Category Archives: rants

It’s Not an AI Hype Cycle – It’s an AI Hype Wave!

Is it yet another AI hype cycle? This question was posed by Vinnie Mirchandani over on deal architect and it’s a good question.

AI has been the dream since the 1950’s when computing was just beginning and Alan Turing defined the famous Turing test (which is, sadly, not necessarily enough anymore because parallel computing and wide-scale sampling of real conversations can be used to pass the Turing test in many situations), but we’re not there yet. Enhancements to computing power, storage, and algorithms have made automated reasoning programs exponentially more powerful than they were sixty years ago, and today’s predictive capability is essentially unmatched … with many algorithms outdoing the best human experts.

But they are still not intelligent. They’re all based on data and trends which, for the most part, are predictable, and that’s where they excel — but where they are not, or where those trends can change on a dime, they still fail … some times spectacularly. Plus, if the data is incomplete or bad, they don’t always see this … and predict a wrong outcome. Human experts, on the other hand, can see missing data, outrageous trends, and when trends are inflecting in unexpected ways.

We’re nowhere near the point where we can even truly start using AI with a straight face, and, as per a quote from Yann LeCun, the director of AI research at Facebook, as quoted by the deal architect, despite these astonishing advances, we are a long way from machines that are as intelligent as humans — or even rats. So far, we’ve seen only 5% of what AI can do. And I’d say that 5% is an overstatement.

But, as the deal architect has indicated, we are seeing a huge increase in they hype about AI. As indicated by the deal architect, IBM has been marketing Watson since before it should have been announced, technologists are claiming that AI can replace entire fields, and Cambridge University has created a Center for the Study of Existential Risk to study extinction-level risks that can emerge from technological advances (like AI). Apparently, the hype is back … but the reality is that it never went away. It’s a continuous wave that crests and troughs and crest and troughs … it never goes away. The believers, for better or worse, won’t let it. There hasn’t been a decade where the hype hasn’t increased unnecessarily since the 50’s, and we’ll never see the end. Because until someone makes a true quantum leap, it will continue to be gradual progress, decade after decade, and the hype will continue ad-nauseum.

Supply Chains Are Complex … and the Earth is Round.

Hey, some of you might not know the earth is round! It’s only been 70 years since the first pictures of earth were taken from an altitude greater than 100 miles in space (and, up until that time, the non-believers could demand visual proof)! (To be precise, the first pictures of Earth as seen from an altitude above 100 miles was on March 7, 1947. Source: NASA)

But to not know that supply chains are complex, when “global” trade is almost as old as civilization (as purchasing is, of course, the world’s second oldest profession until such time as someone can definitely prove astronomy came first), that’s, well, really unthinkable. But yet, APICS and Michigan State University just gave us yet another report that announced yet again that supply chain leaders are citing “complexity” as the top supply chain challenge. Moreover, they decided to dive into the sources of complexity and found, surprise, surprise that they are:

  • customer accommodation
  • operations globalization
  • supplier (local sourcing) complexity
  • supply chain trends

But there’s nothing new here either. Let’s take ’em one by one.

The number of variations of a product desired is equal to the number of customers you ask. Period. Has always been. Has always will be — so the more customers you try to accommodate, the more complex your product variations, and supporting supply chain, becomes. And we’ve known this since long before Marshall M. Kirkman wrote the first Purchasing Manual.

Of course the supply chain becomes more complex as you go more global. Every locale has the potential to add languages, currencies, culture, local regulatory requirements, logistics challenges, border challenges, and so on and so on.

And then there are all the local issues faces by the suppliers — additional regulatory requirements, sustainability and CSR efforts to stay off of boycott lists, local workforce challenges, local disruption and disaster risks, and so on.

And of course trends affect complexity. They are usually the source … but they are not new issue. As we laid bare in our “future trend expose”, of the 33 trends commonly cited as future trends, only 3 were really relatively new, and only 1 was really a future trend.

Complexity has always been here, and the more global we get, the more complex we get. Nothing has changed, and if it’s not completely obvious at this point, you’re in the wrong profession.

That’s why SI has been preaching optimization and analytics since day one, since those are the only advanced sourcing solutions that can really handle the complexity of modern supply chains.

Has the Death of the Enterprise Suite Been Exaggerated?

There’s a big movement towards best of breed in many Procurement, and a bigger promotion on the part of many vendors towards such a movement. It makes sense, given that there are few suite vendors in the space compared to point-based best-of-breed vendors, which constitute the majority of vendors. But does that mean that the end of the enterprise suite is near?

