Microsoft Sees the Light

The Three Mile Island meltdown in 1979 single-handedly killed the nuclear power industry in the U.S. After Three Mile Island construction on nuclear power plants halted, and existing plants were decommissioned.

Today nuclear energy is undergoing something of a renaissance. Many green-energy advocates are realizing that ambitious carbon-free energy goals cannot be met by wind and solar power, let alone be "on all the time" when the sun isn't shining and the wind isn't blowing.

Many tech companies, including the largest and most well-known, joined the alternative energy marching band, IMHO, because it was easy to, since their fortunes weren't tied to fossil fuels like legacy manufacturing and transportation companies.

After 2020, as the race to acquire electricity for artificial-intelligence data centers accelerated, the virtue signalling dampened down.

Big Tech needed power, and ESG (environmental, social, governance) projects did nothing to help and even inhibited that objective.

Three Mile Island will be renamed the Crane Clean Energy Center (WSJ)

In a sign that the worm has turned, Three Mile Island, the bête noire of fission haters, is being revived to power Microsoft's data centers.

A deal between Constellation Energy and Microsoft will restart Pennsylvania’s Three Mile Island, the site of the country’s worst nuclear accident, to help power the tech giant’s growing artificial intelligence ambitions.

Under the agreement, Constellation would revive the plant’s undamaged reactor, which was too costly to run and closed in 2019, and sell the power to Microsoft. The plan signals the gargantuan amount of power needed for data centers for AI, along with the tech industry’s thirst for a carbon-free, round-the-clock electricity source needed to meet climate goals.

Constellation expects to spend around $1.6 billion to restart the reactor by early 2028. Microsoft has signed a 20-year power-purchase agreement with Constellation, the companies said Friday. The deal would help Microsoft pair its 24-7 electricity use with a matching source of nearby clean power generation.

It's nice to know that one of our most successful companies is able to discard green-power fantasies when the virtue-signalling impedes basic objectives.

When I was a child, I spake as a child, I understood as a child, I thought as a child: but when I became a man, I put away childish things.---1 Cor 13:11

It's Not the Same as Lying to a Real Person

(WSJ illustration)

Yes, I have lied, and continue to lie, to websites. Like most others, I do it for my own protection. [bold added]

When you sign up with a social network, e-tailer or other service, you often have to give your name, email or phone number, and birthday. Some sites ask for a street address and more personal or private items, such as gender or mother’s maiden name.

The demands can feel intrusive, and giving sites all that information means that hackers could get access to it. So, a lot of people just sidestep the issue: They lie. When websites ask for information, they make up names, birthdays, street addresses and anything else you can think of.

“It’s a common technique for individuals to use false or pseudonymous contact information and birthdays and other personal details,” says John Davisson, senior counsel and director of litigation at the nonprofit Electronic Privacy Information Center. “So that in the event of a breach, or in the event that the company you’re providing information to wants to use it in some way, like targeted advertising or to sell to a data broker, you’re protected because the information is not high-fidelity information and not going to be of use.”

I give restaurant and retail websites fake identifying data whenever possible because the information doesn't seem crucial to the business that I want to do with them. If they catch me and kick me off, well, I can live without their product or service.

At financial institutions I have to be truthful in order to access the accounts. Although they have much more security than websites in other industries, they're not impregnable; therefore I check the activity and balances once a month.

The virtual, online world has brought many conveniences that also carry new risks. Only give out the minimum information necessary; websites are not your friend.

Apple: the Primacy of Privacy

AAPL sold off 2% today

After its run-up to the Worldwide Developers Conference, Apple stock sold off 2% today, as traders greeted Apple's introduction of its artificial-intelligence products with a yawn.

Apple said its new software will retrieve information from across apps and scan personal information to help users proofread text, call up photographs of specific family members or gauge traffic patterns ahead of an atypical commute. Users can create images and emojis and even convert rough sketches into polished diagrams...

For certain complex Siri requests, Apple will surface a prompt to ask if the user wants to connect with ChatGPT to get a better answer. Apple is also allowing ChatGPT to connect with other areas of the operating system, such as using the AI to help with composing text. Apple said it gives users the ability to control when and if they want to use ChatGPT...

Apple's AI features didn't overtake the competition technically, but as an Apple customer I really liked the emphasis on privacy. (By way of contrast Microsoft Windows' AI tool "Recall" takes snapshots of users' screens every five seconds and stores the images on an unencrypted database that is easily hacked.) [bold added]

Privacy is at the heart of Apple’s new AI capabilities, a feature that could further lock users into its ecosystem. Most processing will be done on a device instead of shipping to the servers in the cloud. But the company said for running larger AI models Apple will keep it private by running its own servers with what it calls Private Cloud Compute. It will only send data relevant to the task to these servers. The data isn’t stored or accessible by Apple for further training, the company said.

