When a good teacher seems like cheating

All of the worrying about cheating with ChatGPT is hiding its biggest potential: being a teacher. As an “expert” (or, at least, well read) in digital transformation, DevOps, cloud native, blah blah, when I ask ChatGPT to do things for me, it’s mostly smell-free garbage, kindergarten stuff. But, when I ask it to tell me about something I don’t know about, it’s great.

For example, I’ve been reading Against Purity, which does the community thing of using a bunch of jargon without explaining it. One word was “syncretism,” which ChatCPT did a good job explaining it to me. What was especially magical about this little moment was that I used the text grabbing feature on my iPhone’s camera app to just copy and paste the text from an actual book into ChatGPT. Magic!

ChatGPT also great at simplifying dense text and ideas I don’t understand. It can be brutal in making a concept overly simplistic, removing all the artful phrasing that experts will use to describe something. Experts tend to either over explain an idea or over explain it.

When the CFO Says No

For example, I talk a lot about legacy software, and the need to modernize it. Here’s the slide I used to give a definition of it:

The longer definition is pretty good for people who are vaguely familiar with software. In focusing on the outcome (what it means), it doesn’t really tell you why legacy software causes these negative problems, nor wha to do about. But, someone knowledgable with software already knows why and what to do about it. Anne is just giving a good definition we can all agree on. Which is very important: just imagine if we all agreed on one defining characteristic/metric for “DevOps” or “platform engineering” (I’d suggest app deploy frequency) and we could move on with our lives.

The Michael Feathers one is an example of stripping away everything but the essence of a definition and make it into a tool. A developer who reads that will say “yes, I get it now. Legacy software is risky to change because you can’t verify that your changes don’t break things.” And then, those seven words also tell you what to do to escape the legacy trap: write tests.

And then my definition - the title of the slide - is just me being cute.

But to an outsider, I’d imagine these definitions aren’t very good. Here’s a definition from ChatGPT. I asked it to reduce a longer answer into a paragraph:

Legacy software refers to outdated or obsolete systems that are critical to an organization's operations but haven't been updated due to high costs or potential disruptions. Despite their antiquated nature, they are often maintained because they fulfill essential business functions. However, these systems pose increasing risks over time, such as system failure, security vulnerabilities, and integration issues, while their maintenance becomes more challenging due to the dwindling expertise in the old languages or hardware they use.

After a little bit of priming, I then asked ChatGPT to explain the Michael Feather’s definition to a CFO who is making decisions about budgets and doesn’t understand legacy software:

This definition highlights an important aspect of legacy software from a technical perspective. In modern software development, "unit tests" are like mini exams we give to bits of code to make sure they're doing their job correctly. If a piece of software doesn't have these tests, it means its functionality hasn't been assured in a systematic, automated way. This could be a risk as it makes the software more prone to undetected errors. So when we say "legacy code is code without unit tests," we're saying it's older software that hasn't been built with these modern safety nets, making it more challenging and risky to maintain or update. For a CFO, it means these systems could lead to unexpected costs and disruptions if they fail or need to be changed.

Coming at from another direction, you a developer could ask ChatGPT to explain things to CFO is an expert at! With a little bit of going back and forth, I had it come up with an explanation a CFO might give for putting off legacy modernization:

As the CFO, I see the value in upgrading our legacy software, but we must prioritize the most impactful investments to meet Wall Street's expectations this year.

Our existing vehicle fleet, which is progressively more expensive to maintain, is a point of concern for both us and financial analysts. Investing in an updated, efficient fleet reduces these costs, satisfies eco-conscious customers, and offers a more immediate return on investment.

Now, when I mention increased amortization, I'm referring to the process of spreading the cost of this fleet upgrade over the useful life of the vehicles. While this means we'll see an increase in our expenses in the short-term due to the costs associated with the new vehicles, this is far outweighed by the long-term benefits.

This immediate capital investment boosts our operational efficiency by reducing the ongoing expenditure related to fleet maintenance. Moreover, newer vehicles would lead to an uptick in customer satisfaction and usage, directly contributing to our net earnings.

