Misunderstanding open source #5: Analyst Challenges
Analysts of the software industry have some challenges when it comes to talking about open source. I know several analysts from a few different analyst companies, and they are good people with good intentions. This post is not a criticism of them – just my view of some challenges their industry faces.
Lack of Experience
Many analysts base their work on the experiences they had prior to becoming analysts. A lot of them worked for decades in large and small software companies. This experience has served them well to date. Unfortunately they now have to talk about open source companies – and none of them have ever worked in one. They now have to deal with new models, new terminology, and new concepts – all without getting direct experience.
This lack of experience leads them to make mistakes and say some odd things. Here is a recent examples (not naming names – you know who you are). This is from an analyst at a very prominent analyst firm, talking about open source companies using the open core model:
Even the very definition of “community” is being adapted to suit the open core narrative. What has largely interested the corporate IT world is the concept of a community as a collection of code contributors working outside a normal project/company structure. But now open core providers are extending the term community to include users and even resellers. That, of course, is what we’ve all been calling a software ecosystem for the last twenty years. Same old, same old – just co-opted terminology used to describe it.
What this analyst is saying is that open core companies have modified the term ‘community’ to suit their needs, by including the user community and not just the developer community. Anyone familiar with open source sees how ridiculous this statement is. Within the free software and open source worlds the community includes all constituents, not just developers – wikipedia definition.
Secondly they state that an open source community is the same as the ecosystem of a proprietary vendor. This is not the case in my experience – the participants and the kinds of interactions are very different. This analyst, due to inexperience, ends up making inaccurate statements that reduce their credibility.
To help counter this issue analyst firms are adding open source specialists. This is a decent initial step but it doesn’t really solve the problem. When it comes to our domain (Business Intelligence) most analyst firms have one analyst that understands open source, and a different analyst that understands the BI market. Unfortunately, neither of them is able to understand how open source will impact the BI market.
The first thing analysts need to do is to review the questions they ask. In many cases the questions asked are about the amount of money spent on software licenses. This is fine as long as 100k spent on package X is roughly equivalent in value to 100 copies of package Y which costs 1k. In the past this has been roughly true. But now open source companies are providing equivalent products for a fraction of the price charged by the old vendors. Additionally, there are companies that sell subscriptions, not licenses.
- Stop asking about licenses
- Stop asking about money
- Start asking about CPU/cores
- Start asking about usage
This is an issue that has to be tackled domain by domain. The analysts in the operating system and database domains get this (took them years). Analysts in the BI space don’t get it yet.
One of the metrics that some analysts use when assessing the impact of a vendor is ‘how often do people call me asking about the vendor’. The reason people call asking this question is that they are not sure if the software will suit their purpose or solve their problem. They are calling the analyst because they have no other way to get an unbiased opinion. When consumers have the ability to download the software and prove to their own satisfaction it does what they need, they don’t need to call an analyst.
An analyst saying ‘no-one is using open source software X’ is like a travel agent saying that no-one goes to Disney anymore. They do, in droves, they just don’t have to call you first.
Impact / Reach
Another challenge facing analysts is that they don’t know how to factor community participation and usage into their models. One tactic has been ‘if you don’t pay for support, we don’t consider you’. Obviously this doesn’t work in the long term. Lets say Apache HTTP server ends up with 90% of the existing web server market, and is used in an equivalent number of installs at companies that could not have afforded WebSphere, WebLogic or IIS. From the perspective of the analyst the web server market has shrunk considerably, whereas it has actually grown considerably.
One way analysts could tackle this is to introduce a new visualization of the market – one that shows usage or impact or reach. Maybe the other axis could be ‘sophistication’ or ‘completeness’. In the BI market this would make sense because now Excel has a spot of its own. Excel is commonly used to solve BI needs but is typically not shown as its own point.