ChatGPT queries from Apple devices will be routed to OpenAI servers, but user information won’t be shared with OpenAI, and the startup won’t be able to identify the queries from specific Apple users. Apple will use a version of ChatGPT that is available free elsewhere online, but those with premium ChatGPT subscriptions will be able to link their accounts.

Users' data is stored on the device itself. If more processing power is needed, then tasks are diverted to Apple's private servers. Users can call on Open AI's ChatGPT, but even here protections are in place.

user information won’t be shared with OpenAI, and the startup won’t be able to identify the queries from specific Apple users.

As an Apple investor and a customer with privacy concerns, I appreciate the fact that Apple will be using its own silicon and not be as beholden as everyone else to external data centers running on Nvidia chips.

This fall I'll be buying a new iPhone 16. My six-year-old iPhone XS Max has served me well, but it's time to move on to the brave new AI world (as well as a much better camera and battery).

I'm not so presumptuous as to believe that hundreds of millions of iPhone, Mac, and iPad customers think like me and will soon upgrade their hardware, but there's a good chance that will be true, so if I didn't already own Apple stock, I'd be buying some.

Like "Sanctuary" Cities, Easy to Proclaim Your Virtue Until You Have to Step Up

Data centers in northern Virginia (WSJ photo)

The tech giants all profess to be "green," but when the choice is between net-zero carbon emissions and the creation of power-hungry artificial-intelligence data centers, watch what they do, not what they say. [bold added]

An explosion of so-called hyperscale data centers in places such as Northern Virginia has upended plans by electric utilities to cut the use of fossil fuels. In some areas, that means burning coal for longer than planned.

These giant data centers will provide computing power needed for artificial intelligence. They are setting off a four-way battle among electric utilities trying to keep the lights on, tech companies that like to tout their climate credentials, consumers angry at rising electricity prices and regulators overseeing investments in the grid and trying to turn it green...

Many new data centers coming to Northern Virginia are known as hyperscale, or facilities that are far larger than previous generations of data centers. The big ones use as much power as the city of Seattle...

Wind and solar can’t serve data-center demand around the clock, so growth will need to be supplemented by natural-gas-fired power generation, said Arshad Mansoor, chief executive of the nonprofit Electric Power Research Institute.

“You can be an idealist,” Mansoor said. “But if you’re a realist, you’ll add a ton of solar and you can balance that with gas.” The only other option to new gas plants is delaying coal and nuclear-plant retirements, he said.

The moguls who run the mega-cap tech giants know very well that the artificial-intelligence gold rush will delay national decarbonisation for years. By talking a good environmental game (and deflecting blame onto the utilities) they hope to retain their ESG credentials and keep consumers and regulators off their backs.

Generative AI: Some Fruit are Poisoned

(WSJ image)

Data integrity, aka garbage-in, garbage-out, has been a problem with computing since its inception. With the rise of artificial intelligence and our projected future dependence on AI, the phenomenon has been given a frightening name: data poisoning. [bold added]

This is when malicious actors insert incorrect or misleading information into the data used to train an AI model with the aim of spreading misinformation, undermining the chatbot’s functionality or getting it to do something bad, such as share sensitive information.

While data poisoning is a concern with all types of machine-learning algorithms, some researchers say generative AI models could be particularly vulnerable because they must ingest vast amounts of text, imagery and other data from the public internet to gain the knowledge they need to create something on their own.

Researchers say this reliance on a vast number of data sources from the open web—rather than curated, locked-down data sets, which are harder for hackers to penetrate—could make it difficult to spot and eliminate poisoned data, only a small amount of which is needed to affect AI’s outputs.

In 2023 lawyers were caught citing fake cases in legal briefs produced by artificial intelligence (ChatGPT).

In recent months Google's Gemini image generator, which clearly had racial diversity as a bedrock principle, produced images that were the object of ridicule:

America's Founding Fathers, Vikings, and the Pope according to Google AI: pic.twitter.com/lw4aIKLwkp

— End Wokeness (@EndWokeness) February 21, 2024

This latter example is not so much "data poisoning" as it is a matter of algorithmic over-ride of good data.

Artificial intelligence is still in its early stages, but already it has given users reasons to distrust its output. The lazy or the indifferent won't bother with performing checks, so I'd bet that we never get rid of the poison in the system.

The Other Dismal Science

(Image from the balance)

I never gave a thought to actuarial science until I went to business school. One of the smartest guys in the class had been an actuary, which I came to understand was too tough a career choice if I had to be as smart as he was and pass difficult, technical exams. (BTW, he's now CFO of a well-known company.)

Digression: so I became a CPA, which meant sitting for two days of exams and working for two years in the audit department of a CPA firm. The latter path was easier, and besides, I've never had to pay anybody to do my taxes.

