3 reasons why MBSE is not dead

Lately I've been reading more and more often that MBSE is in a deep crisis. Of course, this may be because I am in the filter bubble of my own network. It may also be because we are following the Gartner hype cycle and MBSE has passed the peak of inflated expectations. I don't think MBSE is dead, and neither do many others.

Call: I have been working with MBSE for many years and would like to bring it into the mainstream. At the moment, I am particularly looking for new customers or partners who have recognized MBSE as a strategic goal and want to implement it in their company with me. Is that the case for you? Then Let's have a quick chat.

By the way, I will be attending the ReConf on April 22 and giving a talk. The topic of systems engineering appears explicitly in the tracks this time. I'm looking forward to meeting you there (10% discount with promo code REC24_SPRE_1803 when registering).

Who says MBSE is dead?

The impetus for this article came from the following discussion by user Rhedogian at reddit. Various experts and companies took up this discussion in order to contradict it (more on this in a moment). However, MBSE has been called dead in forums for some time now.

Companies, on the other hand, often talk about how difficult MBSE is - and then of course offer their services. However, this is another way in which they are hitting the mark. Because they are at least admitting that MBSE is almost impossible for normal companies to implement without external help.

I have already expressed my personal opinion several times: I don't think MBSE is dead. However, I also don't believe that MBSE mainstream in the next few years at least not with the current approaches.

Points of criticism

However, the problem is clearly the current approaches. This raises the question of the extent to which we can revise them. Here are Rhedogian's main points of criticism:

• MBSE is another silo, the developed artifacts are not directly useful and utilization in the development process is difficult. Integrations do not work properly.
• The tools used for MBSE are not user-friendly and require a significant learning curve and expert knowledge to set up and use effectively.
• MBSE does not work without methodologiesengineers have little interest in learning them.
• The complexity of SysML makes it difficult for engineers to use the language effectively (and the tools are of little help).
• As nothing has changed for years, many engineers have low hopes for the coming years.

The conclusion is that while the concept of MBSE is sound, the practical implementation and tools are not yet ready for widespread adoption. This suggests that MBSE may be a specialized requirement in certain sectors rather than a universally useful engineering tool.

Reaction of the community

I am about the LinkedIn post by Robert Hämisch became aware of this discussion. Robert is the founder of the company Spicy SEwhich is developing a lightweight MBSE tool that does not require SysML. Spec Innovations, manufacturer of the Innoslate, has spoken up.

3 reasons why MBSE could soon become mainstream after all

Many of the MBSE experts with whom I regularly exchange ideas also disagree with this statement. I myself assume that much more will happen in the next few years than I would have dared to hope 2-3 years ago. I see three developments in particular as reasons for my optimism:

Much criticism is directed at SysML, and rightly so. In this respect, the development of SysML v2 is a real milestone. On the one hand, the language is much tighter and more consistent. New views (text representation) and integration options (API) will make it much easier to use SysML v2 data seamlessly in development processes.

At the same time, we keep seeing approaches and tools that manage without SysML. In addition to the aforementioned Spicy SE and Innoslate, these also include tools such as Valispace, iQuavis and Capella. Reducing the complexity of the language by dispensing with SysML is a bet that could pay off.

Artificial intelligence can drastically simplify many use cases in MBSE. Noteworthy, for example Spread.aiwhich extract their own model from the development artefacts behind the scenes and use it to implement use cases. My product Semiant extracts models from texts with the help of AI.

Much of what we are seeing at the moment is still hot air. Nevertheless, solutions are being worked on feverishly behind the scenes. Consulting and training services as well as presentations and events on the subject of AI & MBSE are increasingly appearing on the market.

The "big" tools for MBSE (Cameo, Enterprise Architect, Rhapsody) are all from the last century. And you can see that in the tools. All three are desktop applications that were not designed for integration.

Of course, these tools are regularly updated and expanded. But the architecture is in place. More modern, web-based modeling tools have been emerging for a while now. So far, none has challenged the top dogs for the dominant position. But I speculate that sooner or later a tool with a new paradigm will turn the market upside down.

While Innoslate can hardly be made to work without external consulting despite its architecture, Spicy SE was designed from the outset to be so lightweight that consulting could be largely dispensed with.

It is even more extreme, for example, with trace.spacewhich rely heavily on AI to remove barriers. However, I think the focus on requirements (instead of systems engineering / MBSE) is too limited.

Conclusion: MBSE must become more interesting for less complex systems

The problem with MBSE today is that the investment only pays off for very complex systems. As long as this does not change, MBSE will remain a marginal phenomenon.

But for MBSE to become mainstream, it must also become attractive for less complex systems. For this to happen, we need better tools and modeling approaches that work after a one-hour introduction. SysML v2, AI and new tools will hopefully make this possible soon.

In particular, I see an enormous opportunity through the new SysML v2 and the new possibilities of AI.