Cubegeek

Use Cases for BI

Application Use Case
Each application should be evaluated in terms of its primary use case, reporting, analysis or audit. Applications should be restructured to provide optimum performance for its use case.

Best Practice
Data marts should have a relatively short lifespan which correspond primarily to the dynamism of the business environment or the analysis required to achieve a specific set of business objectives. In this way users will get optimum performance.

Multidimensional cubes should not be designed to ‘boil the ocean’ nor only, at the other extreme, to support very specific reports. Instead they should retain enough information so that they can be properly useful with a 1 year lifespan in mind.

Report, Analyze or Audit?

When selecting a reporting tool or optimizing a data mart, I have found that users are looking for three broad categories of output from their queries. There are lots of different ways to break these three categories down, but I think of these as three different types of queries in terms of what the end users actually do.

Report
When I speak of a report, I am thinking that the user wants to see data in a familiar format. Any time a user asks for a particular number in a particular place and he has an idea what that number is going to be, then he is asking for a report. A report in the traditional sense means headers, rows and columns but that's not necessarily the key. A report can almost always be consumed at a glance and key parts of that report are known.

If you sit a piece of paper with numbers on it in front of a manager, it is a standard report. That manager will look at one number and then another and rather quickly come to a decision about whether things are OK or out of shape. This generally happens so fast when a manager is accustomed to a report that we may assume that they are not paying attention. What I have found is that they are actually making several decisions based upon what they see on the page, and in certain cases it may lead them to go to another page. That is almost considered a 'drill down' and it is often mistaken for analysis. But it is not, rather it is a related report or a detailed report. In the end and in either case, the consumer of the report knows where they are going and the answer they are seeking is generally of the sort, Good, Bad or OK.

That is way I consider dashboards to be reports. More specifically, stoplight charts, thermometers, line graphs, trend bars and the like are all reports. They may be interactive, they may be fixed, but in the end what you come up with is a value judgment on a particular part of the business. The difficulty is that it may or may not be a useful part of the business. It might not tell you nearly enough. But you need reports to be dependable, reliable and formally defined. When you get down to fonts and colors and positions on the page, you're definitely in report-land.

Audit
Audits are detailed reports that match up to something. The end result of an audit is increased confidence in the quality of the data. This is something that managers are not generally interested in, but may come to discover. Audits are all about reconciliation.

The difficulty with audits is that you are generally looking into minutia that doesn't drive the business. It's more like counting fish rather than weighing them. Audits are particularly useful in developing an operational understanding of the business. So you will find that DW requirements that allow for adhoc queries into deep detail are generally fresh to the business and/or have a disconnect between representations of the data in more abstracted reports and the people who are close to that part of the business.

You will often find a disconnect on requirements generation over MOLAP and ROLAP or multidimensional and relational technology over the question of audit. Users who are accustomed to looking at simple green bar data and making successive filters on queries and storing those are more thinking in the Audit paradigm. The idea of aggregating 10 million rows upon load to small aggregate dimensions may be alien to them. I've also found a reliable indication of the Audit mindset when I have a requirement to show the metadata of the calculated metrics.

Audit detail is always going to be a requirement but it is one that diminishes over time when the business implications of the operational data is clear and the feedback process is institutionalized. In the meantime, get ready to audit individual employee hours, individual journal entries, individual SKU counts and the like.

Analysis
Analysis differs from reporting and auditing in that you don't know exactly what your analysis is going to yield before you begin it. A multidimensional analysis is configured to draw one through a series of dimensional factors which contribute to the size of a particular metric. That is to say it is a continuous disaggregation process which gives the user an idea of the causal factors. Ideally, analysis is the reverse of planning. But in any case an analysis is not a result but a process.

Most generally analysis leads one to exceptions and causalities. So exception reports and alerts can be said to be analytical but only to the extent that they can be indications of multiple contributors. In any case, an Analysis doesn't need involve the user drilling into deep detail if the implications of those details are known. Metrics can be established that reveal deep exceptions at a summarized level.

