Creating ROI models for platform engineering is difficult. Here’s three examples of approaches I’ve come across recently.

You’re trying to convince your organization to put an app platform in place (probably either buying one or building one on-top of Kubernetes), to shift your ops team to platform engineering (just after HR finally changed titles from “Systems Analyst II” to “DevOps Engineer”!), or, if you’re like me, sell people a platform.

ROI

“Yeah, but what’s the ROI for it?” the Director of No responds. What they mean by that is “convince me that this change is going to have benefits that we can represent as money, either saved or gained.” A variation is “show me that what you’re proposing is cheaper than the alternatives, including the alternative of doing nothing.” That’s probably more of a “Total Cost of Ownership” (TCO) analysis. Indeed, ROI and TCO models are often used the same way, if not the same spreadsheets. This kind of analysis is also often called a “business case.”1

This is especially true in the post-ZiRP world. When money was “free” and G2000 companies were deathly afraid of Tech Companies, they’d change how they operated based on the capacities they gained, not just on an Excel spreadsheet filled with cash-numbers. Those were good times!

Showing the ROI of a platform is difficult. I haven’t really come across any models that I like, and I’ve seen many of them.

The problem is that platforms don’t generate money directly, so you have to come up with a convincing model that shows how platforms contribute to making money.

Let’s start with the benefits of platforms, and see if we can stick some money to them.

What platforms get you

The benefits of platforms are explained in terms of either:

1. Developer productivity - which leads to improving how an organization can use software to run.

2. Operations productivity - removing the “toil” of day-to-day management and diagnosis of production, but also reducing the amount of time (and, thus, people) needed to manage to platform.

3. “Enterprise grade” capabilities - ensuring security, compliance, scalability - all the other “ilities.”

There’s a fourth category when a platform is a tool in an overall program: usually migrating from on-premises to public cloud or modernization applications. We’ll call this the “enabler.”

These are valuable things, but I’m frustrated with them because they don’t link directly business outcomes, things like: making more money (customer facing), performing a government service in a reasonable way (low cost and good citizen experience), or running a company better (internal operations).

That’s because platforms are just “enablers” of applications. And it’s the applications that directly create those benefits, that “make the money.”

Here’s three approaches I’ve come across recently that are representative of doing ROI, really, for any “enabling” technology.

VCR, IAR, and DTR

In the paper “Measuring the Value Of Your Internal Developer Platform Investments,” Sridhar Kotagiri and Ajay Chankramath (both from ThoughtWorks)2 propose three metrics and an overall way of thinking through platform ROI. This is the most thought-provoking, nuanced/complex/comprehensive, intellectually juicy, and thus all around useful ROI model of the three I’ll go over.

First, they have this excellent chart of linking platform capabilities to business outcomes:

A chart like this is great because it does its primary goal (showing how platform capabilities link up to business benefits) and also defines what a platform does. Here, the three things that directly give you ROI are CX (“customer experience,” I assume, which I’d call “good apps”), innovation (introducing new features, ways of working, ways of solving jobs to be done, and, thus, selling products and services), and cost efficiencies (spending less money).

Cost efficiencies is something you could do directly with a platform. It could cost you less in licensing and cloud fees, it could consume less underlying SaaS, it could require less people. The first two are fine and provable. The third is where ROI models get weird.

ROI by firing people versus doing more with the same

If you’re doing an ROI model based on people working more efficiently (“productivity”) the assumption you’re making is that you’re going to get rid of those people, reducing the amount of money you staff. But are you? Maybe long-term you’ll consolidate apps and platforms and then, a year or so out, layoff a bunch of people, realizing that benefit. If this is your goal, you’ll also need to contend with those future-fired employees reading the writing on the wall and saying “why would I tie my own noose?” and deploying enterprise-asymmetric psyops counter-measures.

Historically, the idea that automation is going to reduce staff costs has been dicey. You encounter Jevon’s Paradox: the cheaper it is to do something, the more of it people will do, often in excess.3

Thus, the more clever thing to do with productivity is to talk about how you can now do “more with the same.” You can give developers more time to work on more features, driving “innovation” and “CX.” Your operations people can now support more apps. Your cost of adding new stuff is cheaper. When you add ten more apps, you don’t need to add another operator or more developers because your existing staff now have more time available.

But, then you’re back to the problem of platform ROI: you’re talking about capabilities you get. And, until those capabilities are “realized,” you won’t know if your platform was useful. Also, there are so many things that could go wrong - or right! - that might be the more direct cause of success.

Nonetheless, I think the framing of “we never have enough time to do everything the business wants, right? If we had a platform, we would!” is pretty good. Instead of ROI, you’re directly addressing a problem, and a problem that’s stressful and probably keeps people up at night.

Three formulas for predicting and tracking ROI

The paper encourages the use of three formulas to track your platform’s value. You could use them to predict the platform’s ROI, but that would rely on you believing the input numbers you, uh, made up ahead of time.

• Value to Cost Ratio (VCR): VCR = (Projected Value / Projected Costs) * 100.

• Innovation Adoption Rate (IAR): IAR = ((Adoption in the current year - Adoption last year) / Adoption last year) * 100.

• Developer Toil Ratio (DTR): (Total Time on Toil / Total Time on Feature Development) * 100.

Here, you encounter one of the basic problems with any platform metrics: how do you collect those numbers?4

• VCR - this is what most people are after with “ROI.” However, how do you figure out those numbers? Proving the “Projected Value” of a platform is the whole problem!

• IAR - counting the apps on your platform versus all of the apps in your organization is achievable, more or less. People struggle with accurately tracking IT assets counting: most people don’t trust what’s in their CMDB, let alone have one, or, worse, even know what a CMDB is. But, I think most people can do some helpful app counting. This metric is tracking how much your platform is used. It assumes that usage is beneficial, though, which, for me, de-links it from ROI.

• DTR - this is the productivity metric and a good one. Collecting those two numbers, though, is tough. It’s probably best to stick with the “just ask the developers” method that DX encourages. That is, don’t waste your time trying to automate the collection of quantitive metrics, and instead survey developers to get their sentiment of “toil versus coding.” What I’d to this is that you should also consider the OTR: Operator Toil Ratio. How much time are you operations people spending on toil versus more valuable things. In the context of platform engineering, this would be product managing the platform: talking with developers and adding in new features and services that help them and other stakeholders.

I like this paper, and I think it creates a good model for even thinking about making the case for a platform and doing some portfolio management of platform engineering. Linking up platform functions all the way up to business outcomes (the big chart above) is great, and in many cases just using that big chart to explain the role platforms play in the business is probably very helpful when you’re talking with the Director of No. If that chart grabs their attention, the next conversation is talking about each of the boxes, what they do, and why doing it in a platform engineering way is better, more reliable, and “cheaper” in the “do more with the same” sense.

The Big Spreadsheet

The second model uses a large spreadsheet to track common developer activities, the cost of operations problems, and staff costs to show platform ROI. If you’re lucky, these are usually large spreadsheets with upwards of 50 numbers you need to input from salary, cost of hourly downtime, number of applications running on the platform, and the benefits of improving apps.

Once you “plug in” all these numbers, a chart with two lines intersecting usually shows up: one line is cost, and the other is benefit. At first, you’re operating in the red with the cost line way up there. Within a year or two, the lines streams cross, and you’re profitable.

Gartner has a pretty good one for platforms which, of course, I can’t share. Here’s another example from Bartek Antoniak:

One line I don’t see often are one-time-ish costs like the cost of migrating apps to the new platform and training people. Even cooler - but hard to quantify - would be the future cost of tech debt in the existing platform and app model.

