Is it? Or is it not?

[Casey Rutland]

Is it? Or is it not?

 For a few years now, parts of our sector and indeed other sectors, have been researching, defining and promoting digital twins.  If we observe anything, it’s that chatter (including within the DT Hub) has been rife with the ‘what is/isn’t a digital twin...’

I’m no expert, and don’t yet claim to offer a resolution to clarify the topic, but I do think a discussion hosted within the DT Hub would be of use.  This discussion is something that will provide greater clarity and implementation for those less involved in this definition process and yet vitally important to the delivery of whatever a digital twin of the future is destined to be.

Let’s learn from BIM implementation

I wear many hats in my career and most of them are related to the implementation and ‘normalisation’ of BIM processes. As Vice Chair of the UK BIM Alliance and Chair of the UK & Ireland Chapter of buildingSMART International, I’m afforded a view of the sector from various different levels of stakeholders and the challenges they face in an ever-changing world as they prepare to digitalise.  The silent majority are perhaps the key to unlocking the transformation to a digital sector and it’s vital that the BIM message reaches them and connects in a meaningful way to each and every one of them... BIM in the UK has been ongoing for over a decade and my feeling is that there is at least another to go before we reach ‘business as usual’.  It’s exactly the same for Digital Twins.

All vocal parties involved here in the DT Hub seem keen to navigate more smoothly through the same sectoral challenges and one of those, in a similar vain to BIM, is “is this a Digital Twin or not”?

Acknowledging that BIM in the UK has formerly been going through the same sector engagement, we can also see similar issues appearing now with the concept behind Digital Twins being taken over by technology providers rather than sector stakeholders and subsequently being marketed in that way.  It’s by no means a UK-only challenge, with many global discussions observed.

Hence, we’re rapidly on the way to Digital Twins being defined by technologies rather than their use and value to us as people.  A human-centric approach to any digital transformation will almost always achieve greater adoption and ultimately ‘success’ than one led purely by technology. Hence the CDBB National Digital Twin Programme envisages the built environment as a system of systems, comprising economic infrastructure, social infrastructure and the natural environment.  The CDBB Gemini Principles neatly position Digital Twins in a way that forces one to consider the overall business need (the ‘why’) and all the potential societal benefits.

Other DT Hub discussions have touched on the possibility of a Turing-type test.  The original Turing test was created by Alan Turing to determine whether or nota machine was discernible from a human.  Whilst the test is valuable for determining artificial intelligence, it’s also one that is evaluated by humans and hence quite challenging to ensure all evaluators are equal. Perhaps a technology-driven test that provides both a score and a ‘time taken’, introducing a level of competition between creators of Digital Twin systems might help adoption.

 

So here’s the proposition... we hold a workshop (or two) to discuss and investigate the need for a test, the type of test, ‘what’ is being tested, what the thresholds might be, and anything else that’s relevant to the topic of ascertaining whether or not someone’s proposed Digital Twin is actually a Digital Twin.

I have three questions to start the discussion here in this thread...

1. Do you feel the need for a ‘test’ to determine whether or not a Digital Twin is a Digital Twin?

Can we continue without a formal ‘test’ or should we actively seek to develop something absolute to filter out what we’ve been able to do for many years and focus on true Digital Twin solutions and the search for the allusive Digital Twin unicorn?!

 

2. If we do need a test, will a simple yes/no suffice? Or does a ‘score have more longevity?

If you ever saw the HBO series Silicon Valley, you may be familiar with the Weismann Score, a fictional test and score for online file compression.  It enabled the fictional tech companies to demonstrate the success of their software and algorithms by testing their performance for file compression.  Would a similar test be suitable for our purposes, with a threshold for determining if a proposed Digital Twin is a Digital Twin and would it then cater for future digital developments and emerging technologies?

 

3. Finally, are you keen and able to join a virtual workshop?

 

 

[Samuel A Chorlton]

Thanks for posting this @Casey Rutland I think that this will stimulate some really interesting conversation and is particularly important at a time when such a high degree of ambiguity exists around what is and what isn't a Digital Twin. @Ian Gordon, @HenryFT and @Matt Edwards would be great to start off this conversation with your views.

[Samuel A Chorlton]

Thanks for posting this @Casey Rutland I think that this will stimulate some really interesting conversation and is particularly important at a time when such a high degree of ambiguity exists around what is and what isn't a Digital Twin. @Mark Coates, @Tanguy Coenen and @Adam Beck would be great to start off this conversation with your views.

[Henry Fenby-Taylor]

Well, first off, let's have that conversation! I think we need to have it.

