Smart watch died, part 5 - 6 March 2025

Credit to Google, the replacement Google Pixel smart watch has arrived.  We now repeat the set-up procedure with the new watch. I will be more observant with the charging process to avoid a repeat of the original problem.  The set-up procedure is taking several minutes; I assume there is a download and update in progress.  There is no indication of what is going on, just "connecting to your watch".  Oh! and something failed, so we are resetting and restarting.  Oh! and again something failed in the set-up process.  Looks like the third time is the charm as I got to the signing-in stage.

This replacement process started on or about 23 February 2025 and completed 6 March. I say on-or-about because I went on vacation in January and I was planning to take the watch.  I went to put on the Google Pixel watch and it was exploded.  I chose not to take the exploded watch on vacation.

As to the photo, we had a dead tree taken down; it had been assaulted during the Seattle Heat Dome a couple years ago.  I hoped it would survive, but not so.  After felling, it became clear that there were internal problems in the tree with bugs and rot, so it is now cut up for firewood.  I did a deep crosshatch in the stump to accelerate the decay process.  Some advise the liberal application of epsom salts but I have not done that.  I am hoping that water, weather, and bugs will complete the process through the added access that I have provided.  I am not sure what the tree was; evergreen, certainly, and coniferous, and the general shape of the trunk suggests a Noble Fir.  It is now a face-cord of firewood.

Smart watch died, Part 4 - 5 March 2025

After going thru the required communications with Google Customer Service, I was finally able to package up the busted smart watch and ship it back to Google.  I was annoyed that I had to supply the shipping material and take it up to FedEx, but the FedEx process was smooth once I got there.  If I had a routine 9-5 job, it would have been a real pain to get to the FedEx office quickly.  FedEx conveyed the shipment quickly to Google and it has been a silent period.  No news from Google.  I received word today that they had inspected the returned item and were shipping a replacement.  Today's notice included a tracking number, but it seems that only the label has been printed. That is, UPS does not yet have the shipment but will pick up the shipment soon (tomorrow?) and they will deliver it eventually.  Note that this is a common technique - Google "prints a label" to make it sound like they are making fast progress, but the tracking number is not meaningful until UPS actually gets the shipment.

I am glad that Google has finally verified that the failure is a valid failure and is covered by the warranty for replacement, but I am still surprised at how s...l...o...w the customer-service process has been for a $200+ item.  

Smart watch died, part 2 - 24 February 2025

I thought everything was all set to get a replacement for my dead-and-exploded smart Pixel Watch.  

Yeah, not quite.  After supplying my address and the serial number, I received a confirmation email that I would get a return-shipment label for my dead-and-exploded watch.  Overnight, I received an email requesting a photo of the serial number (that I previously copied over).  If you are not familiar with the watch, the serial number is hidden under one of the attachment points for the watchband.  It is embossed on a curved surface in small letters, but I took a photo and sent it back in return email.  Because it is so hard to read, I also took a photo of the serial number on the side of the box (I still have the box).  Google Support keeps asking for the same information over and over.  SMH.

The folks are all nice, but their customer management system is - what is the technical term? oh, yeah - the Google CMS is hoarked.  

Smart watch died - 23 February 2025

It is said that all good things must pass, but this is not that story.  Almost exactly a year ago, I bought a Google Pixel Watch, a smart watch.  I tried it out for a while and, really, there is not much benefit.  Clocks are everywhere including one on the phone in your pocket, so the timekeeping function is not really useful.  Various notifications come up on your phone in your pocket, thus notifications on your wrist are just a minor benefit.  The step-counting feature is wildly inconsistent and the heartrate info is only mildly useful.  The sleep tracking is not particularly reliable, so the inconvenience and discomfort of wearing a watch to bed is not helpful for sleep.  And, finally, you have to constantly charge the darn thing - at least daily - and if you wear it to bed, you cannot charge overnight so you have to take it off during the day but then all the other advantages are shot while it sits on the charger for an hour or so.

In the end, the $200 cost is simply not justifiable.  The $40 smart watches are no better.  And who really needs a watch these days?

With all this in mind, I am sad to report that my Pixel Watch baked itself on the charger:  it swelled up and exploded itself.  It was not a violent explosion, there was no shrapnel, but the watch seal is broken and it is no longer waterproof or even water-resistant.  The guts are just hanging out.  I played 20 Questions with Pixel Watch support.  They have been very responsive, but they only want to ask 1-2 questions at a time, so our email exchange went back and forth over several days.  They now seem to have everything they need, finally I thought, and so they have passed the support case to the next level for a decision in yet another deferral.