Getting all of the input numbers is the problem, once again. How do you measure "increased speed of software delivery" and "mitigate security and compliance risk," or something like "optimize the use of infrastructure due to the self-service and developer portals"? How do trust those measurements and even the more straight forward ones like

There's a good trick here though: if it's difficult to gather those numbers, chances are you have no idea what the ROI of your current platform is (the "do nothing" option when it comes to introducing platform engineering). I suspect this is how most organizations are. The Director of No is saying platform engineering is a bad idea...but has no idea how to quantify how well, or poorly, the current "platform" is doing.5

Filling out the giant ROI spreadsheet will probably drive how you think of and decide on platform ROI.6 Tactically, this means that you want to be the first one to introduce a complex model like this if you're in a competition to get a platform in place. This could be if you're battling internal competition (some other group has an opposing platform and/or the option is to do nothing), or you're a vendor selling a platform.

Whoever introduces the ROI model first gets the define the ROI model.

Like canonical ROI calculations, these models are also showing you return over time, usually in three to five years terms. This can introduces an executive ownership problem. While the average tenure of CIOs is actually longer than most people joke about - four or five years, depending on the industry and geography - people move around on projects and within groups in IT.

A positive ROI model assumes you’ll see it through to the end without changing it. So, if the “owner” of the model has shifted and given ownership to someone else, you may not stick to the original plan. There’s also the chance that people will just forget what the point of the ROI model is and, more importantly, the plans that go with it. Pretty soon, you’re making new ROI models. A good test here is to see how quickly you can find the current ROI model (or “business case”) that you’re operating with.

Aggregate ROI

Instead of making a template for your ROI spreadsheet, you can aggregate the outcomes from several organizations. You still have The Big Spreadsheet in the previous example, but the point of the aggregate ROI is to show that the platform has worked in other organizations. The aggregate ROI is trying to convince you that the platform benefits are real and achievable.

Vendors like using these, of course, aggregating their customers. We put one of these out recently, done by ESG.

As ever, the problem with using this type of ROI is getting your input numbers. However, I think aggregate ROIs are good for both figuring out a model and figuring out a baseline for what to expect. Because it’s based on what other organizations have done, you have some “real world” numbers to start with. When vendors do it, these types of studies often contain quotes and testimonials from those customers as well.

You can hire Forrester Consulting to do their “Total Economic Impact” studies. Here’s a very detailed one from 2019 on Pivotal Cloud Foundry (now called Tanzu Platform for Cloud Foundry, or tPCF for short). Because they do these for multiple vendors, it’d be cool if they somehow aggregated all the aggregates. And I wonder if they use the same models for the same technologies?

You notice how I typed Forrester Consulting? That’s because it’s not “Forrester the industry analysts you’re thinking of.” Because you’re commissioning people to work on these TEIs (and other aggregate ROIs), it’s easy to carelessly dismiss them as paid for.

Sure, there’s certainly selection bias in these studies - you don’t hire them to analyze an aggregate of failures. But, these aggregate ROIs are still useful for proving that the platform works. That old TEI report interviewed four companies and based their model and report on them, same for the newer one. As with all the ROI examples, here, the aggregate ROI is also showing you an ROI model for platforms.

Us vendors have an obvious use for these PDFs: to show that our stuff is great! If you’re not one of us vendors, and you’re using these kinds of ROIs to get past the Director of No, I’d suggest looking at PDFs from rival companies and doing a sort of “aggregate of aggregates.” You’re looking to:

1. Prove the concept of platform engineering and the worth of platforms.

2. Show that it’s achievable at similar organizations - it’s not just something that Google or Spotify can do instead of “normals.”

3. Establish a baseline for results - we need to achieve results like these four other companies for it to make sense.

4. Create/steal a model - as with the last two ROI models, just having a model to start with is useful.

Beyond ROI

All of this started because someone asked me to help them put together a developer survey to show the value of platforms. A couple years ago I helped get the developer toil survey out. That survey doesn’t really address the value of platforms. You could use it to track ongoing improvement in your development organization, but attributing that to platforms, AI, or just better snacks in the corporate kitchen isn’t possible. I’d still like to know good survey questions that platform engineers would send out to application developers to gauge ongoing value.

Logoff

That’s enough for today! I’m already late for a call (tangentially on this topic!) so I didn’t even proof read the above. NEWSLYLETTERSSSSS!

Just links and wastebook today.

Relative to your interests

• Spring Boot 3.3 Boosts Performance, Security, and Observability - All these years later, Spring is still in wide use and still evolving.

• ‘You are a helpful mail assistant,’ and other Apple Intelligence instructions - Not that many, but interesting.

• Gartner Predicts 30% of Generative AI Projects Will Be Abandoned After Proof of Concept By End of 2025 - ”At least 30% of generative AI (GenAI) projects will be abandoned after proof of concept by the end of 2025, due to poor data quality, inadequate risk controls, escalating costs or unclear business value, according to Gartner, Inc. ” // Also a chart with rough estimates for initial and ongoing costs: input for an ROI model. // The other take here is that you need to start slow and small with enterprise AI: no one knows what will work, what good business cases are, what customers (and channels/suppliers/employees) will rebel against, etc.

• Nike: An Epic Saga of Value Destruction - The risks of shifting to a direct, Internet-based go-to-market. And, also, of focusing only on growing revenue from existing customers instead of also getting new customers: 'Obviously, the former CMO had decided to ignore “How Brands Grow” by Byron Sharp, Professor of Marketing Science, Director of the Ehrenberg-Bass Institute, University of South Australia. Otherwise, he would have known that: 1) if you focus on existing consumers, you won’t grow. Eventually, your business will shrink (as it is “surprisingly” happening right now). 2) Loyalty is not a growth driver. 3) Loyalty is a function of penetration. If you grow market penetration and market share, you grow loyalty (and usually revenues). 4) If you try to grow only loyalty (and LTV) of existing consumers (spending an enormous amount of money and time to get something that is very difficult and expensive to achieve), you don’t grow penetration and market share (and therefore revenues). As simple as that… ' // A little more commentary here.

• Where Facebook’s AI Slop Comes From - So cyberpunk! // ”the YouTuber who was scrolling through images of rotting old people with bugs on them.” // Related: ‘we don’t need the term “slop”. Consumers have decided that “AI” in its entirety is bullshit.’

• Teaching to the Test. Why It Security Audits Aren’t Making Stuff Safer - Bullshit Work in enterprise security. // Plus, why not start with basics before going advanced: ‘The world would be better off if organizations stopped wasting so much time and money on these vendor solutions and instead stuck to much more basic solutions. Perhaps if we could just start with “have we patched all the critical CVEs in our organization” and “did we remove the shared username and password from the cloud database with millions of call records”, then perhaps AFTER all the actual work is done we can have some fun and inject dangerous software into the most critical parts of our employees devices.’

• The Six Five: Advancing DevOps: Infrastructure as Code, Platform Engineering and Gen AI - “we see in our latest research that 24% of organizations are looking to release code on an hourly basis, yet only 8% are able to do so.”

• Why Is Demand Marketing An Obstacle To Its Own Success? - ‘Too many marketing subfunctions (demand/ABM, field marketing, customer marketing, digital, and events) create strategies unique to their function, independent from the others. Marketers often say that they have a “unified” plan, but it’s more like a PowerPoint deck with “chapters” for each team’s individual plan. This approach prevents marketers from orchestrating programs to reduce overlap and waste, and what’s worse, it has a direct, negative impact on buyers and customers.’

• The Prompt Warrior - Posts on prompts for all sorts of things.

• fabric/patterns - Whole bunch of prompts on a range of topics, even D&D!

