The search bar on Apple’s Safari defaulting to Google is worth billions of dollars per year. What if the US DOJ stops that? CC-licensed photo by beegeye on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

---

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup. Also, if you go to the end of this post you’ll find an longish explanation for why this used to think Wednesdays were Fridays.

---

A selection of 10 links for you. Use them wisely. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.

---

Generative AI is coming for sales execs’ jobs—and they’re celebrating • WIRED

Paresh Dave:

»

Wining and dining, wooing clients with creative offers, and cashing big bonuses provide the glamor to sales work. Drafting answers to hundreds of dull questions posed by a prospective customer’s request for proposals? That’s just drudgery. Mercifully for workers, after months of speculation about ChatGPT-style AI taking over white-collar work, the corporate chore of responding to RFPs is one of the first that generative AI is disrupting.

In April, communications software maker Twilio introduced RFP Genie, a generative AI tool that digests an RFP, scours thousands of internal files for relevant information, and uses OpenAI’s GPT-4 to generate a suitable response. The company’s sales staff simply copy and paste the text over into a formal document and make a few adjustments.

RFPs that once occupied a pair of staffers for two weeks or more are now done in minutes. Twilio, whose cloud tools enable companies to chat with customers, expects to be able to make more and better sales pitches, and isn’t planning job cuts. “This will free up our solutions engineers to focus on more complex problems that demand not just reasoning, but human contextualization,” says Twilio CEO Jeff Lawson of the RFP bot, which has not previously been reported.

Lawson’s sales team isn’t the only one enjoying a sudden windfall of free time. Generative AI RFP response bots also have launched for sales teams at Google’s cloud unit, ad-buying agency EssenceMediacom, and DataRobot, a startup developing software to manage AI programs. In August at IBM, an RFP bot was selected by CEO Arvind Krishna as the winner of an internal AI hackathon called the Watsonx challenge, beating a field of over 12,000 entries. It used a large language model from IBM’s Watsonx.ai service to write answers in a tenth of the time compared to solely by hand, and the company is studying how to adopt the system.

«

If the AI can write the RFP, another one can read it. Maybe one could generate it too? Take the humans completely out of the loop. Though maybe this is how the machines take over: by writing in things for themselves that nobody notices. (RFPs are definitely among the most tedious documents in the world.)
unique link to this extract

---

At US v. Google antitrust trial, the Apple search deal takes center stage • The Verge

David Pierce:

»

“Would I be correct that, at least today, Apple has a lot of leverage in its negotiations with Google?”

Adam Severt, a Department of Justice attorney, asked that question to Google’s head of product partnerships, Joan Braddi, yesterday after a long tour of Braddi’s dealings with Apple over her two-decade-plus career at Google. The two were in a Washington, DC, courtroom, where for the last several weeks, the landmark US v. Google antitrust trial has litigated every corner of the search industry.

Braddi’s response was simple enough. “Yes.” Severt followed up: “Can you think of another search partner who might have more leverage than Apple?” “Not offhand, no,” Braddi said.

This exchange reiterated what has become one of the central themes of the trial so far: the overwhelming importance of Apple to search. Much of the trial is about the deals Google signs with lots of companies, from browser makers to wireless carriers, to be the default search engine on their platforms. But there is no deal more important, more lucrative, or more industry-defining than Google and Apple’s agreement over Safari.

«

A good point that was made about this trial (I forget by who): if the DoJ wins, Google is obliged to stop paying to be the default. But people will still overwhelmingly pick Google from a setup screen. So Google becomes much more profitable, and Apple becomes very much less profitable, to the tune of around $10bn per year each way. How is that (antitrust) success, exactly?

I seriously doubt Apple would build or buy its own search engine. It just isn’t worth the bother.
unique link to this extract

---

Overchoice, and how to avoid it • The Prism

Gurwinder:

»

Most of our everyday choices are between similar things; what movie to watch, what brand of toothpaste to buy. Fredkin’s paradox states that the more similar two choices seem, the less the decision should matter, yet the harder it is to choose between them. As a result, we often spend the most time on the decisions that matter least.

This is illustrated by Buridan’s ass, a mythical donkey that finds itself precisely equidistant from two identical bales of hay. The ass tries to make a rational decision as to whether to eat from the left bale or the right, but since there’s no rational reason to prefer either, the donkey wavers until it dies of hunger.