I am expecting them to offer a refurbished Pixel Watch as a replacement.  I am happy to return the dead watch to them.  We shall see how this plays out.

AI Is Going Great, Part 1 - 3 April 2024

A couple years ago, Amazon.com announced automated stores in which customers pick up items, put them into a cart or bag, and then "just walk out".  A cloud of cameras and scanners would watch the selection process of each customer and automatically total up the bill.  AI would drive the whole systems.  Well, not so much.  It turns out the whole system was driven by 1000 people in India paid to watch the cameras and create the receipts for "automatic" checkout.

According to a report today, 

Amazon ditches "Just Walk Out" tech: The futuristic cornerstone of Amazon's grocery store experience was a lie. Amazon advertised this experience where a completely automated process tracked your moves in the store, watching what you grabbed and tallying your bill as you "just walked out," nullifying the need for any pesky cashiers. However, it turns out the process wasn't automated at all. Instead, over 1,000 people in India were watching the cameras and assembling bills for whatever you put in your basket. 

https://gizmodo.com/amazon-reportedly-ditches-just-walk-out-grocery-stores-1851381116

Web3 Is Going Great (Not) - 16 March 2024

In a move that surprised no one, Starbucks is shutting down its NFTs.

OK, to be precise, Starbucks is passing the Starbucks Stamps NFT Program over to a third-party and washing their hands.  The Starbucks Stamps NFT Program started out as a way to reward regular customers of Starbucks, but it has languished and finally gotten to the point that it is more trouble than it is worth.  So Starbucks is dumping it.

At one point, NFTs, blockchain, and blockchain-contracts were the triplet powerhouses that were going to drive Web3 into the future.  NFTs are failing left and right, blockchains and crypto-vendors are rug-pulling weekly, and e-contracts are often used to steal from crypto-vendors, so the future of Web3 is looking a little disorganized.  #NoSurprise

Reference - Engadget article, 16 March 2024 , subtitled The program ends on March 31 and its NFT marketplace will be shifted over to Nifty.

Sad Tesla News - 10 March 2024

We start today with a tragedy.  According to a news report in The New York Post, "Angela Chao, the billionaire former CEO of dry bulk shipping giant Foremost Group, tragically died at the age of 50 on Feb. 10 after accidentally backing her car into the pond while making a three-point turn." If that name does not ring a bell, note that Angela Chao’s sister, Elaine Chao, is married to Senator Mitch McConnell and served as Secretary of Labor under President George W. Bush and Secretary of Transportation in President Donald Trump’s administration.  Mitch recently retired as Minority Leader (R) in the US Senate.

Angela Chao was hosting some friends on a ranch in Texas and wanted to return to the main ranchhouse at the end of the evening.  As it was cool, she decided to drive her Tesla rather than walk the four minutes from the guest houses to the main house.  At some point along the way, she backed her car into a pond where the car sank.  

The tragedy is multiplied by the design of the Tesla.  Rather than mechanical door handles, the main doors are opened electronically with a button.  Obviously, this is not a great design for an electric car sinking in a pond (loss of power).  There is an emergency mechanical door latch, but from the descriptions, it seems that it is hard to find, especially when sinking into a pond.  Chao was trapped inside the sinking car, unable to get out.  She had enough air in the car to be able to use her cell phone to call friends, but no one was able to help in time.

Many of the prior deaths in Tesla accidents have involved fire, but this is the first to involve water.

Condolences to the Chao family.

Reference: https://nypost.com/2024/03/09/us-news/angela-chao-made-panicked-call-before-dying-in-completely-submerged-tesla-on-texas-ranch/ 

The iPhone is the new transistor - 4 March 2024

Human history is a series of quiet periods interspersed with revolutions.  So, too, science is a series of quiet periods interspersed with revolutions.  In the 1947, the invention of the transistor changed the world and in 2007, the iPhone smart phone changed the world.  It took a while for the transistor to go from laboratory curiosity to common use, but the iPhone moved into common use more quickly.  More recently, people have thought virtual reality (VR), augmented reality (AR), bitcoin (BTC), blockchain, and artificial intelligence (AI, LLM) would be the next revolutionary step.  AI may yet change the world, but the others are racing for the dust heap of history.