• Why CSV is still king - We have a running joke on the podcast that every (enterprise) app needs CSV export. It’s only 10% a joke.

• Epic corporate jargon alternatives - Poetic alternatives to business bullshit jargon.

• 2025 Demand and ABM Budget Planning Guide: Do Better With Less - Enterprise software marketing budgets mostly flat, if not less: ”On the surface, it may appear as if most budgets are increasing, as 82% of global B2B marketing decision-makers report their budgets being increased by 1% or more. But once you adjust for inflation, it’s the same old story, as only 35% of organizations will see a real increase in budgets (with 31% of the 35% saying that the increases would be in the 5–10% range and 4% saying that their budgets would increase by 10% or more).”

The Dutch love Egg Salad and Potato Salad

This is in our small, neighborhood store. The larger ones have even more!

Wastebook

• “Kerfulle.”

• “gas station sushi” ROtL, 547.

Logoff

This is a really well put together presentation, with good content. It manages to introduce one, simple idea (and notice how she returns to/reminds you of it at the end), and yet not make it TED-talk level surface-level-simple. It gives you practical things to do if you want to work on “developer productivity.” And, it’s a perfect example of giving a vendor-pitch that doesn’t seem like a vendor pitch (one of the chief skills of a thought leader): customer cases with ROI, mention of the product being sold, and even screenshot-demos! It’s also re-usable and mine-able for EBCs, sales meeting, etc.

Where the Kubernetes spend is

Here’s a presentation recording going over our recent Kubernetes survey. I’ve mentioned it several times here, but there’s some new charts (like the above) and commentary in this talk I did with Rita and Danielle. Check out it for free in LinkedIn.

Wastebook

• 1 cup toilet chemical solution, three cups water. Add every three days or so.

• This year it’s gonna be beef rib buffet at SDT club.

• ”Seems like something that Netflix would try to turn into a series and then cancel it three seasons in.” Here.

• Its really hard to do “everything” of anything.

• “There are probably plenty of web-based and app based dice rollers already, but as they say, there are also plenty of love songs and people still continue to write them!” Here.

Relative to your interests

• The beauty of concrete - "ornament survives in the mass-market housebuilder market because the people buying new-build homes at this price point are less likely to be influenced by elite fashions than are the committees that commission government buildings or corporate headquarters. The explanation, in other words, is a matter of what people demand, not of what the industry is capable of supplying: ornament survives in the housing of the less affluent because they still want it.”

• A note on the EU AI Act - “The requirements look entirely reasonable considering these products are being positioned to become central to modern software – they’re effectively positioned to become the entirety of modern computing.” // And, this pairs well with this claim: “Sales workflows will fundamentally change. With AI, sales teams will no longer need to spend endless hours researching new leads or prepping for calls – AI will be able to do it in seconds. Reps won’t have to suss out the readiness of potential customers because AI will have automatically compiled a ranked list of primed buyers, and will keep it constantly updated. Need personalized marketing collateral for a deal? Your AI wingman will produce whatever assets you need and feed you live tips while you’re on a call to help you close.”

• Debugging Tech Journalism - Covering a handful of “tech companies,” focusing on the problems.

• The future of healthcare: Why enterprises must embrace AI innovation - Some use cases and applications for AI stuff in healthcare.

• From Burnout to Balance: AI-Enhanced Work Models for the Future - ”Nearly half (47%) of employees using AI say they have no idea how to achieve the productivity gains their employers expect”

• Words to Avoid and Replace - Some good Business Bullshit word translations and optimistic-ization.

• Why Return-to-Office Mandates Aren’t Worth the Risks - From Gartner! But, not much of a slam dunk either way as this promotional post doesn’t put out many numbers pro or con. “The benefits prove to be modest at best, and amid a rising well-being crisis, waning trust between employees and their employers, and a competitive talent market, it’s high time to ask whether the benefits of RTO mandates are worth the risks.” And: “High-performing employees report a 16% lower intent to stay in the face of on-site work requirements.” // There’s a lead-gen’ed PDF I’ll have to look at.

• Unpacking New NIST Guidance on Artificial intelligence

• The CrowdStrike Outage and Market-Driven Brittleness - ”Last week’s update wouldn’t have been a major failure if CrowdStrike had rolled out this change incrementally: first 1 percent of their users, then 10 percent, then everyone. But that’s much more expensive, because it requires a commitment of engineer time for monitoring, debugging, and iterating. And can take months to do correctly for complex and mission-critical software. An executive today will look at the market incentives and correctly conclude that it’s better for them to take the chance than to “waste” the time and money.”

• Don’t miss these VMware Tanzu sessions at Explore 2024 - What’s important to notice here are the number of companies relying on Cloud Foundry (the Tanzu Platform for Cloud Foundry, or tPCF as I think of it) to run their business. More so, that most have been doing so for years. Our PaaS just works. You don’t need to figure out how it works and build a platform out of it, it’s already a platform, a proven one. I’m looking forward to all the real world proof points, cases, and advice in these talks.

• Is Cloud Native a Vibe? Power Users Weigh In - “Cloud native is more than just a way of building apps; it’s a lifestyle and a state of mind. A company that has accepted this fact is working along agile principles with single teams having responsibility for the full lifecycle of their products, which means they own the product. Often, companies like this have flat hierarchies and value the individual knowledge of every employee to drive the products forward. Surely this is only possible with technology that supports these modern operating models, and it’s much more than just the technology.” -Jürgen Sußner, enterprise architect at DATEV

• Cloud Foundry in Action: Real Customers Stories from Cloud Foundry Day - ”We run a tremendous number of applications on top of Cloud Foundry. The ones that really impress me might surprise some people. It’s not just Black Friday sales but also the Amazon Prime events. When these events kick off, there’s an incredible surge in load across everything we care about, including credit card points processing, all handled by Cloud Foundry.” - Tom Brisco, JPMorgan & Chase Co.

• The employment effects of a guaranteed income - “1.3–1.4 hour per week reduction in labor hours.” // Yes, and: if we buy into the premise that a lot of work is Bullshit Work, this isn’t enough, we need more like six to 8 hours of wasted time to convert to living, not sitting in inefficient meetings.

Also, check in on the past two weeks of Software Defined Talk for tech news and commentary: episode 477 and episode 478. I’ll be back on it this week.

Conferences, Events, etc.

Talks I’m giving, places I’ll be, and other plans.

This year, SpringOne is free to attend and watch online. Check out Josh’s pitch for the event. There’s an on-site conference as well at Explore if you’re interested. But, for those who can’t, now you can watch all the fun!

SpringOne/VMware Explore US, August 26–29, 2024. DevOpsDays Antwerp, 15th anniversary, speaking, September 4th-5th. SREday London 2024, speaking, September 19th to 20th. VMware Explore Barcelona, speaking(?), Nov 4th to 7th.

Discounts. SREDay London (Sep 19th to 20th) when you 20% off with the code SRE20DAY. And, if you register for SpringOne/VMware Explore before June 11th, you’ll get $400 off.

Logoff

This summer I looked around at new jobs a bit - it’s good to do that once a year to see what’s going on, make sure you’re managing your career fully and actively, catch-up with people, and figure out if you like your current job. I ended up staying at my job.

During that, I talked with some analyst firms. This gave me some ideas for a long piece I wrote on how big industry analyst firms could change their business model. Well, more like add in more Internet and “the kids” friendly methods of publishing and business generation. There’s a lot of potential for the big firms to put out more of their content for free in new channels (like YouTube, newsletters, and podcasts), introduce some mid-tier pricing, and start addressing looming problems of their customer-base changing. There’s what looks like text-book case of innovator’s dilemma at play.