Apparently, becoming an actuary is even tougher these days: [bold added]

Among people taking at least one exam from the Society of Actuaries—the field’s biggest U.S. credentialing body—15% eventually pass the multiple tests required to become an Associate, one of two designations allowing them to practice. Just 10% pass those and additional tests to become a Fellow, the group’s higher designation, which affords bigger responsibilities and salaries.

It’s such an arduous process that the number of test-takers has been declining in recent years, and the society is making changes to keep candidates from dropping out of the gantlet. It is also adding new “predictive analytics” tests to adjust to the massive amounts of data insurers now have...

There is no limit to how many times a candidate can take the tests. It took one man 50 years to become a Fellow, says Stuart Klugman, an official at the society. The society says a candidate typically takes seven to 10 years to become a Fellow. They must pass 10 exams plus other coursework and requirements.

Every profession is being inundated with oceans of data, and being familiar with data science has become essential to being an actuary. Knowing how to filter, analyze, and model the data is crucial to success. (We noted the importance of predictive analytics three years ago.)

If one is not thorough, knowledgeable about statistics, and honest, one can easily promulgate misinformation by biasing the data sets, selecting or not selecting variables to be analyzed, finding causation in correlation, and not "showing the work" so that results can be replicated.

Why must an actuary be honest? Because as the analyses become more complex, and as the demand for risk assessment (not only for insurance purposes) explodes, there are fewer people who are available to check the work; actuaries must be trusted not to force through results that please the powers that be.

I was glad that I didn't try to become an actuary.

Just for fun: the article presents a sample question involving rudimentary statistics and algebra. Without using a calculator, I could still solve it. (My high school self could have figured it out much more speedily but would also have snickered throughout at the phrase "blue balls.")

“An urn contains 10 balls: 4 red and 6 blue. A second urn contains 16 red balls and an unknown number of blue balls. A single ball is drawn from each urn. The probability that both balls are the same color is 0.44. Calculate the number of blue balls in the second urn.”

Zillow, (Dis) Continued

(Photo from Star Tribune)/td>

Last week we opined that Zillow's house-flipping business failed because it placed too much faith in its computerized decision-making programs:

The algorithm needed to incorporate additional data--the kind of data that's hard to obtain--and Zillow needed to test the program on a much smaller scale before it bet the farm. They had the courage to pull the plug before it took down the company, and it's likely they'll be back some day, probably with a "money" partner that will provide the risk capital while Zillow provided the smarts.

A WSJ follow-up story confirms that the algorithm failed to take into account market nuances and adds that these flaws were exacerbated by the rapid shifts--both up and down--in real estate markets brought on by COVID-19: [bold added]

Computer-driven analysis has become mainstream in stock and bond markets, but buying and selling single-family homes has proved a trickier proposition. The real-estate market varies widely by city, region and type of property, with a range of aesthetic, social and other factors playing into Americans’ home-buying decisions.

Zillow also overstretched its staff as it tried to catch up to competitors and disregarded internal concerns that it was overpaying for homes, according to former and current employees. It operated in an unpredictable housing market, with the pandemic fallout helping to spark the biggest housing boom in a generation. And Zillow suffered from supply-chain and labor issues that slowed its ability to renovate homes quickly...

An unexpected surge in home prices and sales during the pandemic made it harder to predict the market. Buyers began giving priority to space and location in unusual ways.

“That shift in buyer preferences is extremely hard for a machine-learning model to incorporate,” said Dave Meyer, vice president of data and analytics at BiggerPockets, a real-estate investing website.

The final nail in the coffin was trying to catch up to the competition just as the market was turning:

Zillow put together a plan to speed up the pace and volume of home purchases, dubbing it Project Ketchup—which employees took as a play on the team’s mission to catch up to Opendoor. Zillow planned to buy more homes by spending more money, offering prices well above what its algorithm and analysts picked as market value, people familiar with the matter said.

In the second quarter, Zillow Offers bought more than 3,800 homes—more than double the previous quarter. In the third quarter, it bought 9,680 homes. The company was buying so many homes that its overstretched staff started running behind on closings and renovations, people familiar with the matter said.

It struggled to find contractors and renovation materials amid a broader labor and supply shortage. That meant Zillow was in danger of sitting on homes for longer, adding to insurance and debt bills. It also meant many homes bought during the summer would likely have to be sold in the winter, when the housing market is usually weaker.

As for damage to the overall company,

The company’s market cap, which closed at a peak of $48.35 billion in February, is now around $16 billion.

Their misguided efforts to improve matters led to a disaster in less than a year. Fortunately Zillow had the courage to face up to its mistake and change course before it lost everything.

Zillow: Too Early, But They'll be Back

(Photo from Star Tribune)/td>

We've been regular users of Zillow since 2016. It has a trove of real estate information, useful for checking on the value of one's own home, the cost of homes in areas that we might want to move to, or market rents in every zip code.

However, one would be crazy to make a purchasing decision based solely on Zillow's algorithms. One can't know the "feel" of a house or a neighborhood, or whether the roof might leak or the pipes need replacing, without a viewing by a real human being and/or inspector.