Some would claim that the Internet, the World Wide Web (WWW), or computing have changed history, and these have left their marks on life, but nothing has become as pervasive and as impactful as the smart phone.  The Internet, the WWW, and computing have contributed to the march of the smart phone, but so have many other technologies such as NFT, Bluetooth, and Wifi, but the smart phone is the agent of change.

In my smart phone, I can carry my wallet (except for cash, and even Venmo would object to that exception), my keys, my identification, my theater tickets, my shopping list, my maps, and the only things I need carry now are a pocket knife and a handkerchief.  Most people do not even carry handkerchiefs anymore.  And all this fits in one pocket.

We do not always recognize the revolution when it starts; in fact, we rarely do.  Transistors were big, chunky things with poor gain (a measure of transistor quality).  Horseless carriages were hard to start and broke down easily - and they needed roads to get somewhere.  Steam trains were noisy and smelly, too.  Medicine was glorified speculation until germ theory.  

The transistor continues to change our modern world, but the smart phone has surpassed it.

Bad Reporting #1: Radio Tower Stolen - 9 February 2024

No.  One does not just steal a 200-foot tall radio tower.  No.  There are two major problems with this simple-minded assertion, size and power.

The obvious problem is the size of the 200-foot tower.  It takes time and equipment to down it and haul it away.  You need a couple of hours to rig it, to lower it to the ground, to dismantle it, and then to haul it away.  The word "stolen" implies surprise or stealth.  The radio station had hours to respond to any surprise attempt to take the tower.  The police had hours to respond when called.  This tower was not stolen.  

The other problem is the electrical power being pumped into the antenna.  There are thousands of watts of power being pumped into the antenna in order to broadcast, perhaps as high as 50,000 watts, but likely less in this case.  When you walk up to the tower and touch it, you become the path to ground.  Big shock - literally.  Someone had to have the smarts to cut off the transmitter before anyone touched the tower.  Yes, the report talks about evidence of a break-in, suggesting that the thief did have the requisite smarts.  But radio stations monitor their signal - they listen to themselves to make sure they are still transmitting.  Again, no surprise is possible when the transmitter gets cut and the antenna gets "stolen".

Reporters really need to pause and think before they report this stuff.

Source: Alabama station in disbelief after 200-foot radio tower stolen at NBC News.

Career Notes #4: WFH and Remote Work - 30 January 2024

Someone said something on the radio and I am upset.  There is probably only one motive more stupid to write about than "someone said something on the Internet and I am upset", and that is the subject of my first sentence.  Someone said something stupid on the radio and I am upset.  However, I have always barged in where angels fear to tread, so I will share my thoughts on the recent resurgence of the RTO push for workers to return-to-office.  Since shortly after COVID began, there has been a vibrant discussion about the merits of working in an office compared to working from home (WFH).  It is the fodder for serious commentators as well as for the comics.  After hearing someone potificate on the radio this morning about managers who will have to learn how to support and encourage people who are WFH, I rather got to the end of my tether and -- I have thoughts.  I should say up-front that I retired as a senior manager in engineering at a high-tech company and I managed mostly highly educated people (nearly all had PhD degrees from major universities in the US).  These were (are) bright, educated, talented, and highly  motivated people eager to advance their careers and advance the fields of engineering.

When COVID began, there was a medical reason for everyone to avoid crowds and that included avoiding the shared office.  When I decided to close our local ABC Corporate office (a branch office of about 100 people and anonymized), I expected that it would be for 2-3 weeks until the pandemic abated.  I think it was pretty commonly understood that closing offices and sending people home was temporary and we would return to normalcy in a month or so.  As time went on and people continued to die from COVID, it became clear that our initial estimates were naively optimistic and that WFH (work from home) was a long-term state, verging on a year.  

While a high proportion of ABC Corporate workers could be sent home to work, there were key jobs that needed to continue on-site and in the office.  In our branch location, there were about 10 employees that needed to be in the office daily and 2-3 more that would come in occasionally.  The 10 folks needed access to labs and equipment that could only be on-site and could not be taken home.  The 2-3 folks varied each day from the 90 who were WFH - they were not usually the same individuals, but there were consistently 2-3 of them in the office.  