Analysis requires a known framework. You have to know generically what kinds of exceptions and correlations you are going to expose through your analytical process or else your results may not be actionable. You can end up navigating through huge heaps of data which give you indications then all you can do is say 'so what?'. That kind of analysis is exploratory but not often a good use of IT resources, unless you are indeed mining data.

Data mining is analysis of data that allows you to make connections. These connections should be incorporated first into your planning process if you intend to do something about them.

 
Batch vs. Interactive

Hyperion has a set of tools which are optimized for producing batch reports. Using Hyperion Production Reports (PR), highly specific and optimized reports from relational and / or multidimensional sources can be published and distributed via email in many forms including PDF. When there are books of reports to be produced on a daily or weekly basis, this can be far more efficient use of compute resources than generating the same reports via custom methods like VBA.

Within the context of migrating to an officially supported version of the Hyperion suite, IT should take advantages of program features rather than custom programming when the Hyperion version has all of the same capabilities.

OBIEE vs Essbase?

For the second time in a couple weeks, I have heard from some fairly reliable guys who are Oracle Partners that the wave of the future is OBIEE, which is the new name for the technology underlying Siebel Analytics. Ouch!

It is indeed possible, depending on the sentiments and skills of the Oracle sales force that this product could kill Essbase. It doesn't take a genius to recognize that if anyone can do it, Oracle engineers can reverse engineer the repositories and multidimensional models behind the Hyperion Applications. Clearly HFM is already there. But why go technically inferior to please sales guys? (That's a rhetorical question. If Oracle kills Essbase, there is NO technical competition.)

So from a technical perspective I find it very difficult to believe that substituting any other product besides Essbase under the covers is going to give any better performance, especially the 64bit stuff. The catch is that the product stream of Hyperion finapps really are not so very demanding in the main, on Essbase. The real strength of Essbase - where it has yet to be beat, is pure datamart applications.

So I could envision Essbase becoming a specialty reporting engine sold through System 9 portals and that OBIEE (whatever exactly it is) becomes the database of record within Oracle's finapps. This would throw the market into another small paroxysm but enable another class of consultants into the fray. The difficulty is that everybody is going to have to ramp up the OBIEE skills - which means another 2 years before successful implementations are going to be easy to be had.

So the really intriguing question now is who is bumping heads with Gersten - that is if Gersten is still the big dog in Oracle's BI engineering group. Might he have sold out? Is ASO functionality at a standstill? I know that there is at least one old product guy from Hyperion who was less than enthusiastic about the direction of the BI n-tier (for n>2) platform. And I know that some infrastructure guys at my old company were just about run ragged just doing the installations and certifications of Planning, HFM et al. So maybe OBIEE adds to the confusion, or maybe it simplifies things by getting Essbase out of the way.

Either way, the future of BI is getting more and more interesting every day. Lord only knows where I'm going to end up in it. On my head if I keep up this speculative provocation. But you know me. I gotta call 'em as I see 'em. Still I'm going to look for more clues in this matter. Hmm. I should re-subscribe to Stephen Swoyer's RSS.

Two Other Reasons

I have thought of two other reasons why Microsoft would be interested in buying Yahoo - things I simply haven't yet heard in the endless speculation.

Server Farms
This is pseudo insider information. When I was up in Redmond last fall, Microsoft was seriously hard up for server farms. Internal projects that had been working for months would be ready to roll out into production and suddenly people would find their projects facing no-go decisions because there are simply not enough servers to go around.

How does a company the size of Microsoft run out of server farms? It's complicated, but the short answer is that Microsoft internal security is extraordinarily demanding on end-users. Remember what Schneier said about benevolent worms? This is essentially what MS does to everyone on their intranet - they use benevolent worms to keep everybody on the inside inoculated and safe from the barbarian hordes who obviously want to hack Microsoft. 

The other insider bit of info I happen to know is that Microsoft has initiated a new capex system to manage their current server farms better, and this plan is due for a major expansion. So if you were an internal client to the service that doles out servers you get a bill from this internal group. That particular internal group is expanding their capability to manage server resources by a factor of about 20.