Zillow thought that its valuation algorithm could be relied on to buy houses en masse to build a house-flipping enterprise. Now it's getting out of the business.

Zillow Group Inc. reached a deal to sell about 2,000 homes from its ill-fated house-flipping program, the company’s biggest bulk sale as it starts unloading thousands of homes and terminates the business...

Zillow’s home-flipping practice involved buying homes, lightly renovating them and then selling them quickly, making money on transaction fees and home-price appreciation. The company and other iBuyers used an algorithm to make home-price estimates and determine what to pay home sellers.

Zillow said last week that it was shutting down the business because it couldn’t accurately predict future home prices and was losing too much money. The company expects to record losses of more than $500 million from home-flipping by the end of this year and is laying off a quarter of its staff.

I confess to feeling a bit of schadenfreude that a successful (up to now) tech company got a little too big for its britches, but I think they were on the right track.

The algorithm needed to incorporate additional data--the kind of data that's hard to obtain--and Zillow needed to test the program on a much smaller scale before it bet the farm. They had the courage to pull the plug before it took down the company, and it's likely they'll be back some day, probably with a "money" partner that will provide the risk capital while Zillow provided the smarts.

Note: here we know all about being too early. Foster City was the headquarters of Webvan, the highly-touted grocery delivery business that failed nearly 20 years ago.

Nobel Economists: They Opened Their Eyes

(From nobelprize.org)

Economics has been nicknamed the "dismal science", but this year's Nobel Prize winners have not only published insightful studies with real-world policy implications but also have injected creativity into a profession which seemed to have been destined to a future of endless number-crunching. [bold added]

The Nobel Memorial Prize in Economic Sciences was awarded Monday to David Card of the University of California, Berkeley, Joshua Angrist of the Massachusetts Institute of Technology and Guido Imbens of Stanford University.

The prize committee cited Mr. Card’s “empirical contributions to labor economics” and praised Messrs. Angrist and Imbens for “their methodological contributions to the analysis of causal relationships.” Messrs. Card and Angrist have contributed by running “natural experiments” in economics, and Messrs. Angrist and Imbens have contributed to the theoretical basis for these experiments.

Natural experiments have led economists to conclude that an additional year of education substantially raises one’s lifetime income, that small increases in the minimum wage don’t have much effect on employment, and that Medicaid has little effect on people’s physical health.

Why do natural experiments matter? One of the toughest problems in economic research is figuring out whether a relationship between two variables is causal or coincidental. So, for example, economists find that the lifetime earnings of people who go to school for 12 years are higher than those of people who go to school for 11 years. But what if those who stay in school longer are more motivated or smarter than those who are in for only 11 years? Then the earnings of the more-schooled would be higher even if schooling per se doesn’t add much to earnings. What one would ideally like is to compare the earnings of people whose motivations and intelligence don’t differ.

Enter compulsory schooling. In 1991, Mr. Angrist and the late Alan Krueger noted that under compulsory-schooling laws, students born in the first quarter of the calendar year would be able to leave school earlier than students born in the fourth quarter. Sure enough, they found, those born in the fourth quarter had an average of 0.15 year more in school. And the earnings of those in the fourth quarter were 1.4% higher than the earnings of those born in the first quarter. Extrapolate that to a full-year difference in schooling, and you can conclude that one extra year of schooling raises earnings by about 9%.

Messrs. Card and Krueger conducted a famous natural experiment by studying employment at fast-food restaurants in New Jersey and Pennsylvania before and after New Jersey raised the minimum wage while Pennsylvania didn’t. Contrary to what one might expect, employment in New Jersey’s fast-food restaurants rose slightly relative to employment in Pennsylvania’s.

Economist Alex Tabarrok discusses why the approach used to answer the minimum-wage question was so brilliant:

The obvious way to estimate the effect of the minimum wage is to look at the difference in employment in fast food restaurants before and after the law went into effect. But other things are changing through time so circa 1992 the standard approach was to “control for” other variables by also including in the statistical analysis factors such as the state of the economy. Include enough control variables, so the reasoning went, and you would uncover the true effect of the minimum wage. Card and Krueger did something different, they turned to a control group.

Pennsylvania didn’t pass a minimum wage law in 1992 but it’s close to New Jersey so Card and Kruger reasoned that whatever other factors were affecting New Jersey fast food restaurants would very likely also influence Pennsylvania fast food restaurants. The state of the economy, for example, would likely have a similar effect on demand for fast food in NJ as in PA as would say the weather. In fact, the argument extends to just about any other factor that one might imagine including demographics, changes in tastes and changes in supply costs. The standard approach circa 1992 of “controlling for” other variables requires, at the very least, that we know what other variables are important. But by using a control group, we don’t need to know what the other variables are only that whatever they are they are likely to influence NJ and PA fast food restaurants similarly. Put differently NJ and PA are similar so what happened in PA is a good estimate of what would have happened in NJ had NJ not passed the minimum wage.