Buridan’s ass illustrates that there’s a cost to weighing options, which can exceed the cost of any of the options. Thus, the choices we make don’t need to be the best; they just have to be worth more than the time spent making them. If we spend less time making decisions, we can spend more time making whatever decision we made work.

The best way to manage the myriad decisions of the modern age is by employing “philosophical razors,” so-called because they shave away options, simplifying choices.

Naturally, there’s an overwhelming range of razors to choose from. I’ve tried scores of them, and have found that most aren’t workable, either because they lead to poor decisions or they’re too complicated for everyday decisions.

A few, though, have proven indispensable. Here are the five I use most.

«

They are indeed very fine. Recommended, though you might have to write them down somewhere to start making them reflexive.
unique link to this extract

---

MGM Resorts refused to pay ransom in cyberattack on casinos • WSJ

Katherine Sayre:

»

MGM Resorts International refused to pay a hackers’ ransom demand in a September cyberattack that threw its Las Vegas Strip resorts into chaos and crippled its properties and technology nationwide, according to a person familiar with the matter.

Service disruptions from the attack and efforts to resolve the issue will cost the company more than $100m in the third quarter, MGM said in a regulatory filing Thursday.

The cyberattack was detected on Sept. 10 and forced MGM to shut down IT systems in response. The shutdowns hobbled slot machines, interrupted online hotel bookings and required hotel workers to check-in guests using pen and paper for days, among other impacts. The company said Thursday that guest-facing operations have returned to normal.

MGM’s decision not to pay hackers is in line with guidance from the Federal Bureau of Investigation, which doesn’t support paying ransom. Doing so doesn’t guarantee that a company will recover its data, but does reward hackers and encourage bad actors to target more victims, the FBI’s website says.

…Rival Caesars Entertainment also suffered a hack late this summer and paid roughly half of a $30m ransom that hackers demanded, The Wall Street Journal previously reported. Caesars has said its operations weren’t impacted.

MGM said service disruptions would have a $100m negative impact on adjusted property earnings before interest, taxes, depreciation, amortization and rent for its Las Vegas and broader US resorts. The cost of remedial technology consulting, legal and advisory services was less than $10m.

«

Question is, does it make a difference to pay or to not pay? Caesars pays less, has no impact; MGM doesn’t pay, has big impact.
unique link to this extract

---

Meta employee shows how Trending Topics will work on Threads • 9to5 Mac

Filipe Espósito:

»

Officially, Meta hasn’t confirmed that Threads will have a Trending Topics section. However, a Meta employee accidentally posted a screenshot (now deleted) showing the feature in action. The post was noticed by app developer Willian Max, who reposted the screenshot hiding the employee’s name for obvious reasons.

As we can see, the interface is quite simple. It shows a ranking of the most commented topics on Threads with the number of posts on each topic. Trending Topics will appear in the Search tab, which has recently been expanded to let users find public posts from other users by keywords.

Since this screenshot comes from an internal version of the Threads app, it’s unclear whether this will be the final interface when the feature goes live to everyone. However, given that Meta employees already have access to a fully functional Trending Topics feature, we hope to see it available to the public soon.

A few weeks ago, Threads released an important update that finally lets users easily switch between accounts without having to log out and log in. In August, Meta launched the web version of Threads, allowing users to access the social network via a computer. The web version also works great on the iPad, as the iOS app is only optimized for the iPhone.

«

Saying “the web version works great on the iPad” is letting Meta off easily. It has never built an iPad version of Instagram – 13 years on! – and shows no signs of making an iPad app for Threads. It’s quite a gap; the iPad does get plenty of use.

Anyway, Trends – fine. But it should next do Lists, to attract power users, who will attract many more normal users.
unique link to this extract

---

It’s hard to build a durable business selling durable goods • The Diff

Byrne Hobart:

»

it’s easy to find brands that have led to enduring companies. In dishwashers, for example, Whirlpool has been around for a long, long time; they were founded in 1911, and own plenty of durables brands (Maytag, Kitchenaid, Amana, etc.). GE was in the appliance business for over a century, but sold it to Haier in 2016. Smaller brands like Le Creuset have also persisted for a long time, despite selling very durable durables indeed, offering a lifetime warranty for non-business use of many of their products. And there are newer contenders, like the aforementioned Haier, as well as LG, Samsung, and Panasonic.