This article made me think. On reflection, I have two concerns about the direction this conversation is taking.

One, that we don't want to see our businesses and our clients' organisations sold products that don't deliver the benefits of digital twins. The digital twin movement loses, the client loses and the winner doesn't create a sustainable business for themselves so ultimately they lose too. We're taken in by gimmicks, we create wealth in a short sighted fashion and productivity doesn't improve.

Two, once something is decided, that means we can stop thinking about it. That's the point of making decisions. So if it was decreed that only a physical system connected to a digital management system providing real time data counts as a digital twin then we would lose out as well. Suddenly certain data sets, applications and approaches 'don't count'. Everyone loses again, the digital twin movement loses key collaborators and innovations, these brave outsiders lose, because they're out of the fold pushing a genuine solution that the general consensus tells people isn't worth it and the clients lose, because despite having a solution that ticks all the boxes and 'counts', they don't have a solution that serves their actual needs (I'm thinking of logistics and city management systems here).

I don't think there is a complete answer, in the words of Michael Grieves "it's an analogy, so don't take it too literally". So we should be careful about going into the depths of over qualifying what is and what isn't a digital twin. At the same time we should be able to recognise what is and isn't a digital twin and even more than that we should be able to recognise the genuinely ground breaking innovative work when it happens. 

 

Edited October 6, 2020 by HenryFT

[Henry Fenby-Taylor]

Well, first off, let's have that conversation! I think we need to have it.

This article made me think. On reflection, I have two concerns about the direction this conversation is taking.

One, that we don't want to see our businesses and our clients' organisations sold products that don't deliver the benefits of digital twins. The digital twin movement loses, the client loses and the winner doesn't create a sustainable business for themselves so ultimately they lose too. We're taken in by gimmicks, we create wealth in a short sighted fashion and productivity doesn't improve.

Two, once something is decided, that means we can stop thinking about it. That's the point of making decisions. So if it was decreed that only a physical system connected to a digital management system providing real time data counts as a digital twin then we would lose out as well. Suddenly certain data sets, applications and approaches 'don't count'. Everyone loses again, the digital twin movement loses key collaborators and innovations, these brave outsiders lose, because they're out of the fold pushing a genuine solution that the general consensus tells people isn't worth it and the clients lose, because despite having a solution that ticks all the boxes and 'counts', they don't have a solution that serves their actual needs (I'm thinking of logistics and city management systems here).

I don't think there is a complete answer, in the words of Michael Grieves "it's an analogy, so don't take it too literally". So we should be careful about going into the depths of over qualifying what is and what isn't a digital twin. At the same time we should be able to recognise what is and isn't a digital twin and even more than that we should be able to recognise the genuinely ground breaking innovative work when it happens. 

[DRossiter87]

Fab, thank you for posting this @Casey Rutland. 

Interestingly, I listened to a Digital Twin Consortium webinar last night (sorry for cheating on the DTHub!).  The narrative being used was around how BIM had failed to make real change and this (DTs) was almost the next attempt.  As @HenryFT mentions above, we don't want digital twins to be mis-sold. 

I think understanding the distinction of what makes a digital twin "different" is an important one to make. 

[DRossiter87]

Fab, thank you for posting this @Casey Rutland. 

Interestingly, I listened to a Digital Twin Consortium webinar last night (sorry for cheating on the DTHub!).  The narrative being used was around how BIM had failed to make real change and this (DTs) was almost the next attempt.  As @HenryFT mentions above, we don't want digital twins to be mis-sold. 

I think understanding the distinction of what makes a digital twin "different" is an important one to make.

[Guest]

1. Do you feel the need for a ‘test’ to determine whether or not a Digital Twin is a Digital Twin?

Probably no at this stage, as things keep evolving. Firguring out how to test, by whom, and why even borther a test can take some time, that may prevent just doing something and prove its value. Besides a digital model, a digital twin can be a process, a workflow, an extesnion to the physical world, or even an aspiration. 

3. Finally, are you keen and able to join a virtual workshop?

YES, would love to know more details! @Casey Rutland

[Tammy Au]

SAVE THE DATE

We will be hosting a workshop to explore this topic further on the 17th November from 14:00 - 16:00.  Details and invites to follow.

All comments and challenges welcome!

[Tammy Au]

SAVE THE DATE

We will be hosting a workshop to explore this topic further on the 17th November from 14:00 - 16:00.  Details and invites to follow.

All comments and challenges welcome!

[David Roberts]

I think I would also agree, trying of be overly prescriptive in a tight definition of a Digital Twin could be counterproductive.