The addition and annexation of Yahoo server properties and management would be a huge boon to Microsoft which is rather behind the 8 ball internally and acutely aware of a need to expand. There's a new high tech colocation facility in Seattle, I forget the name of it, but people talk about it in hushed tones - and they are booked to capacity out a year. Nobody has the state of the art server farms that Microsoft needs right now, and I'm sure that is a strategic necessity for them.

Hotmail + Yahoo Mail = Giant Captive Audience.
Sometimes it's just that simple. Microsoft wants the eyeballs. Forget search, just think assets and alliances but focus it on the Yahoo customer. I would probably never think of Microsoft if I wanted to get DSL, but I'd think of Yahoo.

Yahoo's customers - the people who use my.yahoo.com and maybe host their homepage at Yahoo, have been around for a long time, and from what I infer from my own high tech crowd, are not the same class of users who would do that MSN butterfly browser thing. In other words, they are a more upscale demographic in terms of their likelihood to purchase additional services, whereas your MSN and AOL guys are just passive eyeballs.

Identity & Deniability

I've been thinking about getting an OpenID. In fact, I've been thinking about getting several. A number of ideas occurred to me as I was gaming online where certain people know me, and know me well as a gamer, but have no idea of who I am in real life.

Positive Doe
The first kind of identity that I considered, I actually have before but I cannot recall the context. Anyway, one of the gamers happened to be Canadian and complained about how much money illegal aliens can get in the form of some social benefits. How do you prove that you are an illegal alien? I mean if I were trying to scam the Canadian health care system as an illegal alien what would they expect of me, to hand over my American ID or my Sudanese ID? Should I simply refuse to speak English and babble in some African-like tribal pseudo-language with a co-conspiratorial 'translator'? It seems to me that there's a big hole in any number of agencies that are specifically funded to keep track of, or ascribe identity, infer identity or otherwise legitimate people who have no papers.

The funny thing is that I can still recall my trip to Axciom in Arkansas many years ago where I was first informed that there are not two but 11 genders in their database. Imputation is an art.

Classes of Identity
The real interesting point about generating multiple IDs is that some should always be, some might be and some should never be attributable to your self. For all intents and purposes, let's call your 'self' as your bodily person where your soul lives - that is the thing that dies when you die. Your identity is a possession. I think, as I said before, that you should have many identities, like keys on a ring, like credit cards in your wallet, like shoes in your closet. The right ID for the right purpose.

If you think of an identity as a suit of clothing, then obviously there are times when you would like to be anonymous. You should be able to wear sunglasses and hats and common clothes so as not to stand out in a crowd. But that's not really a deniable you.

I think of IDs in a few different dimensions of anonymity. Deniability, Traceability, Selfishness. These three make up a rubric into which I would have classes of IDs according to their purpose.

Deniability is the extent to which you can claim that an ID is not you. For example, I can publish a comment at somebody's blog and leave give a false name, forget about the comment and two years later have someone accuse me of authorship. When the name is 'anonymous' I could clearly and credibly say, that could be anyone. This would be less credible if I had published the comment as Joe Cubegeek.

Traceability is the extent to which someone could prove that I had access to the incident in question. I may not even know what my IP address is or the route through which my message was posted. However forensically, someone might be able to trace the message to my house, or someplace it could be reasonably proven that I would be.

Selfishness is the extent to which the content of the comment is similar to my style or attributable to my interests. I might use a completely deniable name, and access a site through an untraceable method, but if the comment is about the nature of my wife's cooking, it stands to reason that I might have something to do with it.

In setting up legends / avatars / identities, these are the kinds of issues I would consider for defending.

For example. In my online gaming, I am known as 'sixoseven'. This is not a primary ID but a somewhat frivolous one. I wouldn't mind if other people knew that sixoseven ties to my self. But in the context of my gaming actions, I think it would have a fairly low traceability to the other gamers. It has high selfishness because I only use it for the games I play, and I use my voice in many of those games, and people could recognize my voice. It has a fair amount of deniability because there's no reason to think that 'sixoseven' is me.