Thus Card and Kruger estimated the effect of the minimum wage in New Jersey by calculating the difference in employment in NJ before and after the law and then subtracting the difference in employment in PA before and after the law. Hence the term difference in differences. By subtracting the PA difference (i.e. what would have happened in NJ if the law had not been passed) from the NJ difference (what actually happened) we are left with the effect of the minimum wage. Brilliant!

Indentifying control variables is tricky, as Prof. Tabarrok points out, and the more control variables one has the more statistical noise is created, and the less powerful the results. With a control group, researchers need not be concerned with control variables.

The Nobel winners aren't superhuman, however. The minimum-wage study likely wasn't error-free:

Unfortunately, Messrs. Card and Krueger’s data weren’t so great—they gathered it by phoning restaurants.

Nevertheless, once the idea was originated, the data could be obtained and analyzed.

At the dawn of the age of artificial intelligence, Professors Card, Angrist, and Imbens, and the late Alan Krueger, have given us hope that human ingenuity is not dead:

Card and Krueger revealed to economists that there were natural experiments with plausible treatment and control groups all around us, if only we had the creativity to see them. The last thirty years of empirical economics has been the result of economists opening their eyes to the natural experiments all around them.

Acoustic Epidemiology

(WSJ Illustration)

Like my pediatrician nearly 70 years ago, the 21st century doctor still places a stethoscope in several spots on the chest and back, asks me to breathe deeply then normally, and pronounces everything to be fine.

It's familiarly reassuring, yet a little disturbing, that almost everything has changed in medicine except the ritual of interpreting the sounds of respiration through a stethoscope...until now. [bold added]

Researchers around the world are trying to turn the humble cough into an inexpensive tool to diagnose and stop respiratory-disease killers like tuberculosis and Covid-19. They’re collecting recordings of millions of the explosive sounds from patients and consumers on smartphones and other devices. And they’re training artificial intelligence to find patterns to try to identify the type and severity of disease from the cough itself.

“We call it acoustic epidemiology,” says Peter Small, a tuberculosis expert and chief medical officer of Hyfe Inc., a Delaware-based company with two free smartphone apps—one for consumers, another for researchers—that use AI to detect and track how often someone coughs.

The sound and frequency of coughs are rich with medical information, he says. Different diseases have some audible differences: crackling in parts of the lung for pneumonia, a wheezing sound for asthma. Makers of these apps say there are sounds and patterns that AI can detect, but the human ear can’t hear.

The smartphone revolution has allowed researchers to amass databases of respiratory sounds and correlate them with various medical conditions.

ResApp Health Ltd. is using an explosion in telehealth services, particularly during the pandemic, to expand use of an app-based test for cough sounds that helps doctors diagnose diseases including COPD, pneumonia, asthma and bronchitis, says Tony Keating, the Brisbane, Australia-based company’s chief executive. A telehealth provider asks a patient to hold a smartphone at arm’s length and record five coughs. The app analyzes the coughs, then sends the results to the doctor.

Using technology developed by researchers at the University of Queensland, the company built the diagnostic tool by training an algorithm on recordings of 6,000 coughs, along with clinical data from those patients in the U.S. and Australia, Mr. Keating says.

Acoustic epidemiology is in its early stages. No expert would posit that a smartphone app has risen to the level of a medical-grade diagnostic tool, but its value as an initial screening device seems undisputable.

Your humble blogger believes that cough-analysis would be a logical next step for Apple's expansion into health care, but at first blush implementation seems tricky: coughing regularly into one's watch or phone is socially dubious, and putting on a sensing device like a smart bra or undershirt every morning seems like too much trouble. Maybe a famous-designer T-shirt? I'm sure they'll think of something.

Comforted by Limitation

Nobel Prize-winning physicist Frank Wilczek reflects upon the "deeply strange" nature of carbon- or radioactive-dating. [bold added]

What makes radioactive nuclei such ideal clocks is that they are reliably unreliable. An isotope’s half-life can be determined accurately by observing lots of decays. For instance, radioactive carbon, which is used to date organic material, has a half-life of about 5,700 years. But it’s impossible to predict when any individual nucleus will decay. In fact, an individual nucleus is a kind of anti-clock: It does not register the passage of time at all. There is no observable difference between old and young nuclei.They remain ideally young, we might say, until they suddenly and explosively die. By monitoring decays within this homogeneous population we measure time statistically, with confidence.

(Graphic from toppr)

In other words scientists can very accurately predict the behavior of a large group of atoms but cannot tell how a single atom will behave.

Of course, it's very easy to analogize this principle of quantum physics to the science of human behavior.

The study of the behavior of groups and whole societies has made significant advancements, but the Holy Grail seems to be predicting how an individual will behave; an immense amount of data has already been stored about each human being who owns a cellphone.