In our group (about a third of the 100 people in the branch), we had long-standing key metrics (KPI - Key Performance Indicator - is a common industry buzzword, but we did not use the phrase KPI).  The primary metric among these was the continued delivery of high-quality results as promised in the schedules and contracts.  As a general rule, the team was able to maintain successful delivery of all the group metrics.  From this, one might conclude that WFH was acceptable as a stable, long term strategy for the team.  After all, the overall team was spread across four primary sites already, so dispersing to homes was of marginal impact, right?  Wrong, but I have gone on too long in this posting and will reserve further thoughts for a second posting.

Although the key metrics, including consistent, timely, high-quality deliverables continued to be met, these are all "hard" metrics.  These are hard metrics because they are countable.  The deliverable was provided or it was not; the deliverable was on time or it was not; and the customer accepted the deliverable or they rejected it.  However, there were other, less countable metrics that were not always met, and many of these "soft" metrics are also on a scale, but hard to quantify.  Without covering all of the next post, consider "innovation".  There is no good measure of innovation.  There are simple measures (e.g., the number of patents authored) of these soft metrics, but simple measures can easily miscount or mislead (e.g., an idea can be patentable but it may be kept as a trade secret instead).  A more critical measure is readiness for promotion or other recognition, which is very hard to measure.

Until next time when I discuss soft metrics.

Innovation keeps rolling, Part 2 - 18 January 2024

Previously, I wrote about a patent idea I had at IBM that had been declined by the internal IBM patent review committee, and that a company had recently announced binoculars that used a strongly similar idea.  Let me return to the more general intellectual property (IP) arena in this note.

The success of new companies of the world is highly dependent on diligently and daily applying innovation to their products and processes.  Bookstores were simple repositories of books until Amazon unleashed technology on the selling and shipping of books and more.  Cars were glorified wristwatches until hybrids and then full electric vehicles (EVs) caused a reinvention of the automobile.  Even companies like Intel were glass manufacturers until ARM, Apple, Nvidia, and AMD showed that fabless semiconductor companies could be a thing.  You can debate specific examples, but the clear trend of business from 1924 to 2024 shows that those failing to apply technology have been left as footnotes; at best, they are quaint companies turning out homespun crafts.  

In recognition of this innovation force, companies set up internal procedures to generate patents.  IBM is probably the cannonical example.  IBM has long had an internal process that collected IP ideas, evaluated them through a committee process, and awarded IP "points" to inventors on patent filings that could be turned into cash and prizes.  (No kidding - my prize for one set of points was luggage.)  Little mafias and cliques grew up within IBM to keep churning out patents.  If your idea had the right co-inventors, you got points.  Even ideas that did not deserve filing were "published" and the authors got IP points as a result.  Perverted results aside, the system worked well, and IBM was the largest patent producer in the world for decades.  The innovations of IBM were world-class as demonstrated in their products, and IBM ruled the computer world.  Somewhere along the line of history, IBM reduced their emphasis on IP and technological excellence and pushed financial engineering.  While other companies continued to grow, IBM sold off parts of the company and shrank.  As I write this, the market capitalization of IBM is about $150 billion, while Apple and Microsoft are each (each!) $2.9 trillion.  

In the last few days, IBM has announced that they are terminating the old program  (that would issue cash and prizes) for a new program that issues BluePoints.  It is unclear what one can do with BluePoints, but cash awards for accumulating points seems to be elimiated.  Combined with other changes over the years, it is not clear that IBM has figured out how to exit their death spiral.  Cold days ahead for innovation at IBM.  Sad. 

Innovation keeps rolling - 15 January 2024

Employed by IBM as an engineer means that there will be continuous opportunity and pressure to create intellectual property, specifically patents.  I was exposed to this during my time there from 2001-2005.  It was one of the better parts of the job.  Engineers would write up an internal document that proposed and described a patent idea.  These proposals would be reviewed by teams of senior engineers under the guidance of an internal patent lawyer (external patent lawyers would be used for the process of drafting and submitting the patents, and the internal lawyers would oversee them).  

Among the patent ideas I wrote in about 2003 or so was the idea that a camera should combine GPS information with compass readings and focus data to determine what someone was photographing.  If you are standing in a particular location (GPS data) and pointing your camera in a particular direction with a focus point about 200 yards away, then you are probably taking a photo of the Eiffel Tower.  And so on.  Clearly this did not work for photos of people or pets, but you would at least be able to say that the Eiffel Tower was in the background of your photo of little Fee-Fee.  This was in the days before AI - long before something as small as a camera could carry the training set or network of a large language model (LLM-AI).  

