Some "Coherent Nonsense" for the weekend
If you think writing the text is bullshit, get the bullshit artist to do it. Or: the AI helps me write an English class essay on Susan Sontag, an inside sales email, & webinar landing page copy.
Current status: I’ve got a good, old fashioned cold. I’ve forgotten how weird it is: you just find yourself sort of staring at nothing. And the snot. I didn’t have time to self-promote my content. Next time! Also, you know where to find it.
English Class Essays, Insider Sales Emails, Webinar Landing Page Copy - More ChatGPT Experiments
Like I said last time, my theory is that the ChatGPT thing is a great tool for brainstorming and writing - for thinking. It knows the structure, the game of all that, and often knows basic content as well. To that end, I keep looking for uses that make me more productive (I don’t do the toil of starting), help me come up with ideas out of the blue, and as I came across this round shorten things.
Here’s three experiments (is that the right word?) I’ve done recently:
Help me write an English class essay discussing Susan Sontag, Joan Didion, Hunter Thompson, and Chuck Klosterman - I’m familiar with all of them, but I am genuinely kind of confused about what Sontag’s deal with “camp” is. I know she defined that idea, but I couldn’t tell you what the point was. The AI’s answer seems OK? And then, the quality of the essay we work on writing about these four authors isn’t that great. However, working on my theory that what you want to do with ChatGPT is have it brainstorm and model with you, it did come up with some interesting notes to work on. Despite liking all of these four authors, it’d take me awhile to just, well, have something to say about them. My theory with this would be that if I asked it compare specific articles/essays it would do a much better job. The part about Hunter Thompson is, you know, coherent nonsense, as some folks at Forrester put it.
Writing better cold-call emails - I get a lot of those emails pitching B2B services to me. I took the most recent one I’d
deletedread and asked the AI to improve it. I don’t think it did a good job - the first email was pretty good. However, once I asked it to shorten it, I think it nailed making a better email. The original and the AI-generated short one are both equally good, I think. This makes me want to try getting it to shorten other things, especially talk abstracts. I tried getting to lengthen other text to make a longer article, but the results weren’t too good.
And here’s another one. I have a talk about the application modernization book Marc and I wrote. It’s a good talk! You’re always having to come up with talk and webinar abstracts for these things which I HATE. So, perfect use for an AI-as-toil-reduction-tool, right?
My prompt:
Can you write a title and two paragraph abstract for a webinar that will explain what legacy software is (maybe mentioning tech debt) and why it's holding companies back from growing (over 78% of management says that legacy is holding them back from improving their software). The webinar will cover the methodology VMware Tanzu has used for ten years to quickly modernize enterprise's legacy software. The methodology focuses on finding the most valuable applications and starting small to deliver early, learn the safest way to modernize, and scale up over time.
It's answer:
Title: Overcoming the Challenges of Legacy Software with VMware Tanzu
Abstract: Legacy software is often a major barrier to progress for businesses. Over 78% of management cites legacy systems as a hindrance to improving their software, and the problem is only compounded by the rapid pace of technological change. In this webinar, we will explore the challenges posed by legacy software and discuss a proven methodology for modernizing it. VMware Tanzu has been using this methodology for over ten years to help enterprises quickly modernize their legacy systems. We will focus on how to identify the most valuable applications and start small in order to deliver early results, learn the safest way to modernize, and scale up over time. By the end of this webinar, you will have a clear understanding of how to overcome the challenges of legacy software and move your business forward.
Three Emerging Principles/Guides
Now, this is where the “I’m not cheating on my homework” thing comes in. The write-up I did is nearly as long as the actual abstract. I just didn’t want to do the work to make it formal and “better written.” Is it better written? You could say that my version, in that I didn’t polish it up is more genuine and “true” (in the Hemingway sense). Or not! You could also say that the need to polish up these abstract is kind of, like, annoying and bullshit. I sort of feel like the second - maybe that’s an asshat thing to think! But, it’s guiding me to this principle:
If you think it’s bullshit that you need to write the text, get the bullshit artist to write it for you.
Here’s how ChatGPT made that more quippy:
If you think writing the text is a load of BS, hire a BS artist to do it for you.
I’d shorten it (and cuss it up) as:
If you think writing the text is bullshit, get the bullshit artist to do it.
That, of course, isn’t the totality of my thinking.
My other evolving theory (to repeat it) is that it’s a good tutor/co-author, for “workshopping” ideas. As I was saying on the podcast this week, if you’ve ever worked with an executive who’ll go up to a whiteboard and draw out a bunch of ideas (slides even!)…and then ask you to go make slides or a memo of it…and then you go back and forth on the slides over and over…the ChatGPT thing is good for that…you, the underling, escaping the shit work of being the executives content-mule. Instead, you should get the AI in the room and both of you can be the executive. I don’t know how to summarize that principle. Maybe:
If you’re going to delegate some content work to a human, why not get the human on your level and both of you delegate it to the bullshit artist?
One obvious concern with all this is that over time, the AI puts us all out of a job, or, worse, make us weak as a society (you know, where did all the strong willed, “grit” kids of the 1950’s go kind of…uh…bullshit).
This is, I don’t know, kind of Evil Top-Hat Capitalist was to thing.
Job displacement and “weakening society” is a risk worth considering. But the chainsaw didn’t make us weaker as a society just because we didn’t need to chop wood anymore. Nor indoor plumbing or electricity. Socrates was afraid that writing would ruin the Athenian intellect, but he was clearly (and ironically once Plato wrote it al down!) wrong about that.