So I might set up a number of IDs, the first would be Class A, which I always want attributable to me. Therefore I might not care about traceability or selfishness. I want people to know it is me and nobody else. This might be something like my national ID, my passport, my driver's license.

Another might be Class X, under which I might acquire contraband under false pretenses and would want perfect deniability.

Another might be Class E, under which I receive privileged information in which deniability is desired bu not necessary, although traceability should be very low.

These are all ways I would think about classifying IDs in the context of anonymity given the opportunity to assume many identities for different purposes.

Bloomberg & Government Transparency

NYC Mayor Michael Bloomberg is a pioneer who is at the very beginning of a revolution started simultaneously with Walter Wriston. The second fruits of his pioneering spirit are beginning to bloom. The first fruits were, of course, his fortune in the securities information industry created by the machine and company that bear his name. The second fruits can be seen here in Citywide Performance Management.

Reversal of Fortune

It's official. I'm moving back to the Essbase world.

It has been about six weeks now since I have moved out of my old digs where I was VP of Services for a Microsoft partner. It was an invigorating experience, rather like a shower that ran uncontrollably hot and cold, alternately luxurious and chilling. I'm sure that I learned a few things, but I haven't bother much to decontexualize them. As folks have been speaking to me recently about my change, it has been difficult to express what has been good or bad without speaking out of school. Nevertheless I did have a good analogy that I used a bit while I was employed there regarding my disposition.

Half of my job was gardener, the other half of my job was catcher. So I wore two gloves, one for tending to flowers and fertilizing, the other for catching fastballs, curveballs and the occasional spitball. Occasionally I would use the wrong glove for the wrong task and a flower would get clobbered. More often it was the other way around and I would get seriously burned by the pepper. Builds character. Of course I wish all of my former employees and colleagues all the best, except for one, and that's all there is to say about that.

Today I'm completely decompressed and back to running on slightly modified versions of my old anti-Microsoft bias and gripes. There's a part of me that cannot wait to get back into Unix, and especially on the Macintosh platform. There's another part of me that cannot wait to get back to big hunkin' enterprise customers and their money and sophistication, and there's another part of me that is almost ready to get back to heads down, hands on work, but not really.

As it turns out I'll be doing that hands on in the short term and I'll be working on front-ends for a change. The thing that has really got my attention is, as I've said, multitouch and new visual paradigms for business intelligence. I've decided that 'glasstop' I will now call 'tabletop' computing and that's what I'll be working on as I come up with some theories and product ideas. They will grow to technology and technique ideas as I get my hands on a Mac later this year and start learning UI programming. This very simple application by CNN of Perceptive Pixel gives you an idea of what I think will be irresistible to BI customers in the next generation.

Immediately, however, I'm going to be working with Tim Tow of Applied OLAP on his new Dodeca product, which is a soup to nuts .NET smart client whose features I helped brainstorm a couple years ago.   It's a comprehensive development environment and the most sophisticated thing you can put on top of Essbase. Tim and his team are world class and I'm really proud to be working with him. In fact, I just bought 3/4 Terabyte drive just to celebrate the data mongering possibilities I will start to exploit as I get back into serious Essbase.

I have to say that in my career so far, the project that helped generate some of the Dodeca features was my most extreme and extraordinary. It was conceived to replace the cost estimating system for a global 100 company that allowed 17,000 Essbase models to be living all at once. That's was not centrally maintained and driven by a DW, that was with 5,000 cube modelers, with dimension building access who were generating their own data. It's the one that got me up with the HP folks in Cupertino running Essbase on the big 64bit Superdome where I did some screaming performance stuff with parallel java servers and the whole nine yards. When I tell you this front end is the greatest harness ever for Essbase, I kid you not. From what I've seen and know about its capabilities, it's everything that Alphablox and Arcplan promised to be, but better. Most importantly, It allows you to build custom workflows, custom forms and custom distributions of all that with more security granularity than any other front-end to Essbase. Again, all in a secure .NET framework. The guys at Applied OLAP have rethunk client distribution for the BI world, and everybody who is doing anything at all in Excel and VB really needs to check it out.

