#57: BUZZWORDS, Architecture, AI Hype
In this issue:
Buzzword Fusion
Titles are like traps for my attention. Sprinkle in "generative AI" or "sustainability" and it's like I can't NOT click. So here's another one…but is it me, or is nothing substantive actually claimed in this piece? Take note of phrases like these:
"Looks to incorporate"
"Benefits should be carefully balanced"
"Presents an opportunity"
"Generative AI can help" [emphasis mine]
"If we can, for example"
These are code words. The code is: "There are no specific stories here - this is a buzzword piece. Its only value is eyeballs."
The analyst gives an example about Airbus, but if you read carefully it's not specific about generative AI or sustainability. GenAI's use is not described, and the sustainability example is a design projected to reduce emissions, if implemented.
I care about the ecological impact of sustainability. I care about the potential of generative AI. Those two things are substantive. This article, however, is not. PM
Nuve Platform
Brought to you by Nuve
Developers! Have you ever thought “gosh, if only I could just get my own SAP Sandbox to play with”? Are you stuck in ECC hell and wondering if your skills still have value in the job market? Want to learn modern ABAP without dealing with learning Basis or wondering if your laptop has enough RAM to run HANA?
Look no further than Nuve Platform! Get into the SAP Sandbox of your dreams with no playground bullies to deal with. Watch this cool video and use code NotBoringDevelopment to get the Developer Plan free for one month!
[PM: I use this product, and it really does work. Read my review here.]
Full-stack Has Failed
I cringe when the word “delicious” is used in the non-edible context, but I’ll make an exception for the delicious article Why the Frontend Kingmaker isn’t Full-Stack: A History by Kate Holterhoff.
For those out of the loop, “front-end” refers to the parts that the users see (like a web page or SAP Fiori app) and “back-end” is not some shady hacker stuff but simply dealing with business logic and data access parts that users don’t see. The developers who can (allegedly) do both, are referred to as “full-stack”.
In the SAP world, ABAPers were the OG full-stack developers, handling everything and the kitchen sink from Dynpro screens and ALV to database updates. Designated front-end development entered the picture only when SAP Fiori became mainstream. And now we are right in time for full-stack development becoming a myth. Which is a good thing.
You can’t learn the whole stack. Nobody can. Maybe it was possible in 1990, the day after the web was invented, but I’m not even sure about that.
And it’s not just front/back development, there is more:
DevOps engineers are expressing a similar sentiment to their disillusioned full-stack developer counterparts. Tired of trying to master the skills required for development and operations, many DevOps practitioners lobby to resegregate these roles into separate domains in order to lighten the mental strain imposed by assuming a mastery of both.
And SAP world just recently got into this DevOps thing. Oops.
“Jack of all trades, master of none” is an old expression. It was foolish to assume the developers were different species who could do everything and be masters of all. JP
All Aboard The AI Hype Train!
As nerds, we are always excited about technology. And you might have noticed an alarming increase in AI content in this Nerdletter lately. But since we live in the enterprise world, it’s good to also look for answers to stereotypical questions like “where is the beef?” and “is the juice worth the squeeze?”. Aka “business value”.
LI article Is Company Intelligence dropping? offers a great quote:
Several times a week I get contacted by a company that wants to do "something with AI and preferably with Generative AI". When asked what business case or scenario they are looking into the answer is almost always the same "We have been told by our top management that we need to be on the GenAI wagon and we need to work it".
“We need to do something with <insert technology name>” is too familiar to the IT folks who’ve been in the industry for more than a few years. One day it’s enterprise services, then it’s blockchain. I’m sure something else will turn up later this year.
Jon Reed from diginomica recently called out “AI circus” (“17% of workers believe AI is about as useful as a screen door on a submarine at work” – LOL, ouch) and even Jon Stewart, who returned as The Daily Show host (hurray!), shared some hilarious but also rather disturbing AI wisdom (“AI: it’s brutal, if you think like as a human”).