We can divide these companies into two categories rather nicely: there are companies that started a very long time ago, often before the category was defined (Whirlpool’s founder started the company to investigate the hypothesis that electric motors would make dishwashers more effective), and there are newer companies that started well after the category had been defined, assembled their products in a country with low labor costs, and won share by achieving rough parity on quality and competing on cost.

This is a helpful split, because we see that more recent consumer durables brands tend to have a more dramatic history. It’s not just Instant Brands’ Instant Pot—there’s also Traeger (shares are down 80% since their 2021 IPO), Hamilton Beach (down by two thirds since going public in 2017), and GoPro (down almost 90% in their nine years as a public company). And even the ones that are doing reasonably well today aren’t in quite as good shape as they looked a few years ago, back when Instant Brands’ private equity owners and their lenders were feeling so optimistic. For example, Yeti has actually put up good numbers since it became publicly traded in 2018, with a total return of 159% since 2018—but it’s down by more than half from its peak valuation in late 2021.

«

It turns out there is a secret to building a durable durables business. However, there aren’t many places where it can be done.
unique link to this extract

---

Mastodon actually has 407K+ more monthly users than it thought | TechCrunch

Sarah Perez:

»

Open source and decentralized social network Mastodon has more users than it thought. The service, which competes with X (formerly Twitter) and other newcomers like Threads, Bluesky, Pebble and Spill, had been undercounting its users due to a network connectivity error, according to founder and CEO Eugen Rochko, and actually has 407,814 more monthly active users than it had been reporting previously. The adjustment also included a gain of 2.34 million registered users across an additional 727 servers that had not been counted due to the error.

The issue was impacting the metrics reported on Mastodon’s statistics aggregator on its joinmastodon.org/servers page, which had been undercounting users between October 2 and October 8. This issue has now been resolved, Rochko said. That leaves Mastodon with a total of 1.8 million monthly active users at present, an increase of 5% month-over-month and 10,000 servers, up 12% — a testament to Mastodon’s current upward swing at a time when the nature of X continues to remain in flux.

«

Soooo… fewer users than Threads? Meanwhile Bluesky has an unknown (to me at least) number of users. The splintering of social media continues.
unique link to this extract

---

‘Keep your paws off my voice’: voice actors worry generative AI will steal their livelihoods • Forbes

Rashi Shrivastava:

»

Voice actor Allegra Clark was scrolling through TikTok when she came across a video featuring Beidou, a swashbuckling ship captain from the video game Genshin Impact whom she’d voiced. But Beidou was participating in a sexually suggestive scene and said things that Clark had never recorded, even though the rugged voice sounded exactly like hers. The video’s creator had taken Clark’s voice and cloned it using a generative AI tool called ElevenLabs, and from there, they made her say whatever they wanted.

Clark, who has voiced more than 100 video game characters and dozens of commercials, said she interpreted the video as a joke, but was concerned her client might see it and think she had participated in it — which could be a violation of her contract, she said.

“Not only can this get us into a lot of trouble if people think we said [these things], but it’s also, frankly, very violating to hear yourself speak when it isn’t really you,” she wrote in an email to ElevenLabs that was reviewed by Forbes. She asked the startup to take down the uploaded audio clip and prevent future cloning of her voice, but the company said it hadn’t determined that the clip was made with its technology. It said it would only take immediate action if the clip was “hate speech or defamatory,” and stated it wasn’t responsible for any violation of copyright. The company never followed up or took any action.

“It sucks that we have no personal ownership of our voices. All we can do is kind of wag our finger at the situation,” Clark told Forbes.

In response to questions about Clark’s experience, ElevenLabs cofounder and CEO Mati Staniszewski told Forbes in an email that its users need the “explicit consent” of the person whose voice they are cloning if the content created could be “damaging or libelous.” Months after Clark’s experience, the company launched a “voice captcha” tool that requires people to record a randomly generated word and that voice must match the voice they are trying to clone.