Sadly, the idea was denied by the local patent team.  They felt it was too easy to fake using false GPS transmitters and the like (as if anyone wanted to disrupt your photos of Fee-Fee such that they would spend thousands of dollars and risk arrest to disrupt your GPS signals.  But I digress.)

Recently, Swarovski has announced new binoculars that will identify birds.  The Swarovski product is a variant of my idea - the one I could not patent.  (Viz., to use geospatial and image data to deduce situational information when taking or viewing an image.)  I wish them well.

Because we do not already have enough nuclear waste dumps - 14 January 2024

Nuclear power from fission, good old Uranium-powered sources, have an on-going level of support that continues to amaze me.  I was previously a supporter of nuclear power, so perhaps my perspective is more intense than it might otherwise be.  This is not an unusual phenomenon;  I know ex-smokers who are vibrantly anti-smoking, more against smoking than people who have never smoked.  But, as an engineer, I try to be practical.  As a modern engineer, I try to understand the full life-cycle of an idea.  And nuclear power may succeed as carbon-free in the short term, but it fails in the long-term safety of the by-products.  We have failed to solve the disposal problem of nuclear power.  We do not know what to do with the continuous stream of low-level waste coming from nuclear power plants - as far as I can tell, we just bury the stuff.  We do not know what to do with the regular pulses of high-level waste coming from nuclear power plants - we just store it in swimming pools and post guards.  That is about the best we can do.  Yucca Mountain failed totally; political or technological, it was a total failure.  There are attempts to reprocess nuclear waste, but those processes create new nuclear waste, albeit more concentrated.  And, go ahead, tell me how concentrated nuclear waste is better than the regular flavor.  Yes, it can create new nuclear fuels, but those will eventually become high-level waste, so it is but a temporary solution.  Finally, the Swedes (I think) are burying the stuff, but, again, that is only temporary.  

Ultimately, nuclear power is a poisonous gift to our descendants, not a solution to anything.

And when the supporters are done arguing (fruitlessly) against these arguments, nuclear power is expensive.  Solar and wind are already cheaper, so nuclear has lost the war.  

We do have a continuing nuclear-waste problem with smoke detectors.  They contain a small amount of Americium (I think it is) to generate ions that help detect the smoke.  Instead of being properly disposed of, instead of being recycled, the bulk of these units are simply discarded in household waste.  That means all of our garbage dumps are becoming nuclear waste dumps.  Clever that.

Recently, some researchers in China announced the development of a 50-year battery...that is nuclear powered.  Oh, joy.  In the article,  Chinese-developed nuclear battery has a 50-year lifespan , the researchers use an isotope of Nickel to generate power for small devices (e.g., phones).  The article reports that "Betavolt says its nuclear battery will target aerospace, AI devices, medical, MEMS systems, intelligent sensors, small drones, and robots – and may eventually mean manufacturers can sell smartphones that never need charging."  Even though the batteries may last 50 years, the small devices will only last a couple years (I found numerous Google articles that quoted 2-3 years based on actual studies).  Let us assume the 50-yr batteries will double the lifetime of the average cell phone to six years.  That still means that millions of little radioactive devices will be discarded every year; more likely tens of millions or more.  All of those devices will be dumped into landfills and garbage heaps to add to the radioactivity of the smoke detectors.  

That is a lovely gift for our descendants.  Lovely.

Never Underestimate the Power of a Demo - 9 January 2024

Early in my career, I worked for a guy named Dave Harms.  Dave supervised a hardware development group at Bell Labs and he was energetic and inspiring.  Furthermore, Dave was a builder and the group was designing a laptop.  We did not really have that word "laptop" at the time as all personal computers and workstations were desktops or "luggables", but Dave was not content with those form factors.  He wanted a laptop built around our new Mac-32 chip.

