MIT: Platform Failures: Lessons for Strategic Management
https://dspace.mit.edu/bitstream/handle/1721.1/99112/921180851-MIT.pdf?sequence=1&isAllowed=y
12 replies (most recent on top)
Platform Organization did not follow a Product Process:
- Strategy
- Planning
- Execution
https://www.slideshare.net/venturestudio/product-roadmap-w-vinod-muralidhar
Mostly it was political haphazard opaque dynamics based on hitler like whims of the vice president. No wonder Platform organization did not survive – a very good decision by the new CTO.
PaaS platforms are dead, thanks to IaaS providers
https://www.infoworld.com/article/3277637/paas-platforms-are-dead-thanks-to-iaas-providers.html
Macy's "Enterprise Platform" group tried to provide PaaS – failed miserably. Then the branding was "Orchestration" (whatever that meant). Anyway million of dollars down the drain with expensive consultants. With GCP migration, some of these mistakes are being corrected by new CTO.
These days, much of what PaaS provides, including quick and easy development tools and quick ops deployment, have been replaced by IaaS providers. Public IaaS clouds, such as Amazon Web Services, now offer features such as container-based development, serverless computing, and machine learning and analytics that have made a feature-rich IaaS platform the best place to build and deploy cloud-based applications.
What also interesting to note is that most major public IaaS cloud providers also provide a PaaS as well.
What’s happed is combination of choice and momentum. Developers, who are these days mostly charged with the migration of application workloads to the public IaaS clouds, have to avoid PaaS. This is because PaaS clouds typically require adherence to specific programming models, languages, databases, and platforms. Thus, while PaaS is good for new cloud-based applications, you can't easily fit some traditional LAMP-based applications into a PaaS platform. To do so means major rewriting, cost, and risk. So, bye-bye, PaaS.
Please read this article:
Go Vertical or Go Home.
https://www.linkedin.com/pulse/go-vertical-home-ardaman-singh/
Since the "supposed"/"ostensible"/"contrived" Enterprise Platform for Macy's Tech could not go effectively Vertical ... it was sent Home (what a righteous beautiful outcome). Quotes from the article:
"In certain rare cases, you can have best of both worlds - build a platform side by side with a friendly army of developers to help you guide the evolution of the platform. But even then, the window of opportunity is small - very quickly that pot of gold (a.k.a. fat wallet) begins to run dry. And when that happens, there is only one savior to feed your army - your check-writing customer, who demands solutions that directly add value to their business. And you’d much prefer to have all of the check come to you, rather than most of it taken away by the developers in the middle."
A Platform team/org should always be challenged with open source. Please read this article (Netflix discontinued use of their own Platform frameworks in favor of Spring Boot 2/Cloud):
https://medium.com/netflix-techblog/netflix-oss-and-spring-boot-coming-full-circle-4855947713a0
Without proper strategic Technical Product Managers, any Platform team/org will meet its demise within 5 years. Often the VP have no clue how to manage Technical Product Managers since their whole career has been on execution and politics (no strategic training). Architects are (or should be) Technical Product Managers by another name ...
The Platform mgr drove the ML platform product mgr out of Macy's and made the engineers execute his own architecture. He had his own vision of the product and didn't listen to anyone else, including the engineers when they often had criticism of the direction. He also wanted the engineers to build in-house many of the tools that were already open sourced and available to use. If he wanted to build in-house, at least be original and look at our actual requirements, not just copy everything that those other tools have... It would take months to get even a working version, with features that our users didn't even want, while our users started exploring open source versions of the tools and didn't even want to touch our in-house tool. Then came even more politics as he tried to convince them to use our tools. Totally inefficient strategy and waste of time/resources.
What is the role of Team Manager in Agile? If Team Manager is supposed to be a Product Person as well then contributing to backlog and strategy management makes sense – otherwise you are stuck with blank stare managers (since technical product management function is absent/non-managed).
That explains a lot. I heard that the SF Platform Team manager rarely participated in sprint planning. It's nice that he gave his team complete autonomy, but that also means that he was clueless about the strategy or who was doing what. When it came time to explain SF team goals, my guess is that he just stared blankly into the middle distance.
There were meetings with the CTO and another team in JC that had been doing similar ML platform work. My understanding of those meetings was to have each team present their work and for the CTO to decide how to move forward. I guess he saw duplicate effort (as there always is between SF and JC) and decided to cut the SF team. Just my observations based on emails and calendar events... someone can confirm or deny.
Platform Managers and Directors were not strategic:
https://martinfowler.com/articles/talk-about-platforms.html
It is better to dissolve a non-strategic organization than to keep a while bloated elephant alive.
For Platform Organization to survive, it MUST have a effective executable strategy – otherwise Platform team becomes an "effective dumping ground" for deflections/excuses of non-actions (regular flow of paychecks). This also means there needs to be deep executive commitment towards Platform funding – after it has been determined what the Platform Organization should do. For a concrete example please consider the strategy for "Experimentation Platform" – what is the go forward pathway?? You may look at Twitter Experimentation Platform strategy as a compare-contrast ...
Three Elements of a Successful Platform Strategy
https://hbr.org/2013/01/three-elements-of-a-successful-platform
- com Enterprise Platform execution is the biggest scam/sham I have personally experienced in my long career in tech.
The Platform Team manager at Macy's Tech was laid off in July. Several of his reports too. For the 'survivors," one moved to a different team and a couple are handling bug fixes, but the team is mostly gone since the focus is on the cloud. (I heard that one of the ones laid off got another job immediately with much better pay and benefits, so there can be a sweet life after M).