I agree. Fragmentation is a major problem. But this stems from Smalltalk’s flexibility and adaptability to various problem domains and special situations. It’s rather like the Linux community. There are a gazillion Linux distros for all occasions, but only a few rise to the top (eg, Mint, Debian, Ubuntu, openSUSE).
Similarly, there are many Smalltalk dialects, but only a few are widely used: Pharo, Squeak, Cincom, and perhaps Dolphin.
I promote Pharo because it has the best likelihood of becoming the de facto standard. If we can gather around Pharo, we can build a healthy ecosystem of libraries and tools that hopefully will find their way into other dialects.