Only Dodeca enables this kind of Essbase farming. There's nothing else anywhere that scales as an application framework for Essbase. Did I say that it does distributed master data management too? Yeah. That's part of it. The really fabulous idea that makes this even more interesting is something of a future, but you know me, that's where my head goes. The fabulous idea is that you can begin to think about the move from unstructured data to multidimensional structured data because Dodeca is going to make it easier to do so on a different scale than people are accustomed to. But that's all I'm going to say for the moment.

It's wonderful to be as old as I am and still get excited about stuff.

The Last ID

Last November, I chanced upon a website that advocated a new kind of contract between persons. They called it a limited liability persona. It was a very cool thing and I think it's the future of identity. At least I hope so.

Part of the problem with identity theft these days is that ultimately it all boils down to a few small keys and then you're busted. Most of your most significant IDs are keyed by your mother's maiden name and your social security number. Most people should know by now that there are duplicates. It also doesn't take much to forge a number. The idea that's cool about the LLP is that you can establish your own authentication independently of every other kind of contract you have.

When it comes to a national ID, there's a certain amount of paranoia and ignorance. It's probably something I should get involved in a bit more in. I checked out a video not long ago - the guy from Sxip talked about identity 2.0, but that whole conversation seems to have disappeared, or at least some of the experts have gone to secret projects. It rather reminds me of the time several years ago when we were talking about micropayments and the 'impossibility' of setting up that thing which has become PayPal. There's a certain inevitability to it, you just don't know where it's going to come from.

I had a conversation with myself the other day merging the four factors, GWOT, LLP, Immigration and Identity Theft all into one.

One of the interesting things I discovered is that one of the reasons we know there are about 12 million illegal immigrants living in the US is because the credit bureaus know. They keep databases and they share databases. Let's think about the capability right now, and let's do it in terms of voter fraud. In approximately 30 seconds, from a million locations, I can create a connection that talks to some number of computers and gets 40 bucks out of my bank account. I'm not talking about ATMs, I'm talking about cash registers. I have a PIN, I have a 16 digit ID number, I have an expiration date and a three digit code on the back of the card for additional validation. Your kids and my kids are used to getting gift cards that just sit in large displays at the local supermarket ready to be activated for 25 bucks worth of iTunes downloads. That's kind of like the LLP, it is a persona you create just for the purposes of music downloads, and the security system allows you to activate it for a preset maximum. There are never going to be concerns about duplicate numbers on gift cards. You just use big numbers.

If I were president, the first thing I would do is to initiate a worldwide census. I would direct the State Department to publish all of the kinds of persons we recognize and start keeping tabs. One of the great difficulties we have in dealing with the War on Terror is that we haven't dealt with a couple realities. Consider them:

1. Detention is the Ultimate Solution

As Edward Luttwak points out, all guerrilla wars, urban or rural (urban guerrilla is a common euphemism, meaning terrorist) can be won by detaining every human being who might possibly be an enemy, holding them securely until the war is over and the winner is clear, and then releasing them without punishment. Like, duh, man. Which has more negative impact on innocent civilians: internment in a civilized detention center, or involvement in a civil war?

2. The Population Is Fixed

...The insurgents require the population to act in a certain way -- support, sympathy, intimidation, sometimes just reaction to provocation, you know? And if you can take that reaction of the population away from them, it's extremely difficult for them to achieve anything.

That's why the surge is not only a matter of putting extra troops into the country, it's what they do when they get there. And what they're doing is going into areas and not leaving. And they sit with the population, partner with them, help them defend themselves. Keep the enemy away. Prevent them from coming back. And if you like, restructure the environment to hard-wire the insurgent out of it.

Three movies come to mind when I think about how difficult it is to find ones family in war, Blood Diamond, Schindler's List and Hotel Rwanda. Forget about American civil liberties for a moment and think about what an enormous service to the world it would be if we went made a huge locater database for every human on the planet. I'm willing to suggest that a proper system wold be of tremendous benefit to humanity if it were done with the LLP concept in mind. That is to say you could absolutely and positively identify people but that through LLP, the people themselves would be in control of the associations known to the system of authentication.