These factors should collectively improve our stock value, addressing Wall Street's expectations for prudent management of our capital resources and strong earnings performance.

Regarding the software modernization, it remains an important long-term project. But given the higher and more immediate returns from updating our fleet, I recommend postponing major software upgrades to the next fiscal year. We'll continue necessary maintenance to keep our systems running smoothly and securely until we're ready to invest in a full upgrade. I hope this offers a clearer picture of our financial strategy for this year.

I bolded the most important thing that the developer might not be considering: yes, but my job is to increase the share price as a function of being a responsible steward of our investors money.

An expert in one field forgets they need to be taught in other fields

The CFO’s explanation is much longer than the corresponding legacy modernization pitch to the CFO. I tried shortening it (scroll down to the CFO part of the chat I’ve been doing here), but then I realized a fun point: my need to have it longer is because I’m not an expert on this stuff and I need that extra simplification and expanding it out. That’s the whole point I’m making here! If I tried to piece that together from wikipedia, Investopedia. reading some CFO Weekly trade publications, or even people commenting on a TwitterMastodonBlueskyThreads question…I don’t think I’d get as good an explanation.

(All y’all DevOps-heads out there can fancy this up by name checking the Dunning–Kruger effect. As a side note, if you look at the part of my chat session [towards the end] where I try to get ChatGPT to use all of this an example of Dunning-Kruger, you can see another fun teaching moment: an explanation that is totally unexpected [by me] due to different framing that I’ve been taught to have when it comes to Dunning-Kruger. That is, in the DevOps world, you often invoke behavioral economics and cool psychology tricks when you want to point out problems, be sarcastic, and otherwise be negative.)

Students are rarely good at being a student

Teachers are in short supply, and it’s impossible to search for answers like this, let alone ask for a new explanation, ask for clarification, etc. And yet, there’s so much worrying about ChatGPT ruining education. As with so many tools that can automate “the basics,” the initial freak-out reaction misses the benefits of doing away with bullshit work. When comes to education, there’s another benefit: scaling teaching, trying break the constraint of the supply of human teachers.

One of the most eye opening things I learned as a student was that to be a good student you have to learn how school works…how to do the actions of student-ing. There are deadlines for turning in work. You will be told what needs to be in your history paper (the KPIs/MBOs/OKRs!). You must write your citations in this one format. You need to show your work in math. You can read a 400 page novel easily if you just read a little bit of each day. You should be taking notes as you read that book, and in class…especially if the teacher has told you what needs to be in your essay at the end. When you take a test, go through any answer all the questions you know the answer to first and skip over the ones you don’t know; that is, don’t get blocked on question three out of ten, when you know the answers to questions 4 to ten.

I don’t think we exactly know how to integrate something like ChatGPT into that “student-ing.” I mean, it’s only been in the mainstream discussion for, let’s say, five months. There hasn’t been enough time to learn what works and doesn’t work.

Socrates famously thought writing was a terrible technology that was going to destroy society. Like so many things, he was being an arrogant dumb-ass. Never mind video games, I think calculators were going to destroy society, and people think that having a computer in your pocket and access to the Internet is going to destroy society. I’m pretty sure people thought type writers were going to destroy society, and for a couple centuries, printing a book always ran the risk of being burned at the stake.

Instead of worrying about students using ChatGPT to cheat, and, thus, destroy society, maybe we should all focus on how to use it as a new tool. That’s what the history of everyday technology would teach us.

You can now buy enterprise support for those millions of Spring apps that are running the world

We’ve soft launched a new business selling enterprise support for Spring. If you’re involved in software and IT in a large organization, I guarantee you you’ll be interested in this. If you know what Spring is, you know why: it probably runs huge chunks of your business.

If you work in IT, even if you have no idea what I’m talking about, you should be interested. Pretty much every large organization builds and runs Java app, as RedMonk shows every year. This means most of them use Spring. Spring is everywhere and runs a large portion of the world as we know it.

