Working Up

by Dwayne Phillips

This is a simple concept. We often confuse the terms.

Precision has to due with the resolution of a statement. Are we talking years, months, day, or seconds?

Here is a precise estimate: we will finish the project at 2:34PM EST Tuesday July 23.

Accuracy has to due with the correctness of a statement. Sometimes the simplest examples sound silly, but they are accurate.

Here is an accurate estimate: we will finish some time after now.

People often confuse the two concepts and use the words incorrectly. Let’s think a little more before we speak and write.

by Dwayne Phillips

Of course it makes sense, it’s only words.

Let’s begin this post with a story. I worked for the Federal government for a few years. Each employee had an annual Performance Appraisal Report or PAR. The supervisor would summarize how a person performed during the past year. The PAR was an important document that played a role in how the person would advance in their career.

Supervisors had to write the PARs. Most supervisors hated writing the PARs. Most supervisors couldn’t write well. Angst.

Enter an old friend of mine. He had worked 30 years. He had kept copies of all the PARs written by his supervisors about him. He supervised several people. He had to write several PARs a year.

No problem. He would open his file cabinet draw, randomly grab an old PAR, randomly put his finger on a paragraph in the random PAR, and copy it word for word. Put the random PAR back in the file cabinet. Repeat this five times. Five paragraphs. The PAR is done. Just randomly copied paragraphs.

My old friend was often complimented on how well he wrote PARs. It was only words. It worked.

Enter the era of AI and Large Language Models and all this wonderful new stuff. Same thing. It’s only words. We have a context. We have preconceived notions of what we should be reading and what the text should say. We read into it what we want.

Wow! That AI system is a genius.

Wow! My old friend who randomly picked paragraphs out of randomly picked PARs was a genius.

It’s only words.

Of course, this post is silly. There is much more to it than random paragraphs and all that. Perhaps.

by Dwayne Phillips

I am a prolific writer, well, sort of.

I am in a temporary state of work and semi-retirement and hobby. For the past couple of weeks, I idle away the hours writing fiction.

In a recent week, I wrote two short stories while doing a dozen other things. I wrote a 3,300-word work of fiction in two-and-a-half hours. Not two-and-a-half weeks, hours. As in sixty-minute hours.

Okay, maybe the story was junk. Then again, maybe if I could type faster I could have cut that time.

There are all sorts of formulas about how many words a minute a person can type times the hours in the day divided by the number of words in a novel or short story. For example, 40 words-a-minute is 2,400 words-an-hour. Hence, a 3,300-word short story is done in under two hours.

People hear those formulas and scoff. “That works if you already have the words in your head.” Maybe my head is overly big and holds too many words to make any sense.

Well, given those formulas, I can write a novel in two weeks. Junk? Maybe. But then, I can take a day off to recharge, and then write another novel. Two novels a month, let’s see, carry the one… that’s twelve novels in a year. No, wait, that’s 24 novels in a year. Can a person write 24 bad novels in a row? Don’t answer that. Anyway, that productivity is impossible. Well, take two weeks off for Christmas and another week for Thanksgiving, and toss in one for the 4th of July, that cuts out two novels in the year.

Of course this is all nonsense. Who would do such a thing? Such volume would indicate AI! No, not really. Maybe it indicates someone who needs a better hobby. Maybe it indicates someone who thinks his stories are worthwhile, i.e., the guy is delusional.

Regardless, the numbers are real and the math is simple. Old folks, like me, have a lifetime of stories in our heads. And when I tire of typing, I can use one of these fancy new software things that let’s me dictate to the computer. The words-per-minute jumps even higher. I would become hoarse.

by Dwayne Phillips

A simple piece of paper on the table next to the keyboard keeps everything flowing — especially new ideas.

I have a productivity multiplier that I use everyday and have for decades. It is not the computer that is a thousand or million times more powerful than what we used to go to the moon and back. It is a piece of paper. A simple blank piece of paper that doesn’t remain blank.

