Andy Kessler

WSJ: Phones That Can Read Your Mind

Follow @andykessler

https://www.wsj.com/articles/phones-that-can-read-your-mind-11580071868

When I first met Larry Page and Sergey Brin around 1999, Google’s accidental billionaire founders (they couldn’t even sell their technology to Yahoo for $1 million) gave a talk about the future of search. Mr. Page rambled for a while, then stopped, grinned and said, “what we really want to do is know what you want before you even think it.” The audience laughed nervously, but that’s been a driving force of technology ever since.

Ad men since “Mad Men” have tried to bend our minds into buying products they’re pitching for their clients. But will we ever get real personalized advertising—ad systems that interact with us closely, in a real-time two-way dialogue about our wants and desires—that it will be a step ahead of our thinking?

I posed this question to advertising giant Martin Sorrell. He built the ad agency WPP in the 1980s based on the “big TV idea,” and today we’re inundated 24/7 with Bud Knight, the Geico gecko and Flo from Progressive. It’s now easy to tune out. Yet with the rise of the “fearsome fivesome” online behemoths—Google, Facebook, Amazon, Alibaba and Tencent—Mr. Sorrell suggests we’ve entered the era of “personalization at scale.”

He defines personalized advertising as “content, not perfect and not tent-pole, tailored to your needs and delivered to you at the time you want and the format you respond to.” He’s even created a purely digital advertising company, S4 Capital, to apply the strategy. Facebook, for instance, figured out the average person spends 1.7 seconds looking at Newsfeed posts on mobile devices, so S4 helped its client L’Oréal create 2-second makeup ads.

S4 also helped Netflix promote its drug kingpin show, “Narcos,” by cutting it into a million different short segments. Surfers at WSJ.com might see ads likening the show to business. Those on a sports or fitness site would see ads with lots of ripped muscles.

It’s the feedback loop that makes it personal.

WSJ: Your DNA Could Catch a Criminal

Follow @andykessler

https://www.wsj.com/articles/your-dna-could-catch-a-criminal-11578863580

In December, Michigan police arrested Robert Brian Thomas as a suspect in two cases of sexual battery—committed in Florida in 1998. These were the latest of hundreds of cold cases investigated since the Federal Bureau of Investigation’s April 2018 arrest of Joseph James DeAngelo, the alleged Golden State Killer. There are still hundreds of thousands of untested rape kits nationwide, but now technology may make it almost impossible to get away with this crime. Privacy concerns are an obstacle in solving cold cases—but not an insurmountable one.

Here’s a quick recap of how crime DNA is processed. The first step is for law enforcement to check samples against the Codis database, the FBI’s Combined DNA Index System. Codis is limited in its use of only 20 DNA markers, and it has records only for convicts and some prior detainees. Only 14% of sexual-assault DNA is matched to a suspect through Codis.

Fortunately, Ancestry.com has the DNA of more than 15 million users, and 23andMe has 10 million. Neither shares data with law enforcement. But many users upload their results to third-party sites, like GEDmatch and FamilyTreeDNA, to find long-lost relatives. DNASolves is another database recently set up specifically to help solve crimes.

After GEDmatch co-founder Curtis Rogers opened access to law enforcement, investigators uploaded DNA from one of the Golden State Killer’s 50 rape and 13 murder crime scenes from 1974-86. They didn’t find a direct match but discovered enough potential close relatives to search the family tree and narrow down the suspects. They followed Mr. DeAngelo to a store and swiped his DNA off the door handle of his car.

David Mittelman, who runs DNASolves, explains the numbers. Using DNA sequencing, he can analyze a sample for $5,000, which covers DNA extraction, genealogy and manpower. “Hopefully costs will halve in the next three to five years,” he says.

Hundreds more cold cases have been investigated since the Golden State Killer arrest.

WSJ: Tech Will Rule These ’20s, Too

Follow @andykessler

https://www.wsj.com/articles/tech-will-rule-these-20s-too-11578262248

Does history rhyme? A century ago, the ’20s boomed, driven by consumer spending on homes, cars, radios and newfangled appliances like refrigerators, sewing machines and vacuum cleaners. Most Americans couldn’t afford the upfront cost of a lot of these goods, so manufacturers and retailers invented installment plans. Debt ruled as 75% of cars, furniture and washing machines were bought on credit.

The supply-side policies of Treasury Secretary Andrew Mellon, who pushed for tax cuts in 1921, 1924 and 1926, increased the capital available for new consumer businesses. The consumption boom also sparked a giant service industry: My grandfather ran Kessler’s Refrigeration Service in Brooklyn, N.Y. Lifestyle advertising was perfected on billboards and in tabloids, which, like today, led to complaints about ads’ ubiquity and influence.

And stocks were hot, hot, hot. Radio Corp. of America shares were worth $11 in 1924, $20 in January 1928 and $114 in 1929, with a peak price/earnings ratio of 72. Turns out equities could be bought on installment too—sometimes a margin loan with a measly 10% down payment.

Like the movie “Titanic,” you know how this ends. Speculation. Tariffs. And finally, the crash of Oct. 28-29, where the market dropped 13% on Black Monday and 12% on Black Tuesday, triggering widespread margin calls for debt repayment. That risk is why the Federal Reserve now caps stock margins at 50%. According to legend, it’s also why windows in New York skyscrapers don’t open. By 1932, RCA stock was worth $3.

So what’s next? My fundamental rule for finding growth trends is that you need to see viable technologies today, and then predict which ones will get cheaper and better over time. Microprocessors, storage, bandwidth—all still going strong after half a century.

The fundamental building block of the 2020s will be artificial intelligence, particularly machine learning. Better to ask what it won’t change this decade.