But, I need to go and make sure they don’t already do all this before I say they should do it. Hah-hah-hah-hah! The Free Advice Racket!

//

I’m back from a two week vacation in Finland with family and friends. We rented an RV drove from Helsinki all the way north to Alta, Norway. Finland is huge! We only drive 4 or 5 hours a day, so it took awhile to get up there. After living in the Netherlands for over six years, seeing all of the pure nature there is almost overwhelming: hills, mountains, fjords, trees! Anyhow, it’s great. I’ll post some photos on Instagram, here’s the first batch.

As far as I can remember, it’s the first vacation I didn’t want to end. I mean, vacation is great, but with when you’re with three kids, it’s more like a “trip,” and getting back to work can be relaxing.

This is just my, personal take on what Tanzu is, not an official statement. But I think it’s a pretty good one! :)

For a more in-depth look, check out Dekel’s recent videos.

Also, if you like the full on corporate sheen take, take a look at the Tanzu Platform Solution brief.

Wastebook

• “the ontological trick of discursive reductionism.” Here.

• Working title: “Working Hard at Nothing - Why Productivity Is Ruining Our Lives.” // I’ve done a while of chatting this week in a category I’m thinking of as “death by productivity.”

• Right now, it’s summer in Finland which is like winter in LA.

• “Reluctant Refusal: When a creature offers the leprechaun the chance to partake in merriment or revelry such as a song, a dance, or a good meal, the leprechaun must succeed on a DC 15 Wisdom saving throw or have the charmed condition for 24 hours. While charmed in this way, the leprechaun partakes of the offering, treats the creature as a trusted friend, and seeks to defend it from harm. The charmed condition ends if the creature or any of its allies damage the leprechaun, force the leprechaun to make a saving throw, or steal from the leprechaun.”

• Related: “Looking for suggestions for how to create a layer of grain in a silo. Just doing it bagged for now and I’m not satisfied.” // I showed this to my wife and she was shocked that I thought it was funny: “It would look better with a layer of grain.”

• “a much-remarked-upon toe ring” - it was a simpler time.

• “The employees murmur what but everyone seems to accept it.” (Translated to English from the Dutch.)

• “vibecession” // sounds like “uninformed decisions that things are worse than they are”?

• Unsettling picture.

• That five blind men and the elephant parable is, like, 1,500+ years old!

• Consumer tech companies drink their own champagne. Enterprise tech companies make dog food for owners to buy to feed their dogs. They do not eat the dog food, and rarely own dogs.

• “The Lock-in Boogeyman.”

• ”Join the Czech and Slovak mechanical keyboard community in celebrating our hobby and come check out just how deep the rabbit hole goes.” From the Mechanical keyboard meetup.

• “no-collar American food.” He didn’t like it.

• J: “I didn’t have to explain. I could have just spilled something on their tie and ran off.” M: “That’s what they call ‘the podcaster’s exit’!” - RotL #544

• L’esprit de l’escalier: “the predicament of thinking of the perfect reply too late”

This week’s podcast: Bring a point of view

On this week’s Software Defined Talk, Brandon and I talk a lot of wasted time at work, so called “bullshit work” and how management might could fix it: “This week, we discuss Google possibly buying Wiz, why "meta work" leads to too many meetings, and why it took forty years to get spell check in Notepad. Plus, we share some thoughts on enjoying your vacation.” This was a long episode (an hour and 40 minutes). We discussed writing and using AI to write in the after show. You can also watch the unedited video of the recording.

Relative to your interests

• White-Collar Work Is Just Meetings Now - “Gloria Mark of UC Irvine has found that workers require an average of 25 minutes to return to their original task after an interruption. By this measure, a 30-minute meeting is, for the typical worker, best thought of as a one-hour detour.” // For 25+ years, a significant part of developer productivity has been the simple of idea of “stop interrupting me.”

• The Silicon Valley Would-Be Vice President - Those capital gains taxes always stick in folk’s craw.

• Next Gen Application Delivery: Getting Started with Intelligent Apps Powered by AI - “Our numbers show that 65% of all enterprise developers use Java, making it essential for business-critical applications. Development frameworks, like Spring AI, enable Java developers to interact with AI models and vector databases through the framework, rather than having to learn new skills.”

• 15 hours a week - ‘And yet, for most of that time, I’ve continued to find that exertion harder to exercise than most, while my flair and ability have gained me attention and love, I’ve seen less of whatever today’s equivalent is of “a satisfactory grade in the public examinations”’

• VMware’s ‘Private Cloud’ Solution Emerges Under Broadcom - “Previously, five different business entities were responsible for delivering our product. We have now consolidated these sectors into one organization with a unified product team, global support team and a single management direction. This alignment ensures a critical focus on product development and seamless delivery.”

Conferences, Events, etc.

Talks I’m giving, places I’ll be, and other plans.

This year, SpringOne is free to attend and watch online. There’s an on-site conference as well at Explore if you’re interested. But, for those who can’t, now you can watch all the fun!

Our analysis of the State of Cloud Native Platforms 2024 survey, online, speaking, July 24th, 2024. SpringOne/VMware Explore US, August 26–29, 2024. DevOpsDays Antwerp, 15th anniversary, speaking, September 4th-5th. SREday London 2024, speaking, September 19th to 20th. VMware Explore Barcelona, speaking(?), Nov 4th to 7th.

Discounts. SREDay London (Sep 19th to 20th) when you 20% off with the code SRE20DAY. And, if you register for SpringOne/VMware Explore before June 11th, you’ll get $400 off.

Logoff

Don’t forget to check out the talk I’ll be part of next week, July 24th on our recent Kubernetes/cloud native platform survey. There’s a few charts in there that don’t show up in the actual report or in my blog post on it, so unique charts for you!

Register to watch it for free here, or in LinkedIn. Also in YouTube, if you prefer that.

//

I’m off to vacation for a few weeks. Finland!

When I get back - which seems like forever from now - I feel like it will be a new year

Survey says…

Kubernetes getting out of appdev improvement slump

This is the chart I look forward to in our annual State of Kubernetes report1:

It’s been a rocky few years as Kubernetes has gone mainstream. I pay attention to the “shortened software development cycles,” which you can see started going down. It’s been going up for the past two years, so that’s good. As more “normals” start using Kubernetes, the tolerance the early adopters have erodes. That’s my theory at least.

You can check out my analysis of the survey in this week’s blog post of mine. And, we have a webinar coming up next week looking at the survey as a whole.

You should watch this talk I’ll be doing in a few weeks, FREE in the comfort of your own home/RTO-job!

Register to watch it for free here, or in LinkedIn. Also in YouTube, if you prefer that.

Relative to your interests

Mostly an enterprise AI edition.

• CIOs resist vendor-led AI hype, seeking out transparency - There’s a lot of AI de-hyping now. First, you have the “they’re stealing our IP” stuff. Second, you have the “no one has come up with (enterprise) apps for it yet” sentiment. Thankfully, there’s no “AI will kill us” vibing.

• GenAI or Die - Big time pro AI for ERP here, with notes of “what’s ERP done for me lately.”

• The “Little Tech Agenda” is Just Self-Serving Nonsense - FTC: “Your therapy bots aren’t licensed psychologists, your AI girlfriends are neither girls nor friends, your griefbots have no soul, and your AI copilots are not gods. We’ve warned companies about making false or unsubstantiated claims about AI or algorithms. And we’ve followed up with actions”

• Measuring the impact of Developer Relations on Revenue - Figure out how to gather leads, and figure out how to get attribution in the sales pipeline. The second is difficult, especially if your company is already bad at it. But, it’s important to figure out.