I put a piece of paper on the table next to the keyboard. A pencil sits on the piece of paper. I work away churning out words and thoughts and designs and all sort of items we produce these days. The pencil and paper sit at the ready.

A new idea. Jot a note. Back to the keyboard in five seconds. The flow of work continues. The flow of ideas on the piece of paper also continues.

A variation on what we are doing. Jot a note. Back to the keyboard in five seconds.

This continues all day.

I sit in a coffee shop. They serve food in little brown paper bags. Ah. No need to bring a piece of paper as they just supplied one.

How about writing on a napkin? That works. How about writing on a table cloth? That works, too. Writing on the table? That works, too.

Just something to jot notes in a couple of seconds. Keep working. Keep ideas flowing. Keep the work flowing. It seems to work.

by Dwayne Phillips

Rogue AI agents? Nope, it’s call malware.

Malware (a portmanteau of malicious software) is any software intentionally designed to cause disruption or destruction to a computer, server, client, or computer network, leak private information, gain unauthorized access to information or systems, deprive access to information, or interfere with the user’s computer security and privacy without their knowledge. — Wikipedia

These AI agents sometimes turn rogue and breach systems on the Internet and leak personal data that shouldn’t be leaked and destroy records and there is no end to the mischief of these rogues.

Bologna.

This is malware. This is software that destroys systems and costs people resources. Turn on an agent and later discover what it did. Oops. Wasn’t supposed to do that. Sorry.

Well, maybe a “bug” is a better description of this software. I mean the folks at those AI companies didn’t intend to do harm and wreak havoc. It was an honest mistake. Sort of like software that deletes all the files because the programmer made a mistake.

Oh, so the programmers didn’t test their software. Well, we are back to the beginning. Stop letting grad students play with software on the Internet to see what it might do. Act like adults. Yes, you are smarter than the average folks who setup the average systems and connected them to the Internet. Take it easy on those average folks. They do things for you that you don’t want to do for yourselves. And they know where you live.

by Dwayne Phillips

Something seems expensive. The question is the same: is it worth it? Will that thing save money in the long run? How long is the long run?

I read this article recently. It is about Apple and one of their recent computers that is loaded with processor power and memory and all such wonderful wonderfulness. The computer would allow a person(s) to run AI locally on local files while completely disconnected from the Internet.

The price tag wasn’t yet announced, but let’s say $10,000. WOW! That’s a lot for a home computer. Okay, I won’t spend that much on a home computer. Or would I?

Move to a business or organization instead. The organization has records that it CANNOT expose to the Internet or to one of these chattering bot companies. The records must absolutely remain in house. Privacy demands it.

Should that organization’s managers spend $10,000 on a computer so they can write little English questions like, “How has participation changed in the last 18 months?”

The answer lies scattered across a couple thousand Microsoft Word documents and emails that sit on the SharePoint system or Slack records or some such thing. Assign a person to the task and come back in a few weeks. We may ask that question every month or so. Let’s automate the task. One programmer can write the software to pull the answers.

How much will that human effort cost? $2,000? $5,000? How often will we repeat the task and pay the person? A contract programmer may cost us $10,000 to write us one program to answer this one question. Well, the hardware pays for itself in one month. And what if we have follow-up questions? What if and what if and what if?

The hardware may pay for itself in a week or a day. That’s a pretty easy decision. Of course we buy the hardware and we stop wasting money.

And what if the question at hand is silly and we don’t need the answers? No need to spend a penny on nonsense.

The question is the same: is it worth it?

We can do better.

by Dwayne Phillips

One of the better solutions to a difficult problem is simply to trying harder.

I was 17. I was a high school senior. It was baseball practice. I was the senior, the guy who knew stuff. The underclassmen were looking at me. I was supposed to teach them how to turn a double play.

My only thought was, “You need to move your hands faster. Concentrate on moving your hands faster. Focus on that. Do that.”

In other words, “Try harder.”

