Google is in the mix now with their own IM client -- including voice traffic. Why? Why not? Its easy now. All you need is plenty of bandwidth and some big servers. Pretty much all the connection protocols, encoders (CODECS), and so on are available now as standards. There are even good commercial and open source stacks which implement most of them. I know, I'm using several in the current project I'm working on it. Making an IM community client would be trivial with the stuff available now.

Tell me then, how IBM or Microsoft wish to make money with their products. SIP servers and CODECS are pretty much a commodity item now, so where is the value add? Is it in screen sharing, group calendaring, and online meeting processes? Aren't these too headed for comoditzation? Isn't their very commodity nature the thing which makes them most powerful -- that ability to join people in these meetings who aren't necessarily part of the same infrastructure?

I'm sure there are some customers who require special items like secure logging and so on -- but it seems to me that is something best done at the network layer anyway. Just scoop the data up as it goes by the routers (true, switches make that a little more challenging, but not hugely so).

Instant message itself is not the real battle here. For both IBM and Microsoft both, a solid communications infrastructure provided by Sametime, Live Communication Server, or Workplace Collaboration Services is going to be critical -- but not as an instant messaging platform. Its critical because it will provide the session control and data pipeline for traffic between developed applications (in Workplace Designer or Visual Studio respectively) and a back end secure storage mechanism (DB2 & Domino, SQL Server & Infopath).

From here, the key development paths a few years out look like this (not in order of preference, necessarily) -- I'm sure I'll get comments and updates, many quite insightful and accurate. I'll add this in another font to indicate that they've been added, so as not to spoil the context of the original.

Andrew's Prediction* of Likely Development Platforms - 2007 - 2008 Timeframe * of course things change, particularly product names -- but largely this is the way I see it looking
	Front End Development	Session & Data Pipeline	Directory Services	Back End Logic	Unstructured Data	Structured Data
Open Source	Java Eclipse	Asterisk	LDAP New Standard TBD	J2EE Perl	JSR170 Repositories	MYSQL, Other RDB's
IBM / Lotus	Lotus Domino Designer IBM Workplace Designer	Sametime IBM Workplace Collaboration Services	Domino Directory LDAP	(J2EE) Websphere, Workplace & Portal Lotus Domino	Lotus Domino Future DB2 Both also as JSR170 repositories	DB2 Cloudscape
Microsoft	Visual Studio	Live Communications Server	Active Directory	IIS	InfoPath	SQL Server

Conclusions

From where I stand, Open Source has catching up to do in many cases, but good solid entries in all categories.

At the front end, IBM and Microsoft have very different platforms, both are quite good but focused very differently. I use and like both. Open Source is weakest at the front end presently.

For the pipeline, I think Asterisk is outstanding (you didn't really think it was just about VOIP did you?) and may be the most compelling -- particularly since I see this as a commodity play, with little benefit to gain from proprietary servers at great expense.

I don't see a good directory solution really -- I think a new standard will emerge from the need to support ever larger SIP implementations. Clearly IBM knows how to do this well with LDAP and Lotus Domino, while Microsoft continues to have a decent enough strategy if you're willing to be locked into a Win32 dominated infrastructure.

At the back end, nothing beats Lotus Domino for cost/benefit, particularly since it incorporates full support for Java and thus encompasses the open source strategy as well. IIS is a neat platform, but has a long way to go to convince me it is robust enough to be a real player without major rewrites.

Domino owns the unstructured data world right now. JSR170 is critical to the future, and I fully expect to see Domino and DB2 as some of the first implementations. I know nothing at all about Infopath so I won't compare it.

Structured data is boring and mature. Nothing to add there.