«

unique link to this extract

---

It’s a global climate solution — if it can get past conspiracy theories and NIMBYs • NPR

Julia Simon:

»

In the 11th arrondissement, a middle-to-working class neighborhood in the east of Paris, if you walk out your front door, you can arrive at a preschool in one minute. A bookstore in three minutes. A cheese store in four minutes. Baguette for that cheese? Bakery’s across the street.

Grocery store and pharmacy, five minutes. Parks, restaurants, metro stops, a hospital: all within a 15-minute walk. I know this because I used to live there, on a tiny cobblestone street with buildings covered in vines.

This is a 15-minute city, says Carlos Moreno, a professor at University of Paris 1 Panthéon-Sorbonne, who met me on the banks of the Seine River. Moreno says that in a 15-minute city, a person can access key things in their life — work, food, schools and recreation — within a short walk, bike, or transit ride of their home.

My former Paris street and much of the neighborhood were built in this dense way more than 150 years ago. But this old idea of areas with many amenities close by has now evolved into an urban planning model gaining popularity with politicians around the world. Moreno says that’s because it not only improves quality of life, but 15-minute cities can also reduce cars’ planet-warming greenhouse gases. Transportation accounts for about 20% of global energy-related carbon dioxide pollution, with cars making up almost 10%, according to the International Energy Agency.

«

But, as the story explains, the fact that multiple people in different cities think it’s a good idea means it must be a Global Conspiracy. I mean, obviously.
unique link to this extract

---

The “Deaths of Despair” narrative is wrong • Slow Boring

Matthew Yglesias:

»

Over the past few years, Anne Case and Angus Deaton have unleashed upon the world a powerful meme that seems to link together America’s troublingly bad life expectancy outcomes with a number of salient social and political trends like the unexpected rise of Donald Trump.

Their “deaths of despair” narrative linking declining life expectancy to populist-right politics and to profound social and economic decay has proven to be extremely powerful. But their analysis suffers from fundamental statistical flaws that critics have been pointing out for years and that Case and Deaton just keep blustering through as if the objections don’t matter. Beyond that, they are operating within the confines of a construct — “despair” — that has little evidentiary basis. The rise in deaths of despair turns out to overwhelmingly be a rise in opioid overdoses. This increase is not happening in European countries that have not only been buffeted by the same broad economic trends as the United States, but are also seeing the rise of right-populist backlash politics.

The obvious explanation is that the US and Europe have very different laws governing pharmaceutical marketing.

That’s why the invention of supposedly-but-not-really safe time-release prescription opioids have wreaked havoc in the United States but not in Europe. Meanwhile, the same right-populist backlash occurs on both sides of the Atlantic because right-populist backlash politics is about the rising salience of conflict over post-material values like cosmopolitanism versus nationalism and has nothing to do with opioids or despair.

…They attribute the rising death rate among middle-aged white Americans to economic insecurity even though:

• There was no similar increase in death rates in European countries, including those that had worse economic conditions
• The rise in deaths began before the economic problems
• Black and Hispanic Americans who lived through the same economic conditions didn’t experience the same mortality trends.

«

Worth considering, if only to mull over what a difference that difference in pharmaceutical marketing makes.
unique link to this extract

---

• Why do social networks drive us a little mad? • Why does angry content seem to dominate what we see? • How much of a role do algorithms play in affecting what we see and do online? • What can we do about it? • Did Facebook have any inkling of what was coming in Myanmar in 2016? Read Social Warming, my latest book, and find answers – and more.

---

Errata, corrigenda and ai no corrida: OK, well sit back if you like programming bugs. The more observant of you will have noticed that at the top of these posts there has been a pointer to my Substack, noting that there’s another one on Friday. Odd thing, though: sometimes it would claim it was Friday when it clearly wasn’t. (Such as yesterday.) I’d noticed this a couple of times, but thought it was a quirk of timing or something. But this was too much. Wednesday is not Friday by any stretch of the imagination.

What was wrong in the code? Let’s have a look.

To generate what I call the “Substackstring” I had a subroutine in the Applescript that compiles the post. The subroutine is called, unsurprisingly,
substack()
Its first action is to call another subroutine in the larger script, which is called
getdate()
whose function is to find out which day the post is scheduled for, and whether that day occurs during British Summer Time (as that pushes the time of posting forward or back by an hour).