To illustrate the design concept, Dave went into the lab with the parts list from the design.  He grabbed a printed circuit board that was about the right size; the standard printed circuit board used in the Bell System at the time was about 8x10 inches (roughly) and all the possible hole positions were drilled out.  (A key challenge during board layout was to find enough room to run the signal paths between all the holes.)  Then Dave went into the parts inventory room and picked out parts that were the right size as called out on the parts list.  These were not the right parts, they were just the right sizes - 14-pin, 16-pin, 24-pin, and so on, as described on the parts list.  He put the parts into the printed circuit board (remember that all the holes were already drilled), assembling a simula of the ultimate design.  He bent the pins a bit so that they would stay in place in the board but still be usable on other projects.  Then Dave went to a meeting to propose the project.

Dave started out with the usual slides.  At the time, these were all hand drawn, but he went through the various points for the design.  The size, who would buy it, the computing and display capabilities, and so on.  People were interested but not very.  They were polite.

Then Dave pulled out his "demo" board and passed it around.  The room was electrified.

When people could "see" the "actual" board, not just slides and talk, but a physical mockup, they went from polite interest in the idea to supporting the idea.  It was like a switch was thrown and the lights came on.

The lesson from this experience was clear to me and I carried it forward.  Never underestimate the power of a demonstration.  People want to put their hands on the idea, and a demo can give that to them.

ETA: What triggered this little story?  A recent news article in FastCompany reports that "recently, Microsoft built a clock."  A research group in the Microsoft Quantum department went out and bought an off-the-shelf clock and "dressed up its enclosure by adding the logo of Azure Quantum Elements".  And...

The point of this little DIY project was to prove the batteries worked in a visceral way: "You want to have a wow moment," explains Brian Bilodeau, the head of partnerships, strategy, and operations for Azure Quantum. And the person the quantum team hoped to wow was Microsoft CEO Satya Nadella. -- FastCompany article.

So there you have it.  The research team wanted to put the batteries on display, so they built a demo using a cheap clock and some artwork, then presented it to the CEO of the corporation in order to get his attention and support.  The Power of a Demo, right there.

Security is the responsibility of the vendor - 6 January 2024

23andMe is in the headlines because they suffered a security break-in that revealed private information for their users.  23andMe has taken to the headlines to claim that their users are the reason, the cause of the failure.  In short, 23andMe claims that users reused their passwords from other sites, thus enabling hackers to break in.  

This is stupid.  23andMe is responsible, not their users.

You do not believe me?  23andMe cannot be that stupid?  Then follow this link from Wired magazine with the headline, "23andMe Blames Users for Recent Data Breach as It's Hit With Lawsuits".  Or google for your own sources.  

I do not know why I have to say this, but if you (corporate or person) collect information from me for a specific purpose, then you have the responsibility to protect that information from other users and hackers, and you have the further responsibility to use that information only for the purposes under which it was collected.  

I protect my personal information at home and I surrender that information only for purposes that I choose.  I expect you to protect my personal information in exactly the same way.  This is made explicit in HIPAA laws in the United Stated and in personal privacy regulations elsewhere (e.g., GPDR laws in the EU).  Should you fail to protect my information, you are subject to liabilities and consequences.  

Furthermore, I provide my information only for specific purposes (again, see HIPAA and GPDR for examples), and you need to limit yourself to those purposes.

Companies and individuals set up security protections for these very reasons.  Everything from firewalls to encryption and more are technologies for security and protection of systems and information.  For 23andMe to turn around and blame users for any data breaches is absurd.  23andMe bears full responsibility.

Why do we have to say these things?  Are there no adults running 23andMe, no adults who can guide the rest of their company to do the right things?  

Rant over.  The photo is of a 10th-11th Century castle built for protection in Portugal.  

 

Musing on Generative AI (ChatGPT, Bard) - 4 January 2023

Generative Artificial Intelligence (Generative AI) has hit the big time.  As evidenced by ChatGPT by OpenAI or Bard by Google, the massive growth of available compute resources has allowed large language models (LLMs) to start working their way into daily use.  Some people use LLMs to generate product descriptions or speeches, some publications use LLMs to generate articles, and others are using LLMs to create novel images and movies.  The Internet is being slowly poisoned by the creeping influence of LLM-generate data.

Leading commentators are claiming this is the end of work.  Everyone from journalists to artists to writers to physicians will be replaced by LLM-driven AI.  In my professional area, people will describe what they want and AI will write the code, displacing millions of programmers.  Poof!  On the dole they go by the millions.  