• Are platforms pointless? - ‘So much of “platform engineering” treats the application process itself as the main event. Sure, great, you make it easy for me to run hundreds of Nginx’s with whatever-the-fuck behind them, and restart and blue-green deploy and autoscale. Great. That’s not my performance bottleneck.’ // He says: (1) It’s just a way to avoid Terraform, and, (2) database management is a more important problem.

Goals of a platform in pictures

This paper by Torsten Volk has a good diagram of the point of platforms, that is, “the outcomes,” the benefits.

Also, the hassles of building you own platform:

Conferences, Events, etc.

Talks I’m giving, places I’ll be, and other plans.

This year, SpringOne is free to attend and watch online. There’s an on-site conference as well at Explore if you’re interested. But, for those who can’t, now you can watch all the fun!

Our analysis of the State of Cloud Native Platforms 2024 survey, online, speaking, July 24th, 2024. SpringOne/VMware Explore US, August 26–29, 2024. DevOpsDays Antwerp, 15th anniversary, speaking, September 4th-5th. SREday London 2024, speaking, September 19th to 20th. VMware Explore Barcelona, speaking(?), Nov 4th to 7th.

Discounts. SREDay London (Sep 19th to 20th) when you 20% off with the code SRE20DAY. And, if you register for SpringOne/VMware Explore before June 11th, you’ll get $400 off.

Logoff

I am continuing to enjoy “Hi-fi relaxation.”

//

Twitter has been slow here in Amsterdam. And I have fiber! Is this some kind of petty revenge, or just standard fail whale problems?

…right?

I’ve been looking around for estimates on how many custom written apps run on private vs. public cloud. There’s a lot of coverage and estimates of people using multiple clouds, but finding breakouts is tough. IT IS VERY HARD TO FIND!

Here’s what I’ve found recently:

• "According to Forrester’s Infrastructure Cloud Survey in 2023, 79% of roughly 1,300 enterprise cloud decision-makers surveyed said their firms are implementing internal private clouds.” Here. // This doesn’t answer my question, but is useful.

• Spend is a bad proxy for workload placements, but: "IDC forecasts that global spending on private, dedicated cloud services — which includes hosted private cloud and dedicated cloud infrastructure as a service — will hit $20.4 billion in 2024, and more than double by 2027. Global spending on enterprise private cloud infrastructure, including hardware, software, and support services, will be $51.8 billion in 2024 and grow to $66.4 billion in 2027, according to IDC." Ibid. // Spend isn’t a great proxy for actual usage, but there’s that.

• IDC Cloud Pulse from last year: it’s something like 40% to 50% public cloud, but this also includes SaaS, which is not exactly what I’m interested in. (See the chart.)

  The IDC numbers are pretty good. I’d want to redo them and throw out COTS apps and SaaS, but good enough.

  So, what’s the split between public and private cloud? I don’t know: 50/50? But, again, this doesn’t track organization’s custom written apps. I could see that it’d go more in either direction.

  Furthermore, if you went off what the Goldman Sachs CIO surveys imply (mentioned last episode), it’d be more like 70% private cloud, 30% public cloud.

  I think I’ll start going with mildly uncertain 50/50 with a percent or so going to public cloud each year.

  Still, if I were to say “half of the enterprise IT world is largely ignored by the chattering class,” you’d hopefully think “well, that’s weird.”

  

  Software Defined Talk #475

  This week, we discuss Mary Meeker's AI & Universities report, the CD Foundation's State of CI/CD Report [see below], and share a few thoughts on DevRel. Plus, Coté gets fiber and is forced to watch soccer.

  Listen to it now! (You can also watch the unedited video recording.)

  Wildly Different Version Control Usage Results in Developers Surveys

  Speaking of estimates and surveys, a tale of being careful with surveys:

  • Slashdata’s survey reports that 30% have used “source code management” in the last 12 months. This means that 70% of people haven’t checked inter code for a year or more, or at all? There is more nuance to it than that, but that’s what’s implied.

  • The 2023 JetBrains survey reports that 76% of people “regularly” use a “source code collaboration tool.” This means that 24% of people don’t “regularly” check in code?

  • The Stackoverflow 2022 survey says that 95.69% of people use version control (it doesn’t say the frequency of interaction). This means that 4.31% people do not use version control. They didn’t track this in the 2023 survey.

  ¯\_(ツ)_/¯

  ---

  You should watch this talk I’ll be doing in a few weeks, FREE in the comfort of your own home/RTO-job!

  

  Register to watch it for free here, or in LinkedIn. Also in YouTube, if you prefer that.

  ---

  Wastebook

  • “Clicks to Bricks.”

  • “IDC Links & IDC Blinks.”

  • “Beloved Austin local Leslie Cochran.” Here.

  • I used to listen to The Lounge Show every Saturday morning. It’s still there! Also, archives here and here.

  • “An enquiy, based on the author’s intimate diary, into the conditions for obtaining happiness and person start of values.” Here, for this.

  • If not better, at least the same. The enterprise software buyer’s lament.

  • “the riffiest of the raff.” Here.

  • I’m usually not “chill out and watch video of people just doing random shit” guy, but I’ve really been liking MrT’s breakfast service marathons. He makes an English breakfast burrito, which I do not agree with, but I’m not here to yuk your mums.

  Relative to your interests

  • Understanding the Rise of Platform Engineering and Its Relationship with DevOps - Printer-friendly - US50199923 - Platform engineering definition from IDC: ”the discipline of designing, building, and maintaining a platform of curated tools, services, and knowledge, called an IDP, that enables development teams' self-service access to the resources needed to build, test, and operate digital solutions. Platform engineering aims to optimize software delivery by removing friction from the developer experience by offering blueprinted, supported approaches to building and deploying software. The platform team, made up of platform engineers, is responsible for building and maintaining the IDP.” // A key point is self-service, you know, less tickets. // This seems like a lot for one team to take on.

  • Does Social Media Cause Anything? - It’s difficult to collect data about social media’s effects (good or bad). // “the ever-present spiderweb of the social graph, the network of accounts, RTs and likes that lets me understand not only what someone thinks but what everyone else thinks about them thinking that.”

  • The Product Model in Traditional IT - ”Outcomes vs Predictability” is good framing for switching from traditional IT to “digital transformation.”

  Conferences, Events, etc.

  Talks I’m giving, places I’ll be, and other plans.

  Our analysis of the State of Cloud Native Platforms 2024 survey, online, speaking, July 24th, 2024. SpringOne/VMware Explore US, August 26–29, 2024. DevOpsDays Antwerp, 15th anniversary, speaking, September 4th-5th. SREday London 2024, speaking, September 19th to 20th. VMware Explore Barcelona, speaking(?), Nov 4th to 7th.

  Discounts. SREDay London (Sep 19th to 20th) when you 20% off with the code SRE20DAY. And, if you register for SpringOne/VMware Explore before June 11th, you’ll get $400 off.

  

  Logoff

  I’ve been thinking about an addition to my Bullshit Business Dictionary entry for “executize.” Maybe something like “the pre-read.” In theory, you put together a memo, document, maybe slides, that you send to an executive ahead of a meeting or for planning. You’ll put a lot of work into this, often with an executized summary at the front (bullet points), and then many pages of longer notes, research, etc. Or, you know, a “slide-bank” after the closing slide.

  In my ~30 years of experience, the pre-read is actually read only 30% to 50% of the time. There are many executives who will never read it. They want to a sort of “have the meeting in the meeting.” It’s “sort of,” because if you’re doing that you will have read the pre-read so that you can discuss your reaction to it, ask questions, and focus on making a decision.1

  You may think this means you don’t need to do a pre-read: who knows what the executive wants, what they’ll ask for, what will be in their head at the moment. Why waste time on things that never get used. However, I think do an extensive pre-read is important so that you know what to say and suggest during the meeting, at the very least so that you have context and can form opinions.