Digital Twins suffer from the same problem as Artificial Intelligence, as the English grammar goes the term is very simple and self-descriptive, but what it is trying to describe is very broad and complicated and so it just leads to an easy misuse of the term as it just “sounds” so simple.

Trying to tightly bind a detailed technical definition to a simple term will never remove the misuse and repeated misunderstanding.

I think I would be more interested in trying to define a spectrum of digital twin capabilities. Again, like AI, where we have Local, Broad and Extreme generalisation ability, perhaps in digital twins we could have tightly coupled and loosely coupled and then perhaps synthetic copies?

If we are striving for complex interconnectivity of digital twins from different sectors and representing different technologies and objects surely, we would be wanting to cast our net as wide as possible?

[Chris Dent]

I hope to join the workshop, and indeed am just writing a blog post on the definitions of AI and DT.

My view is generally that trying to define what is and is not a DT is not a helpful pursuit, at least not in the kind of work I do, as it's irrelevant in most practical questions. The relevant question is what the appropriate decision support analysis should be, not what we call it. The caveat on this is that I mostly work in planning and policy rather than detailed engineering design or system operation - except for specialist interest in power system control room algorithms.

When the post goes up (need to finish editing it) I'll flag in this DT hub conversation.

[Henry Fenby-Taylor]

5 hours ago, Chris Dent said:

I hope to join the workshop, and indeed am just writing a blog post on the definitions of AI and DT.

My view is generally that trying to define what is and is not a DT is not a helpful pursuit, at least not in the kind of work I do, as it's irrelevant in most practical questions. The relevant question is what the appropriate decision support analysis should be, not what we call it. The caveat on this is that I mostly work in planning and policy rather than detailed engineering design or system operation - except for specialist interest in power system control room algorithms.

When the post goes up (need to finish editing it) I'll flag in this DT hub conversation.

Look forward to reading it. @Chris Dent

[Henry Fenby-Taylor]

On 20/10/2020 at 20:49, David Roberts said:

Digital Twins suffer from the same problem as Artificial Intelligence, as the English grammar goes the term is very simple and self-descriptive, but what it is trying to describe is very broad and complicated and so it just leads to an easy misuse of the term as it just “sounds” so simple.

Trying to tightly bind a detailed technical definition to a simple term will never remove the misuse and repeated misunderstanding.

I feel like the popularity of the term AI is probably because of the simplicity, but its progress as fields of study is almost despite it @David Roberts !

[Matthew West]

In the NDT programme we are taking a very broad view of what a Digital Twin might be. This is because we do not want to find ourselves in the position of being faced with a valuable opportunity and having to say "You can't do that here because that does not meet our definition of a Digital Twin". I think that would be absurd. So I work off a Digital Twin being a useful collection of data about something, which might be an asset or a process (or something else).

That is not to say that there are not some particular types of Digital Twin that are more tightly defined. For example, a Connected Digital Twin being one where there is measurement of the Physical Twin and active feedback from the Digital Twin to effect change in the Physical Twin. I think it makes a lot of sense to identify useful types of Digital Twin like that.

[Andy Parnell-Hopkinson]

Ah, good. Saw the event invitation and followed the breadcrumbs to here. Registered, and looking forward to the webinar. 

My 2c is that a DT, like a BIM, should have a functional or data set threshold but once passed, you can pretty much put whatever you like in there (cf Dan Rossiter's thin v fat twins). 

Seems to me most tech suppliers who claim to offer DT solutions actually do offer a part of a DT. The difficulty comes when you start piling these parts on top of each other and they eventually become unstable and unusable.

Actually scratch that - the difficulty comes when you have to ask your boss for a ton of cash to replace all the shiny new things you bought.

[Alexandra Robasto]

On 20/10/2020 at 13:18, Tammy Au said:

SAVE THE DATE

We will be hosting a workshop to explore this topic further on the 17th November from 14:00 - 16:00.  Details and invites to follow.

All comments and challenges welcome!

 

[Alexandra Robasto]

 

[Casey Mullen]

I tend to agree with Chris Dent that "trying to define what is and is not a DT is not a helpful pursuit".  A given system might technically qualify as a digital twin system, and it may completely satisfy one user's use case but fail to satisfy another user's use case requirements.

An owner buying a digital twin system should probably be asking "Does it satisfy my use cases?" rather than "Is it a bona fide digital twin system?"

[Chris Dent]

Here is the blog piece which I trailed earlier - and you get "artificial intelligence" as a bonus!

https://centreforstatistics.maths.ed.ac.uk/cfs/news/buzzwords-unwrapped-artificial-intelligence-and-di