As with anything that helps run there world, as versions of Spring (and especially the wildly poplar Spring Boot) come and go, you’ll want to get bug and security patches, including for very old versions that aren’t exactly supported well for free by the open source community. Plus, you’d like someone to call when you have questions, and it’s always cool to have training and certifications.

Anyhow!

You can buy Spring support now, just on its own, very straight forward and easy. It also includes tcServer, Tomcat, OpenJDK support. Don’t hold me all of this: I haven’t, like, taken the training to know the official menu. I’m just going off the web page.

And, yeah, this is enterprise software so you can’t just put your credit card into gumroad or something: you have to contact us to talk with an enterprise sales rep. I mean, you’re probably going to need to deal with your EA compliance cabal, and, then, procurement department, so, like, we’re already in the enterprise sales game zone. Also, make sure to ask them for a fancy steak dinner (unless you work for the government, I guess). And don’t miss out: ask before and after you close the deal.

Check it out!

I’m writing about this at length because I’m excited about it, both on its own, but as a strategy experiment. Pivotal and, now, VMware1 haven’t really directly monetizing the Spring Framework as much as they (rather, “we,” since I work here) could - as much I thought they should.

This is largely on purpose: the Spring open source community there has done well to remain, like, “pure.” Things have been fin! I work there, so I’ve seen plenty of second-order monetization over the years, both by us and others. Spring runs fantastically in the Tanzu Application Service (previously called Pivotal Cloud Foundry) and is first in line for support in our Kubernetes platform toolkit, the Tanzu Application Platform. And, you pay for Spring directly when you use the Microsoft Azure Spring Apps service. (With all my “hold on, this Kubernetes thing might be going too far?” waffling, I’m starting to think that some of the people I talk with who want want to make Kubernetes easier to use for developers just use that and get on with their lives.)

However, all the time, every year, we encounter the basic question “can I pay for Spring support?” Strategically, it’s also a fun experiment: it’s selling directly to developer groups, a very simple value-proposition (versus the very valuable, but very big and complex platform products we have), and, you know, kind of obvious. It’ll be fun to see how to turns out.

I don’t really understand all of this, but there’s also a lot of stuff going on in the Java world in at the moment (and through the next year) that creates a market window. That’s the seller-side way of putting it. On the buy side: there are some decisions to be made about how you use and get support for enterprise Java applications coming. Some of these are pure business, and some of them are just standard upgrade cycle events.

You should check it out!

Making the “DevOps Engineer” title acceptable

DevOps Thought Leader: if you have a group called "DevOps Engineering," you have no idea what DevOps is. You should at least feel bad if it’s your title. [Crowd laughs.]

Platform Engineering Thought Leader: Got it. Here, hold my beer. [Squints and grins]What if we called the group "Platform Engineering."

DevOps: Wait, hold on, that's not what I meant...

Platform Engineering: Hey, sorry, I was busy. Just launched a Slack and conference. posted 20 sponsored articles on The New Stack, and my O'Reilly book is coming out next week. Gotta go finish that post on Backstage sprawl now.

DevOps: …

Platform Engineering: oh, also: could you retweet my thread on “ten things you’ll got wrong about platform engineering”?

Wastebook

• Has business travel decreased, or it just me (and the tech world)? Ask perplexity: “Overall, it seems that while corporate travel budgets are recovering to pre-COVID levels, the number of trips taken is still lagging behind. Spending on international business trips is expected to pick up in 2023, but the frequency of work trips is unlikely to fully recover.”

• Culture is impossible, tools are easy.

• Threads: finally, an app for centrists.

• “I would dearly love a world full of these things. It takes me back to the days of a more creative time on the early internet, and makes me wish for a future that has more of this in it.” Here.

• What’s the 2023 take on the ethics of being annoyed? Not like, offended and hurt, just, like, annoyed in the way a kid if upset when you serve the Brussels sprouts.