I think the biggest risk/downfall with any productivity in improvement is that “the capitalist” in charge don’t use it as a reason to move from a 40 hour work week to a 30 or 20 hour one. If a productivity improvement cuts down on how long it takes to get the current output, we should work less instead of then focusing on doing more output. But maybe I’ve been living in Europe too long.
Let’s call this last principle the “go home early principle”:
Don’t let it be an excuse for doing more work.
I found the better ChatGPT
I’ve started using the “playground” which (sort of?) has persistence to it. One of the biggest problems with the ChatGPT interface is that it just loses everything when you close the browser. I mean, it’s free, so sure. However, I just tried to look at my “30 day history” in the playground, and it only has today’s in it. Maybe I cleared it by mistake.
You can actually pay for more usage in the playground and I haven’t gotten the “servers are overloaded” or "you’re asking too many questions throttling I did on ChatGPT.
I don’t know how I’ll go beyond the free tier. With all the time I’ve spent with it this week, I’ve only used $1.41 out of an $18.00 credit.
With this pricing, for personal use, it’s virtually free.
I still really, really want to be able to feed it my own text. I don’t know how one would do that. For example, I’d love to just point it at my blog, this newsletter, upload a bunch of PDFs, transcripts…basically all my (only public?) content and anything else I wanted to work on. For example, it’d be great to load up all analyst reports, the technical documents, and so on. Then I could customize the work we do to my view of the world, way of thinking, etc.
(And, yeah, I noticed that I’m saying “we.”)
Oh, and was that too wordy and boring to read? I asked it to shorten it to three paragraphs for you:
I've been using the playground w/ persistence & so far only used $1.41 of an $18.00 credit. It's virtually free for personal use. One of the advantages is that it does not have the "servers are overloaded" or "you're asking too many questions" throttling I did on ChatGPT.
I really want to be able to feed it my own content, like blog posts, PDFs, transcripts, analyst reports, etc. & customize the work to my view of the world.
And yeah, I noticed that I’m saying “we.”
AI Pictures
Also, some of DALL-E image stuff:
I keep trying to get it to make Bruegel-style scenes. It doesn’t exactly do it right, but it’s interesting.
Us recording the our podcast - while this matches what you’d envision use looking like when we record the Software Defined Podcast to look like, it’s not…mostly…OK…it’s sort of close.
I generated one to start using as the header for these emails, as you can see if you’re subscribed.
Hey, why not subscribe if you’re just reading this on the web!
Waste Book
Do more than a one-to-one mapping when you “digitize” a business - you should take advantage of digital stuff to do things in new ways that were not possible in an analog world. You could, for example, easily set spending limits on debit cards you give to kids and restrict types of merchants they can but from. That would serve a customer need (parents that want to give their kids the ability to pay for things, but not just a bunch of candy and Pokémon cards).
Top-line vs. Bottom-line Pitching when selling enterprise software, or anything really - It’s really easy to sell technology that reduces costs (but does nothing new), like virtualization. That is bottom-line, your costs. It’s really hard to sell technology that will improve how your business functions. That is top-line, your revenue before any expenses. The first can be quantified in a number ahead of time, while the second cannot until years later. If you have to compare, or sell against, the two, it gets real annoying.
Relevant to Your Interests
AWS Previews Amazon CodeCatalyst to Automatically Create Dev Environments - DevOps.com - Another go at defining templates for both new applications and the cloud configuration needed.
9 Insights on Real-World Container Use - Good stuff in here.
Apple Expands End-to-End Encryption to iCloud Backups - // Apple’s (helpful) obsession with personal security is sort of fascinating. Why do they do it?
Do you know more about Kubernetes than ChatGPT? - For some people, this would be considered “bullshit,” but if you were just idly wondering what kubernetes is, you can see that auto generated articles would be helpful.
Shareholder Value - “In Douglas Rushkoff’s new book, he acknowledges that he was wrong to say, as he was perhaps the first to do, that to the social media platforms you are not the customer but the product. Rather, he now argues, the company’s stock is the product; you are the unpaid labor that increases the value of that product.” That’s clever!
The Hollow Core of Kevin Kelly’s “Thousand True Fans” Theory - Getting a thousand fans is hard. Also, it will barely cover your base costs (overhead, if tot prefer): you have give up 10 ro 30% cut if the revenue. Then there’s taxes. Then you need health insurance. You need to pay rent, for some other online services. You’ll probably need advertising too. Could you even run a business on $100,00 a year, including your own salary? But, that means you have a foundation to build on. Maybe you need 5,000 true fans.
Twitter comes of age - “More than any other time, if you are not on Twitter, you just don’t know what is going on.”
Beware of “Coherent Nonsense” When Implementing Generative AI - That’s a good way of saying “bullshit.”
KPMG: The playbook for a successful business transformation - “our experts comprehensively reviewed 140 enterprise transformation projects at public companies and found that 73 percent failed to simultaneously deliver top- and bottom-line growth”
Programming Note
You may notice something slightly different. I moved the email newsletter to sub stack. You know me, I like to change it up every six months. We’ll see!
This means we now have all those (annoying?) substack features like comments and whatever an “author chat” is. That last seems horrific.
But, I mean, you know: I want more readers, I crave eyeballs! Why not share it with someone, or, really, everyone you know: