> Then they bought Skype and it went downhill from there. I don't know what happened, but they took a lot of time integrating technologies from Skype (peer-to-peer) with their own tech (which was more telecom/server based)
I was at Skype at the time, working on this very migration, so maybe I can shed some light on what happened.
There were a number of failures (on Microsoft part mainly) that made the whole thing collapse:
- The original Skype developers were used to program in C++ with on-premise Linux servers. When the migration plan came in, the tech stack switched to C# on Azure. Most developers were not really keen or experienced working with C# and Windows. Azure was very much beta at the time, I remember the philosophy was "eat your own dog food" which created an insane amount of frustration working with sub-par tooling, little to no documentation, plenty of bugs on Azure. There is also a mentality/philosophy shift of working with C#/Windows coming from C++/Linux which was not great for team morale.
- Support for group chat was paramount, which meant there was a need of transition from peer to peer to client server.
- The original Lync team was very much annoyed with the rebranding and migration plan and was not really keen on helping Skype developers. I remember us receiving a "dump" of the Lync VCS as tech reference, in the form of a big fat multi GB archive. There were tons of binaries, even videos in there... Lot of fun...
- MS/Lync used a _huge_ amount of proprietary SIP extensions to have Lync work.
- The migration to the cloud and client-server model was accompanied by a huge migration to a brand new microservices in C# project. It meant recoding everything in a new language, a new platform, a new protocol, dealing with hundreds of half baked microservices in a half baked cloud platform. It was slow, buggy, and an architect astronaut fest.
- Overall 99% of the original Skype calling team resigned and joined Twilio.
- Pretty much every manager from MS that I have seen at Skype stayed no more than a year then rotated to a new role, at one point we had 4 changes of management in 2 years.
- Every single interaction with Microsoft management was surreal. Pardon my language, but they were not technical at all, unable to understand the most basic architecture proposals, bullshitting tech talk 9 words out of 10,and overall they were living in an other planet in terms of market share and branding. You could definitely tell that they were talking "product" and "objectives" for Skype, but never used it, understand it's competition, challenges, etc. Pure bozos.
In the end I think Microsoft still extracted some value from Skype, mainly:
- Skype had good PSTN contracts worth a lot of value. The matching engine to select the cheaper one dynamically had value too, and historical data.
MS engineer who worked on Lync. I feel like the Skype acquisition was poorly handed too, but interesting how the perceptions differ.
We were folded into the Skype org, with Tony Bates -who I maintain had no idea what he was doing- reporting to Ballmer. Fine. But he and his henchman (Mark something?) embarked on a quest to make everyone do by-the-book Scrum. That, screwing around with corelib, and the ill-omened liveid migration process meant that Skype didn't improve in any way users cared about for several years. Meanwhile, the competitive landscape changed and the only Skype feature anyone ever used - being able to call your grandma in another country for cheap or free - lost its differentiation. It was clearly over by the time my mom told me she had whatsapp.
Nobody I knew had any animus towards Skype engineers and we were entirely willing to help... but what anyone actually wanted was kind of unclear.
Someone I know who worked there said that after the sale to Ms they had 4 levels of managers between a developer and Ballmer. A year later it was 8 levels. Maybe he was exxagerating, but that was also adding to the problems
Depth of org in a company is logN(number of employees) where N is the number of reports a manager has. At Microsoft there's order of 100K full time employees and N is about 10, so depth is about 6. This seems fairly unremarkable to me, though obviously there are other ways to do it.
I love that alignment description, that reminds me of what it was like when I was at Microsoft. Fancy VPs always did unclear things. Above them in my chain was some kind of president who had no clear role either.
I was at Skype at the time, working on this very migration, so maybe I can shed some light on what happened.
There were a number of failures (on Microsoft part mainly) that made the whole thing collapse:
- The original Skype developers were used to program in C++ with on-premise Linux servers. When the migration plan came in, the tech stack switched to C# on Azure. Most developers were not really keen or experienced working with C# and Windows. Azure was very much beta at the time, I remember the philosophy was "eat your own dog food" which created an insane amount of frustration working with sub-par tooling, little to no documentation, plenty of bugs on Azure. There is also a mentality/philosophy shift of working with C#/Windows coming from C++/Linux which was not great for team morale.
- Support for group chat was paramount, which meant there was a need of transition from peer to peer to client server.
- The original Lync team was very much annoyed with the rebranding and migration plan and was not really keen on helping Skype developers. I remember us receiving a "dump" of the Lync VCS as tech reference, in the form of a big fat multi GB archive. There were tons of binaries, even videos in there... Lot of fun...
- MS/Lync used a _huge_ amount of proprietary SIP extensions to have Lync work.
- The migration to the cloud and client-server model was accompanied by a huge migration to a brand new microservices in C# project. It meant recoding everything in a new language, a new platform, a new protocol, dealing with hundreds of half baked microservices in a half baked cloud platform. It was slow, buggy, and an architect astronaut fest.
- Overall 99% of the original Skype calling team resigned and joined Twilio.
- Pretty much every manager from MS that I have seen at Skype stayed no more than a year then rotated to a new role, at one point we had 4 changes of management in 2 years.
- Every single interaction with Microsoft management was surreal. Pardon my language, but they were not technical at all, unable to understand the most basic architecture proposals, bullshitting tech talk 9 words out of 10,and overall they were living in an other planet in terms of market share and branding. You could definitely tell that they were talking "product" and "objectives" for Skype, but never used it, understand it's competition, challenges, etc. Pure bozos.
In the end I think Microsoft still extracted some value from Skype, mainly:
- Skype had good PSTN contracts worth a lot of value. The matching engine to select the cheaper one dynamically had value too, and historical data.
- The brand itself had value for customers.