• “I need a respite for my mind.” Dear John Letters #23

• I’ve been using HEY! again for email. I almost am there in liking it. I can’t wrap my mind around the idea of no archiving, and I don’t really like the way it lists The Feed. I want my newsletter subscribers to show up as one line and hide the ones I’ve already read. But, as I said to Dave Briggs when he asked me about it: “I have this ongoing theory in my digital life, though, that I should stop using tools that are too perfect, and instead use tools that seem to embody (or limit you to) a certain world view or life-style that looks good. Hey certainly seems like that. It's a kind of attitude about email that's a little less perfectionist/completionist. It doesn't value email as much? I don't know.” This is a twist on the #DefaultsLifeStyle.

• Also, he started a newsletter. You should try it out!

Relative to your interests

• Amazon Prime Video’s Microservices Move Doesn’t Lead to a Monolith after All - The New Stack - At this point, who knows anymore? I think the secret to being a grey beard in tech is to never say “oh, that’s just like SOA!” but instead to just nod and say “wow, that’s a great idea!”

• Facebook’s Threads is so depressing - Fantastic writing!

• Five Nos - “‘yeses’ matter more than ‘nos,’ and taking more chances generates more of each. While there’s a near equal chance of getting told ‘no’ in our endeavors, in virtually all cases, there is no penalty for that rejection beyond some brief discomfort or embarrassment. On the other hand, when you get a ‘yes,’ there is likely a positive outcome on many fronts.”

• How legacy tech can kill recruiting efforts, increase attrition and ruin the employee experience, Conner Forrest - “451 Research’s Voice of the Enterprise: Workforce Productivity & Collaboration, Employee Lifecycle & HR 2023 survey listed ‘candidates expect software or tools we haven’t adopted’ as the top recruitment challenge (52%) they face. To put that into context, the next-highest response was expectation of flexible work (42%) and then compensation expectations being too high (40%).”

• Rocky Linux project details how it will live on - “Nobody chooses RHEL because it’s state of the art. It isn’t. In fact, it’s about as far away from state of the art as it’s possible to be, and in the rapidly changing world of open source, that’s a very desirable attribute for a certain type of purchaser. They choose it because one specific version will be supported for a decade plus, and that in turn is why vendors support RHEL, and often nothing but RHEL.” // Also, the tricky way one of the clones plans to get the RHEL source code to make their clone distro.

• Gartner Survey Finds 79% of Corporate Strategists See AI and Analytics as Critical to Their Success Over the Next Two Years - This seems very true: “Strategists said that, on average, 50% of strategic planning and execution activities could be partially or fully automated; currently only 15% are.” And if you established a stranded memo format, even more so. And you could have it generate presentations for you. Key will be, as always, getting access to the market and internal data needed.

Upcoming

Talks I’ll be giving, places I’ll be, things I’ll be doing, etc.

July 11th How Cloud Native Improves & Ensures Security, Governance, and Trust in Finance, online talk. August 21st to 24th SpringOne & VMware Explore US, in Las Vegas. Sep 6th to 7th  DevOpsDays Des Moines, speaking. Sep 13th, stackconf, Berlin. Sep 14th to 15th SREday, London, speaking Sep 18th to 19th SHIFT in Zadar, speaking. Oct 3rd Enterprise DevOps Techron, Utrecht, speaking

Logoff

Of course I have a Threads account now. Having seen the meager uptake of other social networks over the years, I’m overly pessimistic, but I give this one a 70% chance of being wildly successful. If they actually do good stuff with whatever open standard, it’ll be like a new type of RSS. If they let me merge my Mastodon account, that’d be amazing. (Is that a thing you do with Mastodon? I know you can move to a new account, but you can merge your old account into your existing new account?)

Tomorrow I have the first installment (hey, it could always be the last installment if I get bored with the idea or just forget I was going to do it) of The Podcast Review. Here is the premise I was talking about (also) with Brandon this week:

I listen to a lot of podcasts, and have for, I don’t know, over 20 years. I’ve made and make a lot of podcasts. You might call me both a podcast listening and podcast creating expert. In the classic sense: a critic. I should review podcasts more frequently!

Make sure you’re subscribed so that you’ll get it!