Start Up No.1910: India’s gig workers beat the algorithm, Musk/Twitter (groan), the woman who fell from the sky, crypt0?, and more

Tidal power systems don’t show much above the water, but their generation costs are falling fast – and could undercut nuclear in a few years. CC-licensed photo by Scottish Government 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.

A selection of 10 links for you. Continuing. I’m @charlesarthur on Twitter. Observations and links welcome.

Gig workers in India are uniting to take back control from algorithms • Rest of World

Varsha Bansal:


On October 17, Santosh Kumar, an Uber driver in the south Indian city of Hyderabad was wrapping up an almost 12-hour shift and struggling to find a last ride in the direction of his home. The app showed him a message that destinations in that area weren’t available.

Frustrated, he turned to a Telegram group called CCDA, or Commercial Cab Driver’s Awareness, where he shared his woes with over 5,000 fellow drivers. Within minutes, his peers offered a jugaad — a cheap hack — to game the system: keep trying to book a ride in the direction of your home, and the algorithm will eventually oblige.

Two days later, in the same group, another distressed Uber driver posted screenshots of a “miscellaneous” fee of over 5,000 rupees ($61) that Uber had levied on him. The screenshot indicated that if he didn’t make the payment, he would lose access to his Uber account. He didn’t really understand how Uber calculated this amount and wondered how he would be able to afford the hefty payment.

CCDA members explained that this was a mandatory tax payment and offered a jugaad to offset the hefty one-time charge: keep accepting rides, and Uber will auto-deduct the amount from the daily earnings rather than paying the big amount upfront. “They explained to me that until this amount is cleared, I would only get rides with online payments and not cash rides,” the driver told Rest of World, requesting anonymity fearing retribution from Uber. He managed to clear more than 2,000 rupees of the tax liability in under a month.

Those are just two examples of how India’s gig workers — tired of the obscurity around black box algorithms and technologies that dictate their lives and work — are finding ways to game the platforms to their advantage. Drivers and delivery persons, who work for apps like Uber, Ola, Zomato and Swiggy, are trying to reverse engineer these apps, frequently sharing this information through groups like CCDA and in-person workshops.


Rage collectively against the machine. Neat.

unique link to this extract

How tide has turned on UK tidal stream energy as costs ebb and reliability flows • The Guardian

Tom Wall:


The cost of generating power from tidal streams has fallen by 40% since 2018 – and a report published last month by a government-backed research centre, Offshore Renewable Energy Catapult, forecasts prices could fall below nuclear energy in little over a decade, with one-megawatt hour of power due to cost as little as £78 by 2035 compared with £92.50 for the new Hinkley Point C power plant.

Simon Cheeseman from the research center argues tidal stream energy is at the “point of commercialisation” as companies are keen to scale up production and deployment. But he says the sector still needs careful nurturing to ensure it follows the successful trajectory of offshore wind, which in 11 years has gone from generating only enough energy for 4% of British homes to generating enough for 33% of British homes. “In the early days of offshore wind, you had strong government support. This is the perfect blueprint for tidal stream energy,” he says. “There is no reason tidal can’t follow that same route.”

Orbital Marine, which operates what it says are the world’s most powerful turbines below a plane-like floating platform near Orkney, has secured government funding to deploy three more floating turbines next year. Each platform can generate enough power for 2,000 homes and creates an estimated 100 jobs, according to the firm. “We want this to kickstart a real phase of change for us. We want to start manufacturing consistently and pull in more commercial investment,” says Andrew Scott, the company’s chief executive. “This is the first time in my 20 years in marine renewables that we’ve got a genuine chance of making tidal stream energy work commercially.”


Not quite clear whether tidal will provide a constant “base load” in the way that nuclear does. Though that cost estimate seems a long way off. Presently, the report says, the cost is £178/MWh.
unique link to this extract

Peloton Row review: the price isn’t right • The Verge

Victoria Song:


Any other year, the Peloton Row would’ve made a splash. The long-awaited rower was the “worst-kept secret” in connected fitness, and its launch heralds Peloton’s expansion into a whole new category. But this is a year where Peloton laid off thousands of employees, shuttered its domestic manufacturing, and watched its stock price spiral down the drain. Peloton would have you believe that the Row revolutionizes rowing. But while testing the Row, which costs $3,195, I couldn’t help but wonder how it fits into Peloton’s future.


How much?? The Row does have a useful feedback system which critiques your technique, and where you’re going wrong and right, but is that really worth $2,000? As Song points out, in this economy, you’re not going to get the marginal buyer; they’ll buy a Concept 2 rower and get an Apple Fitness+ subscription and still have $2,000 left over – plus the monthly Peloton sub they won’t be paying.