Now for the purposes of GWOT and Immigration, I imagine that there would be certain absolutes built into the system - which represents the capacity to replicate authentication that we have today. The best ID anyone has, or any garden variety civilian has, is a national passport. That's the thing that is recognized worldwide. So perhaps as a minimum, in the Last ID, your nation of citizenship and all things that State Departments attach to that, would be retained. IE I would have no control, when identifying myself, that I would reveal my nationality and consequently my status with my home country - all the stuff that an immigration control agent would have at customs. But whether or not I wish to give permission to associate my credit report or medical records would be entirely up to me.

Ultimately, the cost of not doing this is what we are up against. Americans will have a very low tolerance, given the openness of our society, to domestic terrorism should it rise above a certain point. I have every expectation that we will, for the sake of security, submit to a regime of national identification which supercedes the present system, warts and all.

Glasstop Computing: Intro & Playing Card Metaphor

I've been doing some thinking about multitouch very seriously since Christmas. (I got an iPod Touch) and I'm starting to crank out some ideas about a new visualization paradigm. It starts with that very premise - that different kinds of data need different kinds of visual paradigms.

Thinking about this in a 'surface computing' UI has unlocked a large number of possibilities and I am developing those ideas now. I will be sharing them from time to time here. It's a bit frightening to do so because I'm not sure how I can keep the ideas to myself. At the same time, I really can't see any other way to develop them strictly in a vacuum, and I think this blog is going to be the best way for me to get in touch with people who can and will help.

At any rate, the basic idea is to make a real desktop - a large translucent horizontal surface for organizing, categorizing and analyzing large amounts of data. Glasstop or tabletop computing is what I'm thinking about, and specifically I am thinking about objects from BI. Not just the presentation objects for end-users but a comprehensive UI which exposes all of the tools that BI architects would use to prepare the data for presentation and that prosumers would use to organize their own or public data sets.

The first rule is that objects will be presented in such a way as to give visual clues to their scale, and the objects may change shape or navigation technique depending on its cardinality.

For example, if I were a marketing manager and I wanted to deal with 1 million customers, and I have and entire six foot by four foot table on which to do so, I would like a tool that allows me to segment them by all of the attributes I have for them. But how do I represent 1 million customers before I perform that kind of segmentation? How might I apply filters and enumerate distinct attributes? These are the kinds of problems I would like my system to solve very easily through multitouch manipulation. 

There are a number of interesting analogs that come immediately to mind. Working with a deck of cards or stacks of paper are manual types of sorting paradigms which are familiar. So let's envision one small aspect of dealing with this problem - attribute enumeration. With our 1 million records, we can represent them as a deck of cards. But imagine that there were 100 cards in the deck instead of 52. If we dealt out the deck, each card would represent 10,000 customers. The thing to keep in mind is that we can zoom in to any card and deal with 10,000 or we could zoom back out to the entire deck. We can also change the size of the deck arbitrarily, but for now let's leave it at 100.

Now we know a deck of cards has suits, and each card has a face value. But if we use our imagination we can think of each of these as dimensions or attributes. For the time being let's leave it at the attribute level. Part of the system that is going to help us out tremendously is that it will have visual paradigms for different scalar systems. When I think multitouch in my iPod, it's easy to do flick gestures through say 100 or 200 albums knowing that each album may have 1 to 20 songs each. But I also have contacts in my iPod, for me that's about 7000. It is very inefficient for me to navigate to the Rs, for example, knowing I have 500 people to flick through before I get to Joan Ruzart. I'd probably have to think, go to the Ss and flick backwards. So the cardinality of a flick list should probably not exceed 200. Accelerating flick speed would not be as user friendly as simply adding a shortcut. Therefore I would want a library of shortcuts which are applicable to all of my data sets for quick navigation. And that's where my attribute enumeration comes in.

Obviously if I've already designed the records for my 1 million customers, I know what all the attributes are and what all the fields are. Let's start with something easy. Last name, as I implied above. Instantly and obviously I can split the deck into 26. So let's say I take my pile of 100 x 10k and drag it through a hoop (rather like a lion tamer) with 'ABC' on top. It comes out into 26 stacks. Or alternately it marks those same cards with 26 different letters. For the sake of our example lets say that our hoop only marks the deck.