  Also, there’s a chance that your pre-read will be converted to a “post-read” if the executive ends up being interested in the topic.

  All that said, if you’re operating an unread pre-read environment, what’s more important is to be spontaneous and use improve tricks to kick around ideas - the old “yes, and” thing. There’s a view that working for an executive means you’re helping them solve the problems they have, no steering them towards the problems they should be focusing on and the solutions you think are right. I think that’s mostly right; it’s a hard thing for nerds to reconcile.

  In the “my job is to augment the executive, not help the corporate achieve outcomes/etc.” mode of operating, you might want to save your energy and time for the post-meeting work, and just do a small amount of pre-read work. Indeed, if you keep things unclear/high-level, you can likely achieve that “executize” level of bullet points right away.

  1

  There’s other executives who will read the pre-read and/or expect a very direct, structured in-meeting “read out.” These executives usually follow the American-style of just wanting to know the conclusions, the exact actions to take next. “Application-first reasoning,” they calls it. They may or may not care why, and will instead use intuition (or trust in the process) to know that something good will happen as a result of taking actions. (The opposite of this is “principles first,” where you build up a case right-side up pyramid style.) Anyhow: figure out your executives style, there are many types.

I don’t know. I’ve been trying to sort out what platform engineering is for awhile.1 It matters a lot for my job! While I haven’t verified it, it seems like it started as a marketing campaign from Humanitec and then took on a life of its own. Now the likes of Gartner have practice areas for it and are hiring analysts to cover it.

This means that people in enterprises are trying to sort out what to do about platform engineering. Surely they need it! The category is now loaded with everything, pulling in the all the stuff, including internal developer portals, CI/CD (I don’t think people [read: vendors] explicitly discuss foundational practices like build automation anymore, but I suspect actually putting CI/CD in place is the primary driver of the benefits people get with platform projects), getting Kubernetes to work, and all the usual cloud native platform stuff.

That is, what platform engineering is has become too expansive and is, ironically, driving too much cognitive load. Here’s my simplification:

And if you don’t have 2 minutes and 36 seconds to spare, here’s a shorter version.

Wastebook

• “the AI-infused journey” Gartner.

• Teamperature.

• “Audience Acquisition Rep II”

Relative to your interests

• The AI summer - Several charts of technology adoption, including the Goldman chart that shows CIO’s intentions to move workloads to public cloud is always high, and not well executed. // Un-clickable citations, though.

• Gartner Survey Finds 64% of Customers Would Prefer That Companies Didn’t Use AI For Customer Service - “Many customers fear that GenAI will simply become another obstacle between them and an agent. The onus is on service and support leaders to show customers that AI can streamline the service experience.” // I mean, that’s the point right: otherwise “productivity” wouldn’t improve. The hope is that the AI things are better at solving problems. The problem is that you usually need a human to actually change things, make things happen, and deal with exceptions. Otherwise, you get stuck on an accountability sink.

• DevRel’s Death as Zero Interest Rate Phenomenon - A list of how to show marketing value. // This whole time all the devrel people just needed to integrate into the finally tuned, perfectly functioning, incredibly accurate, much beloved, and simply existent customer journey management CRMs out there. Also, they should have been listening to all the feedback the sales people gave them about how their activities helped close deals and clamp down churn. And if they had just engaged with the product people who were eager to work with them! Instead, just think of all the money that was wasted on teal-haired people’s sticker collection? // But, yeah. Yes, and: You can definitely always focus on selling the product more.

• Jevons paradox - When you automate something very valuable (or just “costly”), people demand more, and more complex product. This pulls in more need for labor that can do the more complex work. Hopefully.

Logoff

This is a bit of a weird recording since there’s no slides (“technical difficulties”), but if you want to see the second version of my “Why We Fear Change” talk, here it is, from SCaLE 21x. The first time I gave it (at cfgmgmtcamp) I ran out of time because I’d packed some of my Business Bullshit words in as interludes for fun.

//

I’ve been hunting down “private cloud” usage numbers - just anything, at this point. Specifically, I want to know which in-house apps run where. I don’t care about enterprise applications like ERP systems, nor SaaS apps: just the applications that organizations code and run themselves.

There’s not that many out there in the easy to find, free surveys. Essentially, what you see is that something like 70% to 80% of people use multiple clouds - various public ones, on-premises, “private cloud,” etc.

There’s this from Slides Benedict:

If you read this chart, you’d say “something like 25% to 30% of ‘enterprise workloads’ are running in public cloud. This, 70% to 75% of apps are running on-premises.” Does that seem right? If it is, the lack of conversation around on-premises is bizarre. That’s the majority of IT and movement away from it is very slow.

But, it’s hard to have confidence in such a contrary statement because I can’t find the surveys cited to make sure (a) I focus just on apps, not SaaS, etc., and , (b) I’m reading it right, the geographies and industries/demographics (is it 100 F500 CIOs, or just rando’s who answered a survey online?), etc.

In large organizations, this isn’t too much of an insight: they’re so long, so long lived, have so many geographic groups that have their own IT stack, not to mention both centralized/planned IT and YOLO line of business IT, and have acquired so many companies…that of course they have everything. What I’m more interested in are how many apps are in the public2 cloud versus not.

I just got a pile of recommendations from people, so perhaps I’ll have more to report back.

This is a good talk listing some tactics for changing how organizations work. It’s targeted at executives. What’s bonkers is that, as the screenshot below, it has just 27 views.

In the talk, we have a former group CIO at American Airlines. She’s telling you her tips for how to introduce change in a large organization. As Gene “The Phoenix Project” Kim says in his intro, American Airlines is the biggest airline in the world. This is also from the most famous DevOps/IT conference for executives that exists now (probably).

What’s bonkers is that it has only 27 views. And that one like is from me!

Executives and management at enterprises are always asking how to make change easier, and here’s a whole bunch a tips from their peers. And most of them cost nothing, or in the scheme of an American Airlines budget, nothing (like buying some pizza, rubber duckies, stickers, etc.). Also, she tells jokes and has some gags!

Now, maybe those 27 views are from CIOs of Global 2,000 companies who watched the video 100% of the way through. But, come on…

Are these great tips that will instantly solve all your problems, that fit to any situation. Hardly! But, with that few views, no one was even curious enough to check it out. More to the point: probably no executive has come across this video.1

My point here is that getting enterprise IT improvement content like this out is so, so hard as to be nearly impossible. Of course, you have to match your mediums to the type of content and channels you have.2 But, in general: do videos work? How about 60 second videos? Do white papers work? Blog posts? How about a webinar? How about a webinar with an analyst from Gartner, Forrester, or IDC? LinkedIn posts? License a PDF from an analyst shop? Pay a small analyst firm to write a white paper? Maybe emails?! What if I email a short video?

Some of those work sometimes, but I have the feeling that the main thing that works to reach an “executive audience” is just meetings. Which is to say, being face-to-face in a conference room or just having a meal together. And, yes, as in this instance, in-person at a conference.

The enterprise software sales process is incredibly frustrating and inefficient. But, when the hundreds of videos like that are ignored each year, you can see that being in-person is perhaps the most “efficient” way to do the job of marketing and selling because it’s the only way.3

Also, you should watch all my most excellent videos with such tips, not only 100% of the way through, but three to five times each to make sure they really, like, sink into your brain.

Anyhow. It’s a good presentation, you should check it out.