From his eye movements, demographic characteristics, and social networks it would be logical that an algorithm could predict what a person was going to buy, who he would vote for, and whom he will marry.

But hooray for free will, which the ancients said limits the omniscience of a Deity or which moderns might say will constrain the emerging singularity.

If scientists are having difficulty predicting the decay of a single carbon atom, then they are sure to have trouble with a human being. Despite my admiration for science and progress, I am comforted by that idea.

Far Removed From My Wheelhouse

As a retiree and part-time finance consultant, I've posted an abbreviated profile on LinkedIn.

That's enough information for some information scrapers to send me regular emails on jobs (e.g., controller, treasurer, interim CFO, investment analyst) that I might be interested in.

I gotta say that community organizer is far removed from my wheelhouse.

These job algorithms still need work.

Green for Show, Analytics for Dough

Being green: I re-use them at home.

Starbucks is often mocked for its responsiveness to the directional changes in progressive winds, but we need to give credit where it's due. We like the redesign of its cold drink lids, all to eliminate the plastic straws that comprise 0.03% of plastic waste.

[Digression: the straw-waste percentage is about the same as the percentage of carbon dioxide (.039%) in the atmosphere. One of the wonders of our time is the belief that the future of life on earth rests on eliminating the last speck of these "dangerous" substances. For perspective, getting rid of plastic straws would be like ordering a 200-pound person to lose 1/16th of a pound.]

(From imgflip)

Anyway, credit to Starbucks for finding a new use for sippy cups.

Mock them all you want, but Starbucks stock is on fire:

The world’s largest coffee chain posted adjusted earnings of 78 cents a share for its fiscal third quarter, a 26% increase compared with a year earlier. Analysts polled by FactSet had expected earnings of 72 cents a share, excluding certain items.

Starbucks’s sales of $6.8 billion were up 8%, also beating expectations. Global same-store sales growth of 6% was the strongest in three years. U.S. same-store sales rose 7%. Cold beverages helped drive growth in the afternoon, the company said, a time when Starbucks has struggled to boost sales...

Starbucks also remains focused on expanding its business in China. Net store growth there was 16% during the quarter, bringing the chain’s total to more than 30,000 locations world-wide.

FWIW, I think much of Starbucks' growth is due to collecting and analyzing data on its customers. The iPhone app knows my food and drink favorites, the stores I go to most frequently, and the promotions that I respond to. Starbucks also knows how often I do a grab-and-go and how long I linger when I use its WiFi.

With privacy all the rage, Starbucks has to downplay the role of data analysis:

Data collected by Starbucks’s digital programs is helping the company better understand what beverages customers want, said Chief Operating Officer Roz Brewer. “We know a lot more about our customers now,” she said.

I agree. I don't think the sippy cups are why the stock has a 30 PE.

SBUX has nearly doubled the past year.

Vast Benefits, Vast Risks

Microsoft research head Eric Horvitz believes that computers will be able to tell if you're sick. They will know this before you do and before you ask them to look. [bold added]

Eric Horvitz (Microsoft)

He and Microsoft colleague Ryen White, along with co-investigators from Stanford and other universities, have examined anonymized data from hundreds of millions of users of Bing, Microsoft’s search engine. They initially looked at query terms—what you search for—and the time and date of searches. Then, they added IP addresses—your computer’s unique identifier—and other location information. Most recently, they have focused on motor movements such as keystrokes, clicks and mouse activity.

This data can reveal critical diagnostic evidence. People confide intimate secrets about their health—yellow skin, odd-looking stools and other curious symptoms—to their search engine that they do not share with others, even physicians, Horvitz says. And the biometric and geographic information picked up by search engines may uncover secrets of which even users are unaware.

Just because someone did a web search on symptoms doesn't mean that he or she has cancer, but that datum can be correlated with other information, for example, if the user comes from a region with a high incidence of cancer. Similarly, a change in mouse-click speed and accuracy may indicate sleeplessness or a budding neurodegenerative condition, relatively easy to determine from other indicia (age, family history, whether the cellphone was active at 3 a.m., etc.)

We're not there yet, but we're nearing the point when computer networks will know each individual better than they know themselves. The benefits are vast, but so are the risks.

Sodden afterthought--if the Graduate were remade today, the dialogue might well go something like this:

MR. MCGUIRE: I want to say two words to you. Just two words.

BENJAMIN: Yes, sir.

MR. MCGUIRE: Are you listening?

BENJAMIN: Yes, I am.

MR. MCGUIRE: Predictive analytics.

Beyond Human Understanding

Example of complexity analysis on an algorithm

From the Dawn of Man human beings have worshipped phenomena--e.g., volcanos, hurricanes, the sun, moon, and stars--beyond their understanding.