The Peloton diehards will surely go for it, but I think the place this will have in Peloton’s future isn’t big.
unique link to this extract

Hey Elon: let me help you speed run the content moderation learning curve • Techdirt

Mike Masnick:


It’s kind of a rite of passage for any new social media network. They show up, insist that they’re the “platform for free speech” without quite understanding what that actually means, and then they quickly discover a whole bunch of fairly fundamental ideas, institute a bunch of rapid (often sloppy) changes… and in the end, they basically all end up in the same general vicinity, with just a few small differences on the margin. Look, I went through it myself. In the early days I insisted that sites shouldn’t do any moderation at all, including my own. But I learned. As did Parler, Gettr, Truth Social and lots of others.

Anyway, Elon’s in a bit of a different position, because rather than starting something new, he’s taken over a large platform. I recognize that he, his buddies, and a whole lot of other people think that Twitter is especially bad at this, and that he’s got some special ideas for “bringing free speech back,” but the reality is that Twitter was, by far, the most successful platform at taking a “we support free speech” stance for content, and learned over time the many nuances and tradeoffs involved.
And because I do hope that Musk succeeds and Twitter remains viable, I wanted to see if we might help him (and anyone else) speed run the basics of the content moderation learning curve that most newbies run into. The order of the levels and the seriousness of each can change over time, and how it all fits together may be somewhat different, but, in the end, basically every major social media platform ends up in this same place eventually (the place Twitter was already at when Musk insisted he needed to tear things down and start again).

Level One: “We’re the free speech platform! Anything goes!”

Cool. Cool. The bird is free! Everyone rejoice.

“Excuse me, boss, we’re getting reports that there are child sexual exploitation and abuse (CSAM) images and videos on the site.”


Masnick wrote this excellent piece a month ago, and we’re working our way through his 20 (count them) levels of difficulty. Presently we’re at about Level Six. Things get harder as you go, as you’ll have guessed.
unique link to this extract

Musk touts all-time high twitter signups and daily active users on as he promises new features • Forbes via MSN

Siladitya Ray:


In a tweet showing off slides from his company-wide presentation, Musk claimed Twitter averaged over 2 million new user sign ups per day in the past week, a record high for the platform.

Musk’s presentation also reiterated his earlier claims about user growth, noting that monetizable daily active users (mDAU) on the platform have crossed 250 million for the first time.

Musk’s slides also highlighted a drop in impersonator accounts on the platform which spiked after the launch of paid verification, however, it is unclear if this is a result of Twitter improving its ability to take down such accounts or its decision to halt the rollout of the service.

Amid concerns about Musk’s decision to restore banned controversial accounts, the Twitter CEO claimed hate speech impressions on the platform are lower than last year.

The rest of Musk’s presentation talks about his dream of turning Twitter into an “everything app” by touting expanded video-sharing capabilities, encrypted messaging, long-form tweets, and payments.

In a later tweet, Musk stated that he sees “a path to Twitter exceeding a billion monthly users in 12 to 18 months,” which would put it on par with TikTok but still significantly behind Facebook’s number of nearly 3 billion and Instagram’s 2 billion.


Of course you can’t trust any of these claims. Musk isn’t answerable legally if he lies here. Meanwhile, if we do take these as accurate, there’s plenty of wiggle room:
• the signups could well be bots (how many of the signups were then removed?)
• mDAU does include bots (as the previous Twitter admin acknowledged)
• the “impersonator” accounts was measuring *reported* impersonations
• there are fewer moderators to accurately record hate speech.

Trust nothing he says; analyse only what he does. The slideshow also mentioned encrypted DMs as forthcoming (can’t be end-to-end because otherwise you couldn’t read it them a web browser, I think) and “payments”, which remains completely unclear.
unique link to this extract

Crypto’s final price could be zero • WSJ

Andy Kessler:


A cottage industry of firms emerged to lever up crypto. This is when things turned toxic. The first task was to lure customers by paying interest on their crypto holdings. The Anchor Protocol behind the spectacularly imploded Terra-Luna algorithmic tokens was paying up to 20%.

Other platforms such as Binance and would pay 4%, 8% or more on crypto as well, suckering in the masses who could earn only 0.01% interest from, well, real banks. But how could anyone pay interest on crypto? By turning around and lending it out to hedge funds and others who also used leverage. Insanity.

Genesis Global Capital created a lending platform to facilitate borrowing crypto. Lending against what? Again, just air. Firms such as Gemini, set up by the Winklevoss twins, were paying 8% interest, so customers could harvest yields. Why was there any yield on crypto? Good question. It worked on the way up, not so much on the way down. Crypto was lent out like a hot potato until someone got stuck with the value down 90% and everyone else left with defaulted debt. This was probably the only way the delusion could have ended.

