00:00:07 Introduction of the topic of buzzwords and false advertisement in the supply chain industry.
00:01:39 Explanation of the difficulty in adopting a new technology and the need for re-engineering.
00:03:05 Discussion of the superficial adoption of buzzwords in the enterprise software industry.
00:04:51 Explanation of how buzzwords play on the fear of missing out.
00:08:26 Explanation of how a buzzword can be technically true but deceptively so.
00:09:16 Explanation of software product marketing and the difference between client-centered and software engineering-centered perspective.
00:11:02 Explanation of lies in marketing through the example of car manufacturing.
00:12:43 Discussion of how top management in enterprise software companies may not have a technical background.
00:15:51 Explanation of how the lines between selling and lying are blurred in marketing.
00:17:33 Explanation of the history of probabilistic forecasts and how it became popular in supply chain.
00:18:43 Discussion of Loca’s probabilistic forecasting and its origin.
00:19:17 Discussion of marketing and the importance of communication in presenting information.
00:21:18 Discussion of probabilistic forecasting and the importance of tooling.
00:22:45 Discussion of Loca’s adoption of cloud computing and how competitors followed suit.
00:26:49 Discussion of the buzzwords in the supply chain industry and the misleading advertising of AI and machine learning.
00:27:46 Discussions about buzzwords in technology and demand sensing.
00:28:21 M5 Forecasting Competition and the absence of demand sensing.
00:29:37 Vendors using buzzwords to sell their product.
00:31:33 Difficulty of detecting a lie from an enterprise software vendor.
00:35:36 Getting a second opinion from another enterprise software vendor.
In this episode, Nicole Zint interviews Joannes Vermorel, Lokad’s founder, about supply chain optimization software and the use of buzzwords in the industry. Vermorel highlights the importance of letting go of old technology and embracing paradigm shifts to gain benefits from new technologies. He notes that vendors often use buzzwords to sell pretty brochures and empty promises, rather than delivering real value. Vermorel advises customers to seek public documentation and competitor evaluations to assess vendors’ claims. Lokad, an early adopter of cloud computing and probabilistic forecasting, emphasizes the need for proper implementation to maximize the effectiveness of these technologies.
In this episode, Nicole Zint interviews Joannes Vermorel, the founder of Lokad, a software company that specializes in supply chain optimization. The topic of the discussion is whether the supply chain industry is being sold pretty brochures and empty promises or if the software has kept up with the buzzwords of probabilistic forecasting, cloud computing, AI, machine learning, and blockchain.
Joannes explains that software is relatively easy and fast to adapt to new technologies, even bleeding edge technologies. The problem lies in letting go of the old technology that the new piece of tech is supposed to replace, which takes much more effort and time. If this is not done, then the adoption of the new technology will be superficial.
He provides an example of cloud computing, where the vast majority of vendors claim to be doing cloud computing but have simply copied and pasted their traditional way of doing things into the cloud computing infrastructure, with no added value except a little trade-off in terms of capex versus opex.
Joannes suggests that the adoption of a new technology should involve a paradigm shift, meaning that the way of looking at the problem should change. Simply duplicating what was done before in the new paradigm will not provide any benefits. It is necessary to approach the problem differently, taking into account the new situation and the new technology.
Joannes believes that the supply chain industry is driven by buzzwords, and vendors are quick to keep up with the trends. However, there is often false advertising, and customers are sold pretty brochures and empty promises. To truly benefit from new technologies, it is necessary to let go of the old technology and approach the problem differently, taking into account the new situation and the new technology.
They discussed the use of buzzwords in the tech industry and how they can be used to deceive customers. Vermorel explained that buzzwords are often used by marketing departments to play on the instinctive fear of missing out that many people have. He also explained that when buzzwords are used superficially, they can give the impression of being true but often fail to deliver on their promises.
The conversation then shifted to the challenges that new technological paradigms can pose for existing products. Vermorel explained that the initial design assumptions of software products made in the first year can have lasting impacts, potentially for decades. Therefore, when a new paradigm comes along, re-engineering all the other aspects of the product to fit can be a challenge. For example, with cloud computing, it is not just about adopting a new piece of software, but rather a completely different way of engineering software, where you choose a fleet of machines to cope with workload dynamically.