The getdate() code first did this:
set thedate to do shell script "date -j -v+18H \"+%Y-%m-%d-%H-%M\""
Which tells Applescript to run a little Unix terminal script to find out when the post is appearing. The +18H is to make sure that if I’m compiling the post the evening before (as usually happens), I’m allowing for the delay before the post appears in the morning. As long as I compile the post some time after 6am the day before, that code will pick the right day for the next day’s post.

A little wrangling later, and the getdate() subroutine finishes by returning a list:
{theyear, themonth, theday, summer}
which when I ran it today, October 11th 2023, is of the form {2023, 10, 12, true} – the latter for BST. That means that it’s saying the next post will be on October 12 2023, which will be during BST. Correct!

The getdate() subroutine is bombproof – it’s run for years and never been wrong. So the mistake must be in the substack() subroutine, which we have returned to now that getdate() has returned its result.

For the purposes of the substack() subroutine we don’t care about summer, so we dump the latter item of the list and create a string of the year-month-date:

set datestring to item 1 of datestring & "-" & item 2 of datestring & "-" & item 3 of datestring as Unicode text

Then we turn that string into a date (which is a particular format in Applescript), and extract the weekday by name:
set mydate to (date datestring)
set theday to (weekday of mydate) as string

And then do a test:
if theday = "Thursday" or theday = "Friday" then [say that it’s Friday. Logically, if the post is written on Thursday it must be for Friday posting; if the post is written on Friday it must be past midnight, also for Friday posting]
Alternatively, if the above fails:
if theday = "Tuesday" or theday = "Wednesday" then [say that there will be another post on Friday]
(If it’s Monday, we don’t mind, though probably there should be something there to point back to Friday’s triumph.)

So why didn’t this work? I added some logging to the substack() subroutine.
datestring: {2023, 10, 12, true}
Next we do this operation:
set datestring to item 1 of datestring & "-" & item 2 of datestring & "-" & item 3 of datestring as Unicode text
»» (2023-10-12)

Hmm, looking good so far. Next operation:
set mydate to date datestring
»» (Sunday, 15 April 2018 at 00:00:00)

Ah. I think I found the bug. How, you might wonder (I certainly did) does 2023-10-12 get transformed into the 15th of April 2018? Short answer: I don’t know. (* See below) But at least now I knew what I had to fix. Applescript is an odd language, in which date coercion can – as we see! – produce all sorts of odd results.

By now of course all the programmers in the audience (that’s everyone who’s read this far, I think) has been shouting the solution for at least five minutes: just run a shell script, like the getdate() subroutine did, to get the posting day!

And this is indeed the simple way. I avoided it originally because I wanted to be sure the substack() subroutine was in sync with the getdate() one, but that’s solved by the same method of adding 18 hours to the current time.

So my repaired substack() subroutine now says
set theday to do shell script "date -j -v+18H \"+%w\""
This responds with a number: Sunday = 0, Monday = 1, and so on. There’s a small gotcha: the response isn’t actually a number, it’s a text string, so the if/then statements have to allow for that. But that’s a minor point. So, the bug is gone. I might even tweak the subroutine so that on Mondays it will find the most recent Substack and embed that as a link. I’m sure nothing can go wrong with doing that.

* I still don’t know, but I wrote an Applescript to test it which cycles through most of the dates in 2023. See if you can figure out what the association is. I feel I can glimpse it – there’s an order in what comes out – but I couldn’t predict what a string will generate at the moment. Paste the code below into Script Editor on your Mac and see what happens.


set thislist to {} -- this will be a long list of all the generated dates that will be the result of this Applescript
repeat with j from 1 to 12 -- months
repeat with i from 1 to 30 -- days in the month. Not going to bother with accuracy, so 2023-02-30 will be generated
--inserting 0 before the number produces different results than not! Try it yourself.
if i > 9 then
set m to i as string
else
set m to "0" & i as string
end if
if j > 9 then
set n to j as string
else
set n to "0" & j as string
end if
set datestring to ("2023-" & n & "-" & m) as Unicode text
set mydate to date datestring
set end of thislist to {datestring, mydate}
end repeat
end repeat
thislist -- the result of the Applescript. View it and puzzle. You can use different date delimiters ( /, space, double space all work) and get the same result. Why 2006? Why the 16th and 15th? Why starting in July? I haven't tried different years but they probably generate different answers.