Only because I used to work at Xerox am I familiar with the terms 'bursting' and 'decollating'. I'll just use the first term. Bursting means taking carbon copies apart. But in our deck example each card represents 10,000 customers. So to burst a card gives us 10,000. We are not bursting our cards as they go through the ABC hoop, we are marking them. So we make 1 million marks in an attribute with a cardinality of 26.

But we could do the same operation with a burst and resize. A burst and resize would take the 100 cards with the random distribution of 10,000 customers each and then sort them into 26 cards with a variable number of customers each. That sounds like a fairly expensive operation. So let's not do it yet.

Instead we will take another attribute and apply it visually to the deck. Let's apply a net revenue per customer metric to the deck. We calculate net revenue per customer and then take a spread of those values. That spread gives us a histogram. Think of it on the x axis from -$500 to $7500. Now I can stack my cards visually along the x axis and instantly eyeball the median. I can squeeze the scale so that I get only three buckets or I can stretch it wider so that I get more buckets. I can grab any stack in any bucket and then expand that out as well. With any stack I can mark or burst and resize the subset and work with another attribute.

Cool huh?

Microsoft + Yahoo = Monster Zero

What I learned about Microsoft is that it already has three and a half heads. It's now poised to add a fourth. Since I've been purging myself of MS over the past several weeks I don't have any idea where I put my notes that told me exactly what the acronyms are for the three of Microsoft's 10 billion dollar businesses. But the biggest point, which led to my leaving the Kiretsu was the realization that Microsoft is a technology galaxy of loosely federated product groups. This has specific advantages when it comes to providing a platform, but specific disadvantages when it comes to building product stacks, and product stacks are what sell in the enterprise software space. For my tastes, and for the tastes of every SAP, Oracle and IBM customer under the globe, Microsoft's product strategy doesn't have the requisite focus. I won't belabor the point.

What's shocking about this particular idea is that it is a startling admission that MSN is a failure. I can't imagine anything more or less than a huge, huge culture shock.

What's obvious is that this is an attempt to kill Google, or replicate parts of its business model. But Google's business model is actually besides the point. Advertising revenues through online are really means to a Google end which is to build a single computer for the world and world scalable applications for that computer. Microsoft is just adding functionality to a massive installed base that has never and will never talk to itself the way Google customers talk to themselves. Which is to say Microsoft enables software gadgetry with no overarching design. Adding Yahoo to the gumbo.. well it only makes discrete sense not holistic sense.

So Microsoft is going to grow another head. I think it makes that which is ugly even uglier, and that which is dangerous more dangerous still. It will show Microsoft's ability to fight in yet another domain and it will put another lovely bushel of mulch into the compost, but don't expect all the fermentation to yield anything smooth for another four years.

In Search of Multitouch

Now that I have my iPod Touch, I am continuing my search to find what it may take to get into the new killer app (in my view) for BI, which is multi-touch. I've said that BI UI is boring, but what would make it exciting aside from multitouch? Well, that's what I'm thinking about, now that I have a little spare time between gigs.

Today I dropped by a custom audio designer joint here in the South Bay and listened to music for the first time in quite a while. The guy there had a sweet setup with a couple monoblock McIntosh amps hooked up to some B&W speakers. He played a digital source which was absolutely divine. I told him of my ideas and he pointed me towards a couple vendors.

Kaleidescape
They had a sweet media server there pumping some pixels through the Fujitsu Plasma. It was actually a fairly large and impressive library, which suggests to me that theres a serious media server behind it. The young guy swears that Kaleidescape is the number one server of its type, and it does look rather impressive. Of course I was immediatel critical of the UI which in this case was sorta iTunes movie-like but controlled by a hand-held remote that looks like a very large cell phone.

AMX
AMX is one of those companies that builds all the kinds of doodads and controllers and servos that you would expect to find in Auric Goldfinger's HQ. What button do you push to flip the pool table? AMX probably makes it.