The interview also touched on the use of buzzwords in marketing brochures. Vermorel explained that it is not lying, but rather treating marketing like poetry. He added that the top management of many software companies are not software engineers but ex-consultants who may not care about the technical details of their own products. As a result, the marketing brochure is just a form of art that can be interpreted in various ways. Vermorel also addressed the blurred lines between selling something and lying, mentioning that the Roman law had the concept of dollus bonus, or the good lie, where a vendor could sell something by saying it was the freshest and the best.
Towards the end of the interview, Vermorel discussed how Lokad pioneered probabilistic forecasts in the supply chain. He mentioned that the idea had emerged in the late 70s in finance but was unrelated to the supply chain. Since adopting the idea, Lokad has faced competition from others who have copied their work. The conversation concluded with Vermorel urging customers to be better at identifying cheap tricks in the industry.
Joannes Vermorel discusses the company’s innovations and the challenges of communicating complex technical concepts. Lokad pioneered probabilistic forecasting in the supply chain field but faced difficulties in conveying its importance and the technical details to customers. Vermorel notes that competitors often copy their marketing strategies but leave out crucial technical aspects, which results in incomplete or ineffective implementations.
Probabilistic forecasting is useful, but it requires proper tooling to work with the generated probabilities. Vermorel finds it interesting that many industry players claim to use probabilistic forecasting, but their documentation often lacks information about working with probabilities. This gap in understanding makes their approach less effective.
Lokad was an early adopter of cloud computing, which many competitors followed. However, these competitors often simply moved their existing code bases to the cloud, rather than re-engineering their software for a true cloud-based solution. Vermorel highlights the importance of being able to create and access cloud accounts quickly as a litmus test for truly cloud-based solutions.
When asked about falsely advertised buzzwords among supply chain vendors, Vermorel identifies AI, machine learning, and to some extent, blockchain as misleading terms. While these concepts have real substance, it’s often unclear whether individual vendors are genuinely implementing them in their solutions. Additionally, buzzwords with little substance, such as “demand sensing,” become popular despite lacking scientific or algorithmic backing.
In order to assess the validity of a vendor’s technology, Vermorel suggests asking for public documentation and having a competitor evaluate it. He warns against relying on one’s own ability to detect false claims, as vendors are skilled at selling their products and may present convincing stories. By seeking external evaluation, customers can better understand the actual capabilities and limitations of the technology being offered.
They discuss the challenges of evaluating technology and vendors in the early stages of development, using Google’s search engine as an example. In its early days, Google’s technology was vastly superior to competitors like AltaVista, but this superiority was not easily discernable to users without a direct comparison.
Vermorel explains that when evaluating enterprise vendors, it’s important not to underestimate their ability to present their story in a way that highlights their strengths and addresses your weak points. To combat this, Vermorel suggests an adversarial approach: pitting one enterprise software vendor against another. By bringing someone with similar experience and skills to the table, you can expose potential weaknesses or conflicting views.
Vermorel likens this process to seeking a second opinion from a doctor. Just as you might not be able to assess a physician’s expertise without additional input, it can be challenging to evaluate the quality of a vendor or technology without a comparative perspective. By cross-referencing opinions and putting vendor claims to the test, you can make more informed decisions about the technologies and solutions that best suit your needs.
Nicole Zint: The supply chain industry is driven by buzzwords. A decade ago, probabilistic forecasting and cloud computing were not associated with supply chain, but now it’s all the hype. So is AI, machine learning, and blockchain. Vendors are quick to keep up with these trends, but so is false advertisement. Are we being sold pretty brochures and empty promises? This is the topic of today’s episode. Johannes, has the software really kept up with these buzzwords?
Joannes Vermorel: The thing is, and it’s relatively counterintuitive, when it comes to software, it is actually relatively easy and fast to adapt to a fancy piece of tech, even the bleeding-edge sort of things. But what takes much more effort and time is to actually let go of the stuff that the new piece of tech is supposed to replace. That’s the problem. If you don’t actually let go and do all the efforts so that you can let go, what you have is an incredibly superficial adoption of the new piece of tech.