Over on deal architect, the disruptor asks are enterprise suites dying? He notes the big disconnect between vendor and customer talk, including the facts that:

  • the last generation of vendor suites disappointed
    and that most customers ended up buying a number of “ring fence” applications and customizing the packages to meet their needs
  • the concentration of dollars with a handful of suppliers led to “lock-in” and bad vendor behaviour
    and the promise of economies from vendor reduction just stayed a promise
  • enterprises are finding a wide array of cloud applications
    that they themselves are weaving together

And these are true, and yes many burned CIOs are now anti-suite, but does this mean the suite is dead? There are still a number of managers and executives out there with the one-suite solution dream. Plus, as best-of-breed solutions pile up, the number of solution providers an organization has to deal with gets unmanageable for those CIOs who are under-staffed and under-resourced. So even if they don’t like the current suite providers, as too many solution providers creep in, they’ll like that situation even less.

So, while suites haven’t lived up to their promise, as Procurement organizations try to support their operations end to end with platforms, they are going to want less providers, not more. While they may not centralize on one suite, chances are they will want to centralize on a small number of multi-application vendors, and isn’t that just the definition of suite?

The Big Bad Blockchain

It’s not just the big bad wolf you have to worry about, it’s the big bad blockchain … especially when it becomes disassociated with bitcoin.

Bitcoin, which is neither good nor bad (it’s just another currency), is powered by a special type of blockchain — one that is decentralized, open, and auditable by anyone using the system. And one that is extremely hard to alter because altering a block (to steal the currency) would require not only require that every other block after it were regenerated, but all copies of such block (and there can be multiple as all nodes participating in the decentralized system can store a copy of the block) be replaced simultaneously.

Theoretically, this can be used to record supply chain transactions because we could create a virtual currency for any real world unit we want to trade (such as CELLcoins, RAMcoins, etc.) but could also encode other related information in the hash and serve as a fully auditable record of ownership of the corresponding products end-to-end, source-to-sink, but this would require that a similar, truly open, system be developed.

But right now, when you get down to it, the most “open” blockchain proposals are akin to the most “open” supplier networks … and there are no truly open supplier networks. Every supplier network of note in existence is owned by a for-profit company and even those with “open” APIs or “open” integration are not really “open”. Yes, there is an API that third parties can integrate into, but in every case there is a “catch” — either the integration is limited or integration is only permitted at a “price”. It’s not open and free — and “openness” is only “open” as long as the network decides they are “open”.

If any company owns the blockchain, then it is not truly open, and not truly decentralized, open, and auditable by anyone and everyone. And that is necessary for a true block chain solution. So until a global non-profit conglomerate with Procurement punch and tech-chops steps up and creates a truly open, decentralized, non-corporate controlled block-chain solution, let’s stop pretending blockchain is the solution we’ve been waiting for … because all we’re going to get is a prison for our data that will come with hefty prison maintenance fees. And then we won’t just be talking about how “Ariba doesn’t have customers, it has prisoners” [Spend Matters]. We’ll be talking about how BlockChain Company X has everyone’s supply chain data as prisoner.

So while you continue your Bitcoin Buoyancy, we’ll keep our expectations realistic.

Vendor Scorecards DO Work – But Only if They are Done Right!

A recent guest post over Spend Matters by Andy Kohm, founder of VendOp, provided “4 reasons why supplier scorecards don’t work”, which is a terribly inaccurate and a disservice to the procurement space because

  1. They Do Work if done right and
  2. what he was describing was internal vendor surveys, NOT scorecards.

Even worse, if he had said internal vendor surveys don’t work, SI would have totally agreed and hailed the post because, frankly, internal vendor surveys don’t work. Expecting enough people to fill out enough long surveys to get statistically reliable data when everyone is overworked, underpaid, and tired of doing everyone else’s job (because no one has time to do their own) is just ludicrous. It’s not going to happen, and when it does, the data and answers are not going to be that good or reliable because the surveys will be filled out in a rush. And all the reasons provided by Mr. Kohm will hold true.

But you see, a scorecard, at least a proper scorecard, is not a survey, or a summary of soft, qualitative feedback survey scores, but a summary of hard, quantitative metrics built up from hard data over time. A scorecard summarizes hard performance metrics, KPIs, and unarguable (undisputable) incident counts, not subjective scores on reliability.

We have to remember that just like anchoring can be a problem in negotiations, it can be a problem in subjective ranking. If the last couple of interactions with the supplier were problematic, the recipient is likely to fill out a fairly negative score even if the 20 interactions before that were great and, overall, the supplier is batting 800. Similarly, if the last few interactions were particularly good (because the supplier knows their review is coming up and making extra effort just to score enough to pass), the recipient may rank the supplier very positively even though 8 out of 10 requests are ignored on average. In short, for reliability, surveys suck.

But hard scorecards, built on on time statistics, reject rates, incident counts, billing accuracy, and so on are unbiased, anchored in fact (and not fiction), and work. They allow both parties to zero in on true issues, problems, and disagreements, and work collaboratively to fix them. They are the best supplier relation management tool the average organization has at their disposal and should not ever be discounted. Proper scorecards are the solution, not the problem.