This is silly and people should know better.  To borrow a concept from (I think) the Gartner Group (GG) consultanting group, LLMs are approaching the Peak of Inflated Expectations in the GG Hype Cycle, and we will soon see the Trough of Disillusionment.  For more details, the reference is in the Gartner Group Hype Cycle chart.  How can I, a mere pleb, a common person, claim this to contradict the great Gartner Group geniuses and the wisdom of the great commentators of our time?

One word:  Hallucinations.  The LLMs we use today commonly inject massive errors into their results, errors that are commonly called hallunications in polite company, and 🐄💩 in more direct company.  You can feed good stuff into an LLM and get garbage out (aside, the new meaning of GIGO for AI is Good Input Garbage Output).  So hiring poor programmers or non-programmers to run your great LLM Programming Machines will tend to generate garbage that only experienced programmers will be able to detect and fix.  Programmers are not going away, at least not because of LLMs.

Goals without plans are just wishes, or not - 1 January 2024

Annual cycles are common in nature.  The migration of whales and birds.  The leaves that drop in the Fall and come anew in the Spring.  Snow skiing gives way to water skiing gives way to snow skiing.  Corporations across America have employees complete a self-assessment of the prior year and start to write goals for the coming year.  In many situations, at the beginning of a new cycle, we are encouraged to think of goals for the coming year and we are reminded that "goals without plans are just wishes", but I choose to challenge this supposed wisdom.

We are told - without proof - that goals can only be achieved by careful planning so that we can follow that plan to success.  Random motion can only lead to stagnancy or regression.  If the drunk starts from a lamppost and proceeds along a random walk, they can only end underneath the same lamppost.  But this is not so.  Science tells us this is silly.  Practicality tells us not to expect to follow a plan.  Military strategists tell us that a plan is perfect until the first shot is fired.  Planning has its place, but it is far from a guarantee.  Let us explore a bit.

It is a common joke in research science, mathematics, and engineering that the most interesting results come when someone says, "Huh, what's that?" when they notice an anomalous result.  Clearly, science and engineering try to create predictable results most of the time.  However, it is not the expected results that are interesting, but the unexpeceted results that lead to new questions, new answers, and new insights.  Thus, one can plan a series of experiments or design stages, and they will often play out, but the results are not always very interesting.  When the results stray from the plan, they create moments of insight - aha!  This does not mean that someone should head off with no ideas, skip over some degree of planning - afterall, one has to assemble the right equipment to make any progress - but the planning does not ensure success.  In fact, an interesting result will often cause the investigator to toss the plan and follow the new results to uncover the insight.

Similar tales abound in the military world.  All the soldiers line up and go at the enemy, but most plans are quickly replaced with improvisation after the first engagements.  In fact, the enemy will put great effort into creating surprises to distrupt the plans of the opponent.  Yes, this is planning, but it is also likely to be quickly tossed if the results are not as desired.  

In short, make a plan, but be prepared to ditch it.  the plan is not required for success nor does it provide any assurances of success, even petty success.  So where does that leave us with goals?

Goals should describe the desired results.  The goal should describe a faster process, a cheaper process, a better position, and may identify some of the steps to get there.  There are many guidelines about how to write good goals - "S.M.A.R.T." is common and well-know - but nowhere is there any requirement for a plan in the goals.  Aha!, you say, that is because goals and plans are separate things! and I will agree.  That is precisely why goals without plans are still goals, not wishes.

Excel and Big Decisions - 13 October 2023

Friday the Thirteenth is famous for its association with bad luck.  That makes today a fitting day to explore the use of Excel in corporations.

In a recent report, mistakes in Excel programming caused major, embarassing errors in hiring doctors in the UK (reference at the end).  The headline screams "Excel recruitment time bomb makes top trainee doctors 'unappointable'" and the sub-head explains that "Mangled mismatch of formats, macros, and VLOOKUP practice hits wannabe anesthetists".  Although this particular failure is significant to those affected, there is a larger problem that affects us all.  We start from two facts.

One.  Microsoft Excel spreadsheets are nearly impossible to debug, thus they contain numerous errors.

Two.  Microsoft Excel spreadsheets are used to make key decisions in nearly all businesses and institutions.