The advice worked on some of them. Some others rolled their eyes.

If you want to complete a task at work in less time, one of the better things to do is focus on completing the task in less time. Forget process improvement and management seminars and all the rest. Try harder on that one area. It might be what works. In my experience, I see it working often.

If you want to complete at task at work with higher quality, one of the better things to do is focus on completing the task with higher quality. Try harder on that one area.

I could go on with examples from my past and different tasks at work. Often, I simply had to try harder. Simple. Not always easy. Harder is the opposite of easier. I find it unfortunate at times, but this works.

Careful how this advice is given. It is easy to hear the big, mean, old boss saying, “You are lazy. You aren’t trying hard enough!”

Instead, try, “I am not managing this work well. We are failing in speed or quality or something. I want to focus on speed or quality or something. Let’s try again.”

We can do better.

by Dwayne Phillips

Attempts to put unstructured information into a structure had their day. Goodbye to them.

Much of the information that people use is unstructured. Consider a novel you like. Consider a text that is the foundation of a field of study. Not much structure that is easy to ingest into a computer.

Until now.

Since 2008, I have kept records of job searches, interviews, and the like. I typed those records in a Google Docs file. If I were to print it, and I am not printing it, it consumes 600 pages. It is unstructured. It is readable by humans only.

Until now.

I drug a copy of the text file into several of these chattering bots. When did I apply to a job with ACME Inc? What job was it? What was the outcome? Bang. Boom. Pow. Out pops the answer. The chattering bots understand the unstructured structure.

For years, I contemplated making a JSON or XML or something file with this information. That would allow me to query the structured information using a Structured Query Language or something. That would be the only way for me to find information.

Until now.

New tools. Some old problems solved. Waiting had its benefits. YAGNI (you aren’t gonna’ need it) came true in this case.

Goodbye JSON, XML, and the like. You worked in your era. Times change.

by Dwayne Phillips

College professors and other educators continue to struggle needlessly with writing assignments and AI-regurgitated text.

Here we have yet another story of a teacher with the best of intentions trying to have students write papers in the era of the current slate of AI systems that pump out text at the push of a button. Alas, some of these teachers with good intentions just don’t get it.

The writing assignment is a waste of time. It is needless. Admit that and omit it.

Assignment: research the writing of the Declaration of Independence or the use of robotic technology in delicate surgery or some other topic. Write a ten-page report on the topic.

Good grief. That is a mechanical exercise. And some folks are surprised that we have built machines that can do the mechanical exercise. Why don’t you just have the students look up the times tables and report on them? Needless. Machines can spew the answers.

Counter: learners should be able to write. I agree. Let’s have them write. AI slop? Come on. Think.

Assignment: watch this movie. Tell me what you think of it. Tell me something that only you know down deep inside you. What person in your life comes to mind while watching this movie? What did that person do to you? No one else has the answer you have. Write it.

That is not a mechanical exercise. We can’t build a machine that knows the deep core of beliefs of an individual.

Tell me something that only you know.

Another counter: that is okay for the fine arts. What about technology?

Assignment: perform an experiment that only you have performed—no one else has done this. Only you know what you did. Only you know the results. Write about that.

Tell me something that only you know.

by Dwayne Phillips

Sometimes, you have to work hard to untangle what someone is trying to convey. Then again, if their words are tangled, maybe they don’t have anything to convey.

I spend much of my days untangling the words of others. Long documents; long talks, and the like. What is that person trying to convey? Print the document with a wide margin. Note key words and phrases. Draw pencil lines connecting similar thoughts. Key on the active words. Find the message!

Sometimes if the other person, the person who tangled everything, tangled all the words, they didn’t have a cogent thought to express. Perhaps all that tangling indicates confusion.

Untangling can reveal the key thought. Untangling can also reveal confusion. Which is which? I wish I knew ahead of time. Still, that’s the task at hand. Give it a good effort and either learn the idea or learn that the person could use some advising and assistance.