So, you say, "Okay this class corresponds to this table and this property corresponds to this column," and it just does the rest for you. So, right away we're database independent, we run on MySQL, we run on Oracle or we're on (Inaudible) any of those guys. And I can change our database structure writing Java code and it's so easy. I never thought that it would be that exciting, but it's just made us so much more productive.

Robert Scoble - The ScobleShow
It sounds like Microsoft has a lot to worry about.

Ken Gardner - SOASTA
Well, it's interesting, the way I would characterize it is, Microsoft has the best application development in the tools, tools in the world, but they only run on their platform. And there's a real strong push on the Open Source side for this wide collaboration and that's a real powerful thing and so who wins ultimately, the world's kind of split 50-50 now, right? So, you're either a .NET guy or you're over on the Open Source side, almost all the startups that I see though are going the Open Source and as I was discussing before this...

Robert Scoble - The ScobleShow
There are a few still that are .NET.

Ken Gardner - SOASTA
Right. Reference implementations of all of these widgets that you can then pull together quickly to assemble your unique value proposition, is just as powerful I would say, as the ability to very, very quickly do things in .NET. And I don't know I worked at -- I was in early 90s, I was the VP of Products at Borland at one point and was there when Delphi was released. And that really, really beautiful stuff and that pulled everybody to the Microsoft platform for a very, very long time. This opens standards and open collaboration the real powerful stuff and I don't know what Microsoft's response is going to be to it. I think they have to respond at some point. And that's obviously a real hard tough decision -- it's cannot like very analogous to IBM when they had Mainframes that were very high priced and everything started to go PC's, at what point do you make the break.

How do maximize the return to your investors from that switch, because that switch is happening and if you wait too late to jump over. In earlier transition, I had a company that was at the Windows 3.0 launch event now, this was 3.0 not Windows 3.1. There were 25 companies that presented that day; there was just little theater in downtown Manhattan. I met Bill Gates that day, I met Bill Gates' mom that day its pretty clear, it wasn't quite all there about nine months later, 3.1 shipped and they took over the whole world. Lotus and dBASE and Borland dBASE and software publishing, the runaway category winners in a previous generation, completely missed that transition. And Microsoft took over that business and that's the tradeoffs they've got to play with. So, tough problem, I'm glad I don't have it.