Wastebook

• The catsup paradox. Everyone loves catsup on fries, on burgers, hot dogs, etc., even meatloaf. But, on its own, most people find catsup disgusting, it stinks. For example: after a meal, no one ever wants to be the one who cleans the ketchup ramekin. (Mustard and mayonnaise don’t invoke that kind of reaction, let alone pickles.)

• “I would love to be a, sort of, Mr. Rogers for adults.” Noah.

• “I feel like the world is burning, and I’m wearing polyester.” Ibid., but Peter from Germany.

• “I have no idea who that is!” said ever closet Gallagher fan ever. // He probably could never tour Europe because the watermelons were too small. “That just won’t be funny,” he said, taking off his hat and rubbing his head, “It’ll be like I’m smashing a baby’s head. Just cancel the shows, Garry. Just do it.” // Months later, Garry sits in front of Pauly Shore: “now, I like what you got, kid. Your mom says great things. Say hi to her, by the way. You don’t smash any large fruit in your act right? No? Great. Have you ever thought of the Europe circuit?”

• ”My gf, upon finding out there is not only one but in fact TWO sequels to Taken: “Somebody needs to put an AirTag on her”” kwon.

• “Premium Economy – access a lounge where they provide materials to make your own ham sandwiches.” Here.

• “He was scanning the horizon.”

Relative to your interests

• AI transforming banking, reports UBS executive - ”The Swiss bank has integrated AI into its services, launching a pilot last year for instant credit tailored to small and mid-size companies requiring liquidity. This service enables bypassing credit officers to expedite the process for this standardised product.”

• Japan’s digital minister claims victory against floppy disks - ”Japan’s digital minister, Taro Kono, confirmed that the Japanese government has finally rid itself of floppy disks.”

• The World Is Eating Software - The software bubble. Also commentary on power steering.

• Why Supercloud Architectures Could Upend Cloud Computing – Or Not - ”IDC’s March 2024 Cloud Pulse Survey (n = 1,350) shows that 74% of cloud buyers have multicloud strategies. It’s no longer a big deal to use multiple clouds.”

• Join the public beta for GenAI on Tanzu Platform today! - Get your own private AI in a box sandbox going on.

• Accountability Sinks - ’“accountability sink”: a situation in which a human system delegates decision-making to a rule book rather than an identifiable individual. If something goes wrong, no one is held to account.’ // That’s a good phrase to point out something people seem blind to. Humans set policies, humans can decide to un-set policies. // It’s a version of “it’s just business,” which people who have/use that sentiment forget is a catch-phrase for mafia movies when they kill people.

• Frontline workers split on AI’s workplace impact - ”Nearly 3 in 5 workers using generative AI say the technology has saved them around 5 hours of work per week.”

• How To Measure Platform Engineering - Suggested metrics: “Market share. Onboarding time. Net Promoter Score (NPS). Key customer metrics.”

• The Blue-State Wealth Exodus Continues - State income tax flight: “The IRS last week published its annual data on the migration of taxpayers and adjusted gross income (AGI) between states. California ranked, again, as the biggest income loser ($23.8 billion) in 2022, followed by New York ($14.2 billion), Illinois ($9.8 billion), New Jersey ($5.3 billion) and Massachusetts ($3.9 billion). The top gainers were Florida ($36 billion), Texas ($10.1 billion), South Carolina ($4.8 billion), Tennessee ($4.7 billion) and North Carolina ($4.6 billion).” // It’d be buck-wild if far right cultures in southern states changes because progressive people migrated due to taxes.

• DevOps Isn’t Dead, but It’s Not in Great Health Either - Round-up of surveys showing that the benefits of DevOps are flattening/slowing. Perhaps it’s not dead, it’s just had the final victory. That is: it’s much better than 15, even 10 years ago.

• How Much Revenue Must a Company Generate to IPO? - “Before 2018, only one company IPOed with more than $200m in revenue. In fact, the median revenue at IPO at $90m. Today, the median revenue at IPO is $189m (corrected for inflation), more than double.”

Logoff

A few items in my “topic ideas” queue:

• I’ve had a lot4 of conversations recently about the fallacy of collapsing silos in large enterprises.

• This is different than the “DevOps is Dead” meme that’s going on around, which I’m slowly building up some commentary around (I think my take is: what if it’s not “dead,” but just wildly successful? AKA: do you not remember how terrible things were in the 2000s? I don’t really like that take too much - way too counter-counter-take - hence still rolling around in it.) Recent DeathOps pieces: the Eulogy one, and some survey digging. Brandon and Matt discussed them on last week’s Software Defined Talk.

• If we were to look at the original, NIST PaaS definition, is it different than the cloud native platform definition? Is it some kind of really nerdy trolling to start using the phrase PaaS again?

Using Surveys for Marketing

Wastebook

• They dress like they’re Chaotic Good, but they’re totally Lawful Good under all that sloppy couture.

• This is some buck-wild ABM: a direct appeal to Delta to buy their AI stuff.

• “I wrote this one when my family was in Hawaii, too. I kind of had a meltdown there, to be honest.” Blue Cheese.

• If you were to watch Furosia and Fury Road back-to-back, as one movie, it might just be the best action movie ever made. You’d need two intermissions, though, not between the two movies, but right before each major action sequence.

• Why Not Give Robots Foot-Eyes? Here.

• “How are you today?” “Legally prohibited from complaining due to a non-disparagement clause, you?” Via.

• “The problem is the darker side of the word, when we use it to express ‘merely.’” Here.

• “nothing fun happens on a Monday night.” Here.

• “Unsavory”

• “How does (or will) AI impact this decision?” // When AI reaches the office or the CFO. // If your CFO asked this, how would you even answer it? Would a CFO know how to interpret, verify, and trust the answer? // Maybe the right answer is “who knows. Have you actually used those things? How about we put ‘fuck around and find out’ on the ledger for this quarter?”

• There’s no fucking around, just finding out.

• “Why Not Give Robots Foot-Eyes?” Here.

• “It just wasn’t their world anymore. So they had to leave.” Hilda, s1e2.

• “There’s that well observed effect where you have better conversations if you’re not looking each other in the eye. Driving. Walking. Walking meetings. Watching telly on the sofa.” Russell Davies.

• And: “they still have a selection of newspapers you can read. That means something, they’re not going to hurry you out if they’ve offering you a newspaper.” Ibid.

• “How are you today?” “Legally prohibited from complaining due to a non-disparagement clause, you?” Via.

• “The problem is the darker side of the word, when we use it to express ‘merely.’” Here.

• “A turd for your argument” Here.

Relative to your interests

• Platform Engineering: The First Hype Cycle - I should probably read this.

• Gartner Survey Shows 66% of Finance Leaders Think Generative AI Will Have Most Immediate Impact on Explaining Forecast and Budget Variances - What’s also remarkable is how low they rate other use cases. I’m a bit confounded about reports that consulting firms are getting billions in sales for AI stuff. What could it possibly be?

• A Eulogy for DevOps - “Feels like somehow Developers got stuck with a lot more work and Operation teams now need to learn 600 technologies a week. Surprisingly tech executives didn’t get any additional work with this system. I’m sure the next reorg they’ll chip in more.” // New idea for conference talk: The Nine Lives of DevOps.

• Kubernetes Security Report: Evolving Landscape of DevSecOps - Red Hat Kubernetes surveys results, here, looking at the security part. The theory is: there’s too many different parties working on security, causing “confusion and delay.” See also our recent State of Cloud Native Platforms survey (5th year in a row): security is always on the top. I guess that problem will never get solved.