Nicole Zint: What do you mean by superficial?
Joannes Vermorel: I mean that most novel technologies come with paradigm shifts, meaning that suddenly, you’re not looking at the problem the same way. The way you approach the problem can change dramatically. It’s very frequently easy to just duplicate what you were doing before in the new paradigm, but if you do that, there are zero benefits from being in the new paradigm. Just to give a concrete example, let’s talk about cloud computing. If what you’re doing is simply doing a cut and paste of the software that you had and you just move it to a third-party company that just happens to rent computing, then you can actually run a COBOL program that was implemented in the late 70s in the cloud. It’s not because it is now in the cloud that the COBOL program has anything more than what it used to have when it was not in the cloud. If you just cut and paste stuff, then usually, novel paradigms, novel pieces of tech, are as powerful as the old one in many ways, and thus, you can just copy and paste and transfer, but again, you will not gain anything. The vast majority of enterprise software vendors say that they are doing cloud computing nowadays, but what they have actually done is they just literally copied and pasted their way of doing things from the traditional way, which was in-house hardware, into cloud computing infrastructure where they’re just renting the hardware. But besides the fact that they’re renting the hardware, there is nothing that has changed and there is no added value except maybe a little trade-off in terms of capex versus opex, but that’s very minor in the end. So basically, the buzzwords keep up with the trend, but the tech does not necessarily do that.
Nicole Zint: So, the best way is something interesting. A buzzword is essentially something used by the socially marketing department to play on a very instinctive aspect of the…
%Nicole Zint: “What is the human psyche that is behind buzzwords in software development?”
%Joannes Vermorel: “The human psyche is simply the fear of missing out. There is this instinct that things should be better somewhere else, that the news is always better. And when you’re playing with these buzzwords, you’re just playing with this instinctive fear of missing out, that there is something that you should be doing and you’re not. By just adding the buzzword, you can make your offering look better, and if you decide to have just a very superficial adoption tack-wise of the buzzword, then it can be also very cheap and fast to actually give some modicum of truth to this statement, but only on a surface level.”
%Nicole Zint: “And that’s a problem because…?”
%Joannes Vermorel: “That’s a problem because many choices are typically made within the first year of engineering a software product that will have an incredibly lasting impact on the product, potentially for decades. And when there is a new paradigm that comes in, the challenge is that suddenly all your initial design assumptions might be completely off. It’s not about adopting the piece of software or tech that is difficult, it’s to re-engineer all the rest so that it fits into this new paradigm. For example, if we go to cloud computing, you could cut and paste your code and bring virtually nothing except a small Opex versus Capex trade-off. But if you start thinking about cloud computing as having dynamic access to computing resources, memory, CPU, storage, bandwidth, and you can dynamically adjust all of those elements to deliver a superior supply chain performance, that becomes a complete game changer.”
%Joannes Vermorel: “Nowadays, the vast majority of enterprise software vendors are claiming that they are doing cloud computing, but if you look under the hood, their product is still entirely driven by pre-cloud constraints that don’t really make sense in this new world. The initial constraints that were put in place by the first tech that got implemented gets carried on, and with the industry, chances are yes. And in terms of buzzwords, it is just about ticking boxes. You can just do the minimal effort to superficially do a little bit on the side so that you can take the box. Your brochure doesn’t have to be specifically reflecting the effort or the number of lines of code that you put into every single piece of tech that you put on display marketing-wise on your website. How easy is it to trick?”
Nicole Zint: That’s the interesting thing, is that are you tweaking, you see, and I’m asking this question because maybe you know, doing the difference advocates, but there are several layers to this question.
Joannes Vermorel: First, if you adopt somewhere in the organization in your product, you know the piece of text, then technically it’s not a lie. You know it is true that if for example, if you say “I am using cloud computing,” and you have let’s say a program that runs on, let’s say AWS, it may this piece of program might just be your website, but technically it’s true. So, it’s kind of technically true, so it’s not necessarily such a lie. That’s the first problem.