When we combine these, it is easy to see that key decisions in business and institutions are made based on bad data.  These include billion-dollar decisions as well as smaller decisions in hiring and promotion.  I spent years building budget and planning spreadsheets on behalf of my boss, often working with another, more skilled Excel practitioner and budgetmaster (hi, Nick!).  After many, many errors, we taught each other to put debug checks into our spreadsheets.  Not only did we sum across the matrix, but we summed down the matrix and compared the results.  As one example, we would build large matrices of spending or staffing numbers, and if the result of sum-across did not equal the result of sum-down, a cell in the sheet would turn red with a warning.  We were careful to cut-and-paste links rather then values, so that if the original values changed, the links would update.  (This cut-and-paste is a manual operation, so it was subject to errors, but we tried.)  These sheets would be used to create plans for hiring of staff and interns for the coming year, a critical decision that could cause us to fail to me business objectives if the numbers were wrong.  If our budget was too low, we might lack the staff necessary to do the work; if our budget were too high, we would overspend (and no one ever got Executive sympathy for overspending).  

In this particular report (from The Register), some bad hiring decisions were made for doctors in the UK, but we all know that mergers and acquisitions are decided based on Excel calculations.  M&A can be measured in billions of dollars.

To be fair, Excel, itself, is not the direct cause of the problem.  Excel is merely doing what the programmers are telling it to do.  Excel will happily sum 11 months of costs and report the sum to a reader expecting to see 12 months of costs.  And so on.  The fault is that Excel is not designed to allow programmers to detect and fix errors.  Further, most of the "programmers" are business people, not trained computer programmers, and so they lack many of the programming disciplines required to produce reliable, usable code (Excel macros, in this case).  

To fix this problem, Microsoft needs to add checking and debugging features to Excel and the Excel programming community needs to learn to use them.  I do not expect this to happen any time soon.

https://www.theregister.com/2023/10/12/excel_anesthetist_recruitment_blunder/?td=rt-3a

Twitter shitter - 16 Aug 2023

Absent from Blogger for a while, a couple recent events have forced me back to record the progress of Twitter-X as it swirls ever downward.

Earlier this week, it was announced that Twitter-X had resisted federal supoena efforts to obtain information about Trump's tweets-xeets.  This was evidently an effort by Elon Musk to protect TFG from Jack Smith's investigations.  

Also earlier this week, it was announced that the Twitter-X URL shortener (t.co or something similar) was injecting delays of 5 seconds when redirecting to sites that were not in Elon Musk's good graces, including newspapers that had reported unflatteringly about Elon.  Shortly after this was documented and reported, the delay went away.

And just today, it was reported that a podcaster and professor at NYU, Scott Galloway, had been locked out of his Twitter-X account after he released a podcast unflattering to Elon Musk.  Evidently Galloway said something about unsubstantiated (inflated) range quotes from Tesla.

The only conclusion is that Elon Musk is a thin-skinned adolescent-minded twerp.

Photo is of blackberries coming into ripeness.

https://gizmodo.com/scott-galloway-locked-out-twitter-musk-1850743635  

Trouble in AI Land - 13 July 2023

Forgetting.  As I understand the European data rules, one has the right to be forgotten.  At first, this sounds odd.  Although it can be abused, it makes sense when you think of it.  If a mistaken or outrageous screed about you gets injected into the Internet, you should have the right to get it removed - to be forgotten.  Now think about AI engines.  

Similar to search engines, AI engines today will crawl the web looking for input materials.  No single AI company can generate the vast amount of input data needed for AI training, so they pretty much are left with crawling the web.  The information they find is incorporated into the training models and used in the "predictive" part when the AI model generates output.  It is likely that somewhere in the depths of ChatGPT that there is a copy of Moby Dick, Shakespeare, and your master's thesis, all wrapped up in "the model" and existing as a latent copy of your work.  

Now, suppose you wish for your thesis to be forgotten, removed from the Internet, relegated to nothingness.  You can get the data pulled from a search engine, but how does one get data pulled from an AI model?  The developers do not know how information is incorporated into the model (whle they know in vague terms, "the algorithm", they cannot trace any particular data item into the model).  I suppose they would have to retrain using the old input dataset but without your work, and that would be very expensive.  Training is often the most expensive part of developing an AI model.

To emphasize this problem, there is already a lawsuit by someone who claims that their copyrighted material has been included in an AI model without their consent (ref. Sarah Silverman).

So the AI folks are in a pickle.  

https://www.theregister.com/2023/07/13/ai_models_forgotten_data/?td=rt-3a