• YouTube dominates streaming, forces media companies to adapt - “YouTube made up 9.7% of all viewership on connected and traditional TVs in the U.S. in May — the largest share of TV for a streaming platform ever reported by Nielsen’s monthly ‘The Gauge’ report. Netflix ranked second, claiming 7.6% of viewership. Among streamers only, YouTube’s total viewership was close to 25% market share.”

• Don’t be results-oriented - Make sure your factory is functioning well.

• Enterprise hits and misses - gen AI spending meets ROI pressure, AI readiness is a thing, and AR/VR has a vision problem - “Enterprise hypothesis: AI can’t fix what ails you (e.g. legacy landscape and ‘messy’ data bottlenecks), but applied properly, it can accelerate what you do well.” // And, as said elsewhere: “So you aren’t just adopting AI, you’re doing data management as well.”

Logoff

Someone finally figured out why procrastination is therapeutic:

New idea for a conference talk: “How platform engineering stole all your sponsorship money.” It’d kill as a lighting/ignite/whatever talk on the DevOpsDays circuit.

It’s July!

Consumer-grade Chat Gen-AI Churn Update

Overall, the generative AI things have been a disappointment for me. For writing, they seem helpful for generating and starting content, but it ends up being more work to fix and, then, rewrite what they do. Over the years, people have ghostwritten things for me. Like, humans! I’d too often have to (want to) go through and inject my voice and style, and also add play around with the core ideas. Now, when “byline” opportunities like this come up, I write just write it all myself.

Using AI for writing is like that. It’s sort of good for bouncing off ideas.

And summarizing text is hit or miss. First, it can’t always retrieve web pages. I mean, I get it with people not wanting to give their content to Evil Nerds and stuff. But the result is that I don’t get what I want: a faster way to read text.1 The summaries are “fine,” but they usually tell you what the article is about instead of making the actual contents easier to read.

Should I cancel my subscriptions? I haven’t been able to get my mind around canceling ChatGPT, but I did cancel Gemini.

Gemini just never felt like it was doing more than a quarter-assed job. This was the final deciding factor: I asked it to tell me what my oldest email in GMail was and who wrote it, and it claimed it couldn’t find that, and did something like just telling me the last ten emails in my GMail inbox. I mean, come on: what the actual fuck? And this is when it actually looked at GMail. Often Gemini would tell me it didn’t even have access.

And, I’ve more or less given up on using it to play D&D. It just can’t be creative enough. This one is more complicated to evaluate. If you put a lot of work into it, it sort of works - hence my early enthusiasm. But, it’s like some of the other use cases: it’s only as good as what you bring to the table. Both your time in managing it (crafting prompts, giving it the right text to start with) and coaching it. Instead, you could just all of that yourself.

Now, having it generate code for me works well. But it’s not like I’m writing big enterprise apps. I’m just making one off scripts.

My theory is that it’s just better search, slightly better summarizing. For example, you can upload the endless lore about The Forgotten Realms, and it sort of does good at answering questions for you. It’s also good at explaining things, being a tutor. I had a long session to understand net-present value and discounted cash flows, and I feel like I understand that better. Having it explain why the Fed interest rate is so important would be interesting.

With all of the billions(!) being spent on consultancy firms, I’m not sure…why. CEOs and other executives who’re so thrilled by AI should try actually using it for week, even in just these chat interfaces. If Google can’t search it’s own services like Gmail, good luck to the rest of you.

Here is a picture of my daughter enjoying delicious Dutch fries at the zoo

Wastebook

• This meeting could have been an AI talking to itself. At worst, two AIs talking to each other.

• “Life in general being more software-mediated.” Sharp Tech, June 20th, 2024. // This is a good framing, and simplification of what “digital transform” aspires to describe.

• This meeting could have been an AIs talking to itself.

• AI: have the bullshit master do bullshit work. Yes, and: if you can have AI generate it, it’s probably not worth doing.

• Socialism applied to developer productivity metrics.

• “…the one that you appreciated, which: I appreciate your appreciation.”Sharp Tech, June 20th, 2024s

• “There is, what we say, ‘the escape of the brains,’ in Italy.” Sara.

• “We’ll have extra slides to explain the bingo to you.” DevOpsDays Amsterdam evening activities.

• “Succor borne every minute.” The FTC.

• ‘Perhaps they have a regex that replaces “Willie Nelson” with “Lavar Burton”?’ Here.

• “the Let’s Only Drop Half A Bomb On Belgium Party” NL politics.

• Mixed metaphor du jour: festering tech debt.

• New number to track: how much of your IT budget is spent on “tech debt” (or “legacy”)? // “CIOs estimated that tech debt amounts to 20 to 40 percent of the value of their entire technology estate before depreciation.” McKinsey in October, 2020.

• And: “Some 30 percent of CIOs we surveyed believe that more than 20 percent of their technical budget ostensibly dedicated to new products is diverted to resolving issues related to tech debt.” Ibid.

• “lift-tinker-and-shift.” Stephen Orban, AWS, 2016.

• Do you self identify with these sales plays? If so, dial 1–800-CUSTMRJRNY right now. Call now!

• “Fake Buddha Quote.” Here.

• “The median image could be described as ‘a poster you’d expect to see on the wall of a teenage boy in a movie scene where the writers are reaching for the standard stock props to show that the character is a horny teenage boy who has poor social skills.’” On AI images.

• “Next: cranium rats.”

• Maybe there’s a rubric that: if it takes more than a day to be delivered and to your house, it’s actually a luxury good.

• “He always carried a little grater and a nutmeg to flavor the glasses of port he drank.” On Henri de Toulouse-Lautrec.

• “In fact, XML is possibly the only announcement in the development world to rival the impact of the Java platform. It is fortunate for us as developers that these are complementary technologies rather than competing ones.” June, 2000.

• AI is the enemy of done.

• Any person at a vendor who’s actually worked on the submission to a Wave or Magic Quadrant will tell you: if we COULD have paid them instead of doing all this work, we WOULd gladly have paid them, even out of our own wallets.

• AI Will Not Fix Your Broken Culture (and Other Hard Truths).

• “Reduced Snowflake expenses by 50% by clicking a button that I am pretty sure that 90% of companies also need to click. A$500,000 savings, approximately.” Here. A resume is as much about filtering out potential employers as it is the recruiting pipeline.

• “geomyth” Here.

• They really wear their strategy on their sleeve.

• “The second thing I do is decide whether the country is worth wasting a meal on breakfast. I might just skip it.” Tyler.

• And: I hate to be the grumpy old man but one day, not too long from now, I will be telling my children, “I remember when you didn’t need a visa to travel abroad. I also remember when there were no anti-money laundering and know your customer laws!” The kids probably won’t believe me.

• “Are we up to 7 “R’s” now?” // it feels like there are only seven people who will get this joke, and I know ow three of them. From Seroter.

• “Many prey animals will have no coherent image of what their greatest predator actually is” Here.

Logoff

I feel like I’ve been busy, in a hectic state for awhile. But I’m not sure what I’ve done. Trying to look at it objectively, over the past two weeks I’ve (1) given a talk at DevOpsDays Amsterdam, (2) written 90% of an article about one of our recent surveys, (3) edited a co-workers first-ish blog post, (4) filed some expenses,2 (5) polished up my resume,3 and (6) done all the usual family stuff. This is actually a lot of work! But, it still feels like not enough. Anything that does not result in publishing, and then acclaim (or at least more than a handful of feedback) feels like “not working.”

I don’t know what that is. It’s not imposter syndrome or something from the bag of DevOps pop-psycology. I don’t think it’s, like, lack of GTD’ing or any kind of nudges-habits airport book lore. It’s weird, and unsatisfying.