GenAI is a much bigger and more interesting concept than blockchain, so this time, “it’s complicated”. I would say though that AI Kool-Aid intoxication can have much more damaging results than just sending your IT on a useless wild goose chase for “how can we use blockchain”. JP
[Two of my stories have some mention of AI, as well. Reality is warping around it. PM]
Architects Gonna Architect
Josh Greenbaum posts about the Next Generation SAP Enterprise Architecture Forum, and wonders about what exactly is enterprise architecture? You should just go read his piece. It's full of goodies. (Maybe read Jelena’s earlier architecture writings as well!)
It makes me think back to earlier times in my career, when I was first encountering architecture from a technical perspective. I owe a great deal to all those I was able to learn from. Here are a few things I took away from those formative years:
Architecture is knowing when to open a black box, and when to let it sit. The first [solution] architect I worked with as a software developer had mastered this skill. I'd give her a laundry list of things I was concerned about, and she'd blow right past everything I thought important. She knew those things could be solved (she'd often done it herself in previous roles), and instead drilled into the things I thought least important. And she was right.
Architecture is being able to visualize more than one perspective, in service of a goal. An exemplary architect is always running a mental simulation of the other stakeholders. Every time you think you've drilled into the heart of the issue, they say "X is not going to like this, because he's neck deep in such and such problem". Then, the next time you have a deep conversation with X, you are enlightened. PM
15 Years A No-Test
The other day I discovered the slides Ethan Jewett shared on LI. (Ethan happens to be my manager IRL, so I need to tread lightly here. Sweaty palms.jpeg) At first I scrolled through wondering why hasn’t Ethan shared this fantastic material about testing in SAP (TDD, unit testing, etc.) with the team. And only then I noticed the date: 2009. Yikes.
It’s been 15 years and these slides seem fresh not just because they were ahead of their time but also because these ideas still haven’t caught on among the majority of ABAP developers.
The reasons are likely the same I pointed out in this 2018 blog post that followed the controversial openSAP course on ABAP Unit. On one hand, it's a chore for developers, hence they don’t want to do it. As this tutorial describes, you need to write more code to test the code, so it’s a rather hard sell. (TDD fans be like: “See!!! This is why you start with the test!” But this is an even harder sell, to be honest. There is a reason some edgy videos call it Toxic Developer Disease.)
On the other hand, the ROI or any tangible value is not clear to the people who handle the budgets and would be in a position to apply pressure on developers “to do the needful”. Code with unit tests will take longer to write. And it’s kind of the point because you spend more time upfront, so that you don’t spend as much time later maintaining the code. This is not very appealing to the “stakeholders” who think in terms of the current quarter and current project though.
To have a shot at making Ethan’s 2009 utopia a reality, both ends of the unit test dilemma need to be figured out. It should be easier for developers to create unit tests and at the same time the value has to be better understood by the purse string holders. There is no shortage of testing evangelism. (Paul had a story about this and as I’m writing this, another post popped up on LI.) But it seems a stronger push is still needed. JP
Amazon’s Just Walk Out Just Walks Out
Amazon’s Just Walk Out store feature, allowing customers to just pick items from shelves and walk out, is being phased out. It was apparently still too dependent on human workers reviewing and labeling videos. We all suspected this was happening, right?
The Bloomberg piece calls out the first thing that came to my mind: x.ai (no not that one), a personal assistant tool that debuted in the mid 2010's, but allegedly couldn't release the reins to the AIs - so they kept more humans in the loop of their assistant's automated emails than they could afford.
(I thought that x.ai was the bee’s knees! At that time, it was very much one-way: I loved having an AI do the scheduling comms, but others felt like I was shrugging them off only for my own convenience. Without ubiquitous use from all parties, people found it weird that I diverted dinner scheduling to some maybe-human-maybe-bot entity. Everybody getting a Copilot of some flavor changes this equation.)
AI itself isn’t phony, I use it daily to help people automate tasks. But us pathetic meatbags still need to go over many results and correct things. Have we reached the point where the ratio of human review to automated action is low enough? For some things, like Amazon Just Walk Out, the answer is still “no”. PM
Exciting new Show and Tell video is out: Lars ‘The Legend’ Hvam shows some cool ABAP stuff he does in VSC. Not to be missed.
Watch the Nerdletter Talk 57 video for some bonus material and edgy commentary.
Would somebody please thing of the nerds? uWu Consider supporting this Nerdletter by buying us a cup or two of coffee. Thank you for your continuous readership and support!