Now mankind's own creations are on the verge of joining the ranks of the incomprehensible. Algorithms--step-by-step instructions that computers use to perform tasks--have become so complex that they defy analysis by the most expert programmers. In other words "God is in the Machine". According to an anonymous algorithm-builder in a very large, well-known company [bold added]

Within his tech giant, algorithms rarely stand alone. Instead, they exist within webs. “I rely”, he said, “on signals that are produced by other algorithms.” His algorithm was fed by outputs that were shaped by other algorithms. It was like a car assembly line. He, like his colleagues, worked on a small, specific part of a much larger process.

The algorithm was also constantly changing. The data inputs were flowing into the algorithm in real time, but the actual weights, measures and trade-offs that the algorithm made weren’t static either. Some of the functions that the researcher had woven in used machine learning – techniques where the machine constantly learned and adapted to what the most important patterns, correlations and relationships were. It meant that the algorithm was constantly changing and moving as the world moved around it, and its diet of data changed to reflect that.

We sat there, looking at the computer, his creation laid out in multi-coloured type. “This is all to do with complexity,” he said contemplatively. “Complexity of input. Complexity of analysis. Complexity of how outputs are combined, structured and used.” One of the reasons that he’d been employed to build a process like this was exactly because it could handle complexity by being complex itself. It grasped the blinding number of factors, signals and influences that bounced off each other at every moment in ways that we simply cannot.

Algorithms have changed, from Really Simple to Ridiculously Complicated. They are capable of accomplishing tasks and tackling problems that they’ve never been able to do before. They are able, really, to handle an unfathomably complex world better than a human can. But exactly because they can, the way they work has become unfathomable too. Inputs loop from one algorithm to the next; data presses through more instructions, more code. The complexity, dynamism, the sheer not-understandability of the algorithm means that there is a middle part – between input and output – where it is possible that no one knows exactly what they’re doing. The algorithm learns whatever it learns. “The reality is, professionally, I only look under the hood when it goes wrong. And it can be physically impossible to understand what has actually happened.”

We can still pull the plug but won't because we can't give up the benefits of the brave new world that has been built. Just pray that the God we are creating will be merciful.

Protection

"Free" privacy & security comes at a price.

One technology contretemps largely escaped notice, but it's useful for highlighting the differing philosophies of two of the largest companies in the world.

The issue concerns Virtual Private Networks, which are a method of protecting internet privacy:

A virtual private network (VPN) is a technology that creates a safe and encrypted connection over a less secure network, such as the internet. VPN technology was developed as a way to allow remote users and branch offices to securely access corporate applications and other resources. To ensure safety, data travels through secure tunnels and VPN users must use authentication methods -- including passwords, tokens and other unique identification methods -- to gain access to the VPN.

Last year Facebook offered a free VPN app, Onavo Protect, to its users:

Onavo allows users to create a virtual private network that redirects internet traffic to a private server managed by Facebook. The app, which bills itself as a way to “keep you and your data safe,” also alerts users when they visit potentially malicious sites.

However, as with most "free" internet services there was a price: [bold added]

Facebook is able to collect and analyze Onavo users’ activity to get a picture of how people use their phones beyond Facebook’s apps.

Onavo Protect did not comply with Apple's new privacy guidelines, and Apple and Facebook mutually agreed to remove Onavo from the App Store. There has been some posturing back-and-forth but the bottom line is:

Apple’s decision widens the schism between the two tech giants over privacy and is a blow to Facebook, which has used data gathered through the app to track rivals and scope out new product categories.

Except for brief experimentation with Google's Android in 2010, I've always used Apple's iOS, iPhone, and iPad products, and I'm glad I did.

Don't Be Typecast

Michal Kosinski

Stanford Business School professor Michal Kosinski demonstrates how it was possible that the techniques used by Cambridge Analytica influenced Facebook users in the 2016 election.

In an older study the researchers figured out a user's personality traits (for example, introversion or extroversion) based on his Facebook likes:

Kosinski and his Cambridge colleague, David Stillwell, were able to correlate “likes” with other basic personality traits: openness, conscientiousness, agreeableness, and neuroticism. Armed with only 10 “likes,” they could evaluate a person’s traits more accurately than that person’s coworkers. With 70 “likes,” they could do better than a person’s close friends.

Armed with personality information, [bold added]

Kosinski and his colleagues — including Stillwell, Sandra Matz of Columbia Business School, and Gideon Nave of Wharton School of Business — confirm[ed]the next logical step: Ads are indeed more persuasive when they are tailored to those psychological traits....

In promoting a line of cosmetics, for example, they ran dueling ads aimed at introverts and extroverts. All told, the ads reached 3 million people.

The ad for extroverts featured a woman dancing and the slogan “Dance like no one’s watching (but they totally are).” By contrast, the ad for introverts featured a woman contemplating herself in a mirror and a quiet slogan: “Beauty doesn’t have to shout.”

Sure enough, people were 50% more likely to buy the cosmetics if they saw the ad aimed at their particular type.