Nicole Zint: And then we can really argue exactly on what sort of degree, and that’s gets very, very fuzzy, because obviously when you want to present a software product, you cannot have a marketing description that is faithful to the implementation of the product. You’re not going to have an enormous description about low-level libraries that are completely inconsequential for the users, etc. So, you have to have a description that makes sense from the client perspective, not from the software engineering perspective who is actually engineering the product. And that’s fine.
Joannes Vermorel: Again, when you buy a car, the person who is the car manufacturer doesn’t tell you how many microns of paint there are on every single part, and what is exactly the specific process to actually paint every single metal part in the car. You see, this is kind of irrelevant beside the point. You just trust the car manufacturers to do a good job when it comes to painting. You see what I mean. So, there are tons of things that are very important, but you can’t just translate it to sales language.
Nicole Zint: Yes, you have to make choices. You can’t produce a 100,000-page document that says everything there is to say about your product. So, you have to make choices and obviously lie by omission if only by necessity, because the full description would be insanely long and also insanely uninsightful. Just think about if you were to describe a car starting by all the thickness of paints on every single part. That would tell you almost nothing about the car, and you would be very, very confused to make even a judgment about whether you’re actually doing something reasonable or not.
Joannes Vermorel: Now, there is a second layer to that. So, that’s just necessity, etc. But there is also a second aspect that is very interesting, which is, is it a lie, and the question is lies in the mind of the person who is doing the marketing effort.
Nicole Zint: And here there is something that is very peculiar in what I’ve seen in enterprise software, is that as far as tech is concerned, usually it is completely disregarded in the sense that typically the top management of many software companies are not software engineers. They were ex-consultants, they were people who fundamentally don’t really care about the software. It may seem puzzling because this is a software industry, I mean, they do care very much about the sort of problem and industry they’re working on. I’m not saying that they don’t care, it’s just that they don’t care exactly about the same thing, and they’re not naturally caring that much about the nitty-gritty details that are going under the hood. So, as a consequence.
Nicole Zint: I have found a particular trait in my discussions with my competitors. They are very ignorant about the fine print of the technical details of their own products.
Joannes Vermorel: The executives of most enterprise software vendors are not knowledgeable about the technical details of their products. They approach marketing as poetry, where truth is beside the point. It’s just a matter of making something look good.
Nicole Zint: I’ve had discussions with top executives who were supposed to be in technical positions, but even they were completely baffled by questions about the processing power of their software. It was as if I was asking them about their sumo skills.
Joannes Vermorel: If you’re not really interested in the implementation details of your product, most technical questions are beside the point, and your marketing brochure is just a form of poetry or art.
Nicole Zint: The lines between selling something, the art of selling, and simply lying are blurred. This is something that I address in one of my lectures. Even Roman law had this concept.
Nicole Zint: “Uh of uh dollus bonus the good lie you know when you go to a market and there is somebody selling fish to you it says this is the freshest fish ever and this is the best fish you will ever test this is kind of you know an accepted lie and it’s it’s okay. But again what is very interesting here is that um buzzword is going beyond that and it’s playing you know it’s not just saying I have the best product uh it is playing on a very specific trick which is again this fear of missing out, keeping up with the trends.”
Joannes Vermorel: “Exactly, and uh and uh and and basically you you’re trying to create some degree of urgency.”
Nicole Zint: “And uh. And I think it’s um the it’s kind of okay vendors do what vendors do. I think that’s that’s uh that was you know the lesson from the roman law on on this uh dollar’s bonus is that um you’re not going to have a law to condemn vendors for doing what vendors do. It’s kind of beside the point this is just in the nature so the problem that I see is more on the client side is to be better at identifying this sort of uh of I would say relatively cheap tricks.”
Joannes Vermorel: “Out of curiosity, so locate we pioneered essentially probabilistic forecasts but now it’s all the hype as we mentioned um has luck had been copied so yeah as um to narrow the claim to narrow the climb um probabilistic forecast emerged by the very late 70s um and but in areas that were completely I would say unrelated to um supply chain um it was it was it appeared first in finance and then in the 90s it became I would say gradually adopted um in in in areas like meteorology and and some and some I would say heart sciences look at pioneer the idea of applying this uh in supply chain and we and that was basically a decade ago and indeed uh and and to use that um as a as a primary I would say technical mean to address uncertainty indeed um since that time we have been I would say um copied by by many competitors who are now putting you know probabilistic forecasting on their website on the brochure etc and so so yes in this sense you know we have been copied but again this idea did not originate from loca did it originate from other people exercising the idea to bring probabilistic forecasts to supply chain that’s sort of what we pioneered essentially yes.”
