,
16 tweets,
3 min read
Read on Twitter
@jeffborek and @gossmanster regularly argue that Cloud Changes Everything in the open source world. This response needs its own thread. Let's first separate the 'market' from the software. 1/x
In the past 40 years since copyright landed on computer software, we've seen enormous changes in computer hardware technology. PCs, the Internet, cheap servers, the Web, then Mobile, now Cloud, and IoT devices. 2/x
In each case, these computer changes have brought about enormous business and market changes. Each of these computing hardware platforms has created new markets for software (to run the platform and run on the platform) and new challenges for delivery channels. 3/x
Let's consider the software challenges separately: First, the 'to run the platform and run on the platform' problem. This is a very Clayton Christensen economics sort of problem. Value moves around the product 'network' over time. 4/x
Each new HW platform has created enormous markets in the application spaces (run on the platform). But things change. You now pay a few dollars for your phone app or it's not free with subscriptions/ads behind it, compared to the cost of PC/business apps in the '90s. 5/x
Delivering infrastructure software at scale is still lucrative. Databases, operations, and operating systems (e.g. IBM buying RH for $34B) are all lucrative markets requiring Big Software Engineering to scale. 6/x
But if you're delivering the HW platform itself, you can't be tied to 3rd party fee-based software over the long term. Which brings us to the 2nd problem of changing channels. As hardware platforms have evolved, application models have evolved. 7/x
In the '80s, a customer bought an application or tool (e.g. database), and installed and ran it on the computer 'in front of them.' The channel had COGS. The pricing was relatively straight forward. 8/x
Now, a customer reaches through an interface (possibly mobile) to take an action (possibly stateless) that effects data (possibly reached through an endpoint) and the entire software dynamic can be distributed across different computer platforms in different locations. 9/x
A company delivering that solution to a customer needs to figure out how to deploy (channel) and charge for the solution to the customer's problem. Does COGS now include CPU/Storage/Network charges from the Cloud Provider? Is the fee per transaction? Time-based? 10/x
If I deploy my app in a container on a cloud, and it's elastic wrt its compute needs, how do I think about charging my customers for it? If the container is actually a mix of containers and endpoints, and maybe other 3rd party software, what does my pricing look like? 11/x
We would not have the Web (or Cloud) we have today without LAMP. No developer pays for languages & tools anymore. Engineers collaborating in well-run communities using OSD-based licenses on their projects have enabled software to evolve to run these new hardware platforms. 12/x
The same holds true in this latest new new computer world of Public Cloud. But the other thing that is equally true is that the COGS, pricing, channel challenges each new computer hardware platform wreaks upon the applications industry of its day will continue to happen. 13/x
This will cause grief in companies misunderstanding the engineering economics of open source software project use/creation and community engagement. Good companies have failed for less in each computer generation. 14/x
Public Clouds will differentiate. Some will be app model based, some may be datacenter or partner-centric, some may work to be the cheapest delivery, and who knows how Alibaba, Baidu, Tencent, Huawei work to differentiate. 15/x
This means 'Cloud isn't eating Open Source.' Software companies still need to be companies first. Many will fail. Engineers collaborating on software as they have back to 1950 will continue to rule the day. The OSD describes the broadest surface area for collaboration. 16/16