While these studies show how the melding of big data, artificial intelligence, and social networking can influence mass behavior, now that the public has become aware of it, IMHO, they'll be less susceptible to manipulation.

For example, when I watch the Hallmark Channel the ads for Medicare supplemental insurance and long-term care insurance come as no surprise because Hallmark and its viewers know the demographic of its audience. When it comes to Facebook and other sites, we did not know the extent to which they were targeting messages to us based upon our web behavior.

This is all the more reason to get out of our silos and look at news and opinions that we don't agree with.

Whatever your opinions or tribe, don't be typecast, and don't be manipulated.

In Tim I Trust

The Apple Watch has already breached my defenses.

The inputs: a person's

internet searches,

driving patterns ("the number of trips taken, duration of trips, left turns versus right turns and time spent on the highway versus local roads"),

how he or she uses a computer mouse or touchpad.

the time it takes to fill out a weekly online health questionnaire.

The black box: artificial intelligence and big data.

The output: the probability that he or she has early-stage Parkinson's or Alzheimer's Disease .

The above is a trend sweeping health care: the use of artificial intelligence (AI) to help better diagnose and treat patients. [bold added]

A Duke University doctor working with Microsoft researchers sifted through data on the physical movements of computer users that came from millions of internet searches. Their study found links between some behaviors—such as tremors when using a mouse, repeat queries and average scrolling velocity—and Parkinson’s disease. They used artificial intelligence, or a computer analysis, to identify which of the metrics separated a control group from those searching for Parkinson’s disease symptoms.

It used to be that family members were the first to notice that something is amiss. The machines that we touch every day--cars, phones, computers--are now more accurate, quicker, and smarter diagnosticians than loved ones. But what about individual privacy?

I HOPE that Apple lives up to its standards.

For those who are too cautious to buy a TV that can watch them, or a smart speaker that monitors what they're saying, the promise of real-time health monitoring may the lure that allows the Internet of Things(IoT) into personal space.

As for me, it's too late. My Apple Watch, which has been tracking me for 1½ years, knows more about me than my doctor.

In Tim Cook I trust (because I don't have any choice).

The Dam is Breaking

TWTR: We've had a nice run for two years, but it's time to take profits.

I don't trade stocks based on daily events, but this time I'm making an exception. I sold my 2-year investment in Twitter stock on Monday.

Having missed out on Facebook (I thought in 2012 that its IPO valuation was pricey at $100 billion--owning FB would make sense if the company was worth a crazy $200 billion by 2015, now it's north of $500 billion!), I did buy some Twitter in March, 2016, at $15.95 per share as an entrée to social media investing.

TWTR had continually disappointed analysts since its 2013 IPO at $26 per share. I rationalized the purchase by thinking that it did not have much further to fall; if it continued to under-perform someone would buy it for its brand recognition. Whatever the reason, it's been a happy ride to $36.80 last week.

Then the dam broke. Facebook has been called on the carpet by media and government officials not only because data on millions of users was "harvested" by the Trump campaign but also for allowing outside companies access to that data as well. Congressional hearings have been called, and its clear that both Democrats and Republicans are gunning for the social-media giant, albeit for different reasons which we won't get into here. The selling wave affected all social-media stocks, including Twitter.

There will undoubtedly be a lot of Congressional grandstanding over the next few months. There will be calls to regulate internet platforms like Facebook, Twitter, and Google. None of this will be good for those stocks, whatever one believes about their growth prospects.

I was happy to sell Twitter at $33.14 and will look to buying it back if it goes under $30.

If at First You Don't Succeed, Lie, Lie Again

(No, this post isn't about politics.)

One of the latest trends in fraud is to create fictitious people to take out loans. How can this happen in the age of Big Data, i.e., when large companies know more and more information about everybody? [bold added]

(WSJ graphic)

Synthetic-identity fraud exploits a vulnerability in America’s consumer-credit system. Lenders often consider a loan applicant legitimate if the applicant has a credit report at Equifax Inc., TransUnion or Experian PLC. But a new “credit file”—essentially a precursor to a credit report—often gets created when someone simply applies, even if the loan doesn’t come through.

Some lenders approve loans after reviewing credit files, which helps turn those files into full credit reports. That’s how a fictitious person, or 300 fictional people, can end up with a credit card.

Crooks know that the first application will be rejected, but it will allow the fake identity to get into the system and maybe some business-hungry lender will approve the next application.

Three comments:
1) The consumer-credit business is much less secure than other lending. Recently I've helped shepherd two real-estate deals through closure; in both cases the loans were under $500,000, the borrower's equity was 25-50%, the document requests were voluminous (tax returns, W-2's, brokerage statements, and balance sheets with questions about every line item), and numerous phone calls between borrowers and lenders.
2) Financial institutions who make personal loans to fake people deserve the consequences. No bailouts for them!
3) Fake identities are better than stolen identities. At least no consumers were directly harmed.