Nicole Zint: “Um and following up from that so to do probabilistic forecasts you do need probabilistic algebra has that been also included in this copy pasting?”
Joannes Vermorel: “That’s something that is very interesting is that again as I was saying earlier in the interview you have to make choices of what do you put forward you know going back to the car you can’t really detail that you have to have all those details about the paint and the thickness of paint on every single particle so so you need in your communication to leave out details some of them are actually very very important to for the thing to work because the reason we’re leaving out detail isn’t it because we simply doesn’t have to because if we if we we can in fact tell our customers these paint layers on these carbs yes but by the way we do we do we have a technical documentation that details that but if we do we also have to sort of we have essentially made a claim that we can directly check and yes we don’t make this claim we can sort of yeah but that that’s again that goes back to
Nicole Zint: And so it’s interesting because our competitors they copied the marketing aspects but they completely leave out the sort of technical details that are necessary, that’s the backbone, basically.
Joannes Vermorel: Yes. And where it gets even funnier is that when they kind of spin off on the buzzword on their own, using the same buzzword, but they get a bit inventive in ways that are actually not really working in terms of tech. But because the lack of technical knowledge is absent, it’s not really a problem, it’s just a pure marketing exercise.
Nicole Zint: What is very interesting if we want to go back to this very specific point of probabilistic forecasting is that indeed probabilistic forecasting is interesting, but if you don’t have the tooling to work with all those probabilities, then you’re not doing anything with that. And very quickly, it is losing all practical interest.
Joannes Vermorel: Yes, I mean, if I tell you we can generate huge matrices of probabilities for your supply chain, the answer of a reasonable supply chain practitioner would be, “So what? I’m not going to do anything with those probabilities.” Those are artifacts, they don’t do anything for my supply chain. Only through a careful exploitation of those probabilities can I do something of interest, and it turns out that you need some tooling to do so.
Nicole Zint: Besides realistic forecasting, has anything else been copied from Lokad?
Joannes Vermorel: Lokad adopted cloud computing, so we were early in this area. I would not say we were pioneers, at least not in the realm of enterprise software. There were clearly people like salesforce that were before us, specifically in supply chain, I believe we were very, very early, maybe not the earliest, but among the earliest. And when Lokad was putting cloud computing on our websites, plenty of other companies followed suit. But again, what they did was mostly just cut and paste all code bases into the cloud and then say, “We did it, and it’s now cloud-based.”
Nicole Zint: By the way, a simple litmus test to know if it’s cloud-based is the following: can the vendor create an account for you in two minutes and let you work with a blank state instance of cloud computing that you can access at any time?
Joannes Vermorel: Yes, I mean, it doesn’t define cloud computing today, but fundamentally, if you have engineered your stuff for a cloud computing platform, there is zero reason why you shouldn’t be able to do that.
Nicole Zint: So at the moment, what do you see gets falsely advertised the most among supply chain vendors?
Joannes Vermorel: I guess I would say the trendy 20 buzzwords that are very misleading are probably AI, machine learning, and maybe a little bit blockchain. Blockchain tend to not be misleading in the sense that they’re not doing blockchain, they’re only misleading in the sense that they can easily create added value for companies. But nowadays, I would say AI and machine learning are the interesting things, those buzzwords are kind of real. The question is, when an enterprise software vendor advertises buzzwords, the question becomes, is there any substance for this particular vendor with respect to this buzzword.
Nicole Zint: That’s the difference. You’re not saying, for example, the research team of Facebook has been publishing continuously for probably something like five years, papers that very much qualify as progress towards something that would one day maybe be qualified as artificial intelligence.