Most of these platforms are now frozen and might disappear as customers caught with a hot potato frantically demand withdrawals in the wake of the FTX collapse. Of course, all these crypto lenders had to do was ask: What’s the underlying collateral? Where are the assets? With no good answer, no sane lender would have lent against it. But no one asked.

…Technology, like Red Bull, is a supercharger until it wears off. Debt, like milk, can kill you when it spoils. They don’t mix.


People have been acting shocked at the headline, but it’s obvious in itself. To call crypto a speculative asset is to misuse the word “asset”.
unique link to this extract

Juliane Koepcke: how I survived a plane crash • BBC News

Koepcke was aged 17 when she got onto a plane that was to fly over the Peruvian rainforest:


It was Christmas Eve 1971 and everyone was eager to get home. We were angry because the plane was seven hours late.

Suddenly we entered into a very heavy, dark cloud. My mother was anxious but I was OK, I liked flying.
Ten minutes later it was obvious that something was very wrong. There was very heavy turbulence and the plane was jumping up and down, parcels and luggage were falling from the locker, there were gifts, flowers and Christmas cakes flying around the cabin.

When we saw lightning around the plane, I was scared. My mother and I held hands but we were unable to speak. Other passengers began to cry and weep and scream. After about 10 minutes, I saw a very bright light on the outer engine on the left. My mother said very calmly: “That is the end, it’s all over.” Those were the last words I ever heard from her.

The plane jumped down and went into a nose-dive. It was pitch black and people were screaming, then the deep roaring of the engines filled my head completely. Suddenly the noise stopped and I was outside the plane. I was in a freefall, strapped to my seat bench and hanging head-over-heels. The whispering of the wind was the only noise I could hear.

I felt completely alone.

I could see the canopy of the jungle spinning towards me. Then I lost consciousness and remember nothing of the impact. Later I learned that the plane had broken into pieces about two miles above the ground.

I woke the next day and looked up into the canopy. The first thought I had was: “I survived an air crash.”
I shouted out for my mother in but I only heard the sounds of the jungle. I was completely alone.

I had broken my collarbone and had some deep cuts on my legs but my injuries weren’t serious. I realised later that I had ruptured a ligament in my knee but I could walk.


Incredible story of survival; surviving the crash was only the beginning, because now she was lost in the rainforest. She probably wouldn’t have survived if her parents hadn’t been zoologists who had worked in it.
unique link to this extract

June 2016: The iPhone’s biggest threat isn’t Android—it’s Amazon’s Echo • WIRED

Davey Alba, wayyy back in 2016:


On slide 133 of her much-anticipated annual Internet Trends report, venture capitalist Mary Meeker made a curious comparison. She put a graph of iPhone sales side-by-side with a sales estimate for the Echo, the newish wireless speaker and voice-activated personal assistant from Amazon.

That juxtaposition might seem strange, but Meeker was making a point. Sales of the iPhone have been slowing, and according to Meeker’s projections, they’ll go into decline by the end of 2016. Right as this is happening, sales of the Amazon Echo are starting to take off.

It’s a sign that using voice as a way to command your tech is steadily gaining traction. By 2020, according to Andrew Ng—chief scientist at Chinese Internet company Baidu, who Meeker cites in her report—at least 50% of all searches will make use of images or speech.

…she said as of last month, 20% of searches on Android smartphones were voice-based.

Meeker seems to be suggesting, however, that the traditional smartphone won’t necessarily rule all when it comes to seeking digital assitance. As the Echo’s popularity shows, there’s a burgeoning opportunity to go not just hands-free but screen-free. Just yesterday, Amazon CEO Jeff Bezos said more than 1,000 people are working on the Echo and Alexa, the software that powers his company’s voice-activated assistant.


Ah, what might have been, but wasn’t. As we now know, Alexa has been a multi-billion pound bust. Meanwhile, the iPhone has done OK (in terms of revenue, for certain).
unique link to this extract

‘iSpoof’ service dismantled, main operator and 145 users arrested • Bleeping Computer

Bill Toulas:


The ‘iSpoof’ online spoofing service has been dismantled following an international law enforcement investigation that also led to the arrest of 146 people, including the suspected mastermind of the operation.

Over a hundred of these arrests, including that of the platform’s leader, were made by London’s Metropolitan Police.

iSpoof offered cybercriminals so-called “spoofing” servers which allowed them to mask their phone numbers with one belonging to a trusted organization, making it appear to the victims as if their bank called them.

This call number spoofing made it possible for the crooks to conduct social engineering, phishing, and carry out “bank helpdesk” scams, stealing money, banking account credentials, and one-time codes.

“The services of the website allowed those who sign up and pay for the service to anonymously make spoofed calls, send recorded messages, and intercept one-time passwords,” Europol said on Thursday.

“The users were able to impersonate an infinite number of entities (such as banks, retail companies