Joannes Vermorel: Yes, that’s right. The research team of Facebook is absolutely real and truthful in their characterization of the intent and effort and result that is being delivered. However, I would not say the same thing for most of the enterprise software vendors. And then we even have buzzwords that are lacking any substance behind it, such as demand sensing. Demand sensing is just supposedly a technique that gives you more accurate forecasts, but if you look at the last forecasting competition of the M5 competition, working on Walmart, there is like zero scientific publication, and zero algorithms that have been pre-published or recognized.
Nicole Zint: What was very interesting about the data set?
Joannes Vermorel: The data set was interesting because there were zero competitors who claimed any results using anything that could be qualified as dimensioning. I checked the first 100 competitors and none of them published their results. Even though there were 900 teams, not a single one claimed to use anything that looked like dimensioning.
Nicole Zint: How is it possible that nobody is claiming to use dimensioning?
Joannes Vermorel: It’s very interesting. If you come up with a good buzzword, like “demand sensing,” then vendors will copy that, regardless of whether it’s real or not. They will present their product as if it was poetry and reuse those terms. As a result, you end up with a trail of companies using the same buzzword, even if there is still zero substance attached to it.
Nicole Zint: What sort of questions can you ask vendors to challenge their tech?
Joannes Vermorel: That’s a tricky question because you’re not an expert. When you’re buying software, you’re discussing with vendors who are trying to sell their stuff five times a week, whereas you only buy software once every five years. Vendors are professionally trained to convince you that their stuff is good, and enterprise software is immensely profitable. As a result, it’s difficult to detect the truth. I suggest going for the public documentation and asking a competitor to make the assessment.
Nicole Zint: What if you ask them what broke in their tech when they implemented a new tech?
Joannes Vermorel: They will give you an inspiring story about how initially everything was broken, but then they realized all the problems and the tech got so much better in the process. But don’t trust your capacity to detect the truth too much, instead go for the public documentation and ask a competitor to make the assessment.
Nicole Zint: They are incredibly good at that, but what if the answer is that nothing broke, essentially we implemented it and now it’s working fantastically? Is that a red flag?
Joannes Vermorel: Maybe, but again, you see the thing is that maybe it’s true. Maybe it’s true. You see, when Google started to have their search engine technology, the interesting thing is that one of the earliest investors, a business angel, realized that at the time, we were in something like 1996. Essentially, search on the web at the time was like Alta Vista as a search engine. Web search was completely broken. When you were searching for something like IBM, you would get a website called ibmibmibm.com three times IBM instead of getting ibm.com. So, the search on the very early web was completely broken. The interesting thing was that the Google search technology was radically better from day one. It was massively superior to anything that was outside Google.
So, yeah, you see, again, if they were, if you were asking Google, “Is your technology better?” They would say, “Yeah, and how long has it been better?” They’re from day one. So you never really know. So now, again, I would say, I don’t underestimate that. If you’re dealing with an enterprise vendor, the person is going to be able to tailor the story they tell to you so that it fits your weak points. And they can adjust, and they will know whether to sort of…
You see, again, you need to have a question where they can’t cheat, even if they are better than you. And what I’m saying about this sort of technique, which is an adversary technique, is that how do you defeat the lie of an enterprise software vendor? The answer is with another enterprise software vendor, you know, fire against fire. And so, you will have somebody you just bring to the table, somebody who has the same sort of experience and skills, and then you let those things do the work for you. And then, in the end, what you will have is conflicting views, but exactly, that will give you the angle of attacks for what should you be pressing, instead of just going blindly and trying to trust your own intelligence to be able to figure it out.
I’m not saying that you’re not intelligent; I’m just saying that if I go see a doctor and this doctor tells me that I have a condition and I need to do that, and they will give me an explanation, this explanation is very likely to sound incredibly reasonable just because I know nothing about medicine. And so, if I want to do that, there is no question I can ask this physician to know if this physician is good or bad. I won’t be able to assess that. The only thing that I can do is to go to another physician and have a second opinion. And that might work. Actually, this is, in the real world, this is how you do it. You get a second opinion, and you can even cross-reference the opinions.
Nicole Zint: Very interesting advice there, Joannes, on how we can assess our vendors better. Thank you very much for your time today. Thank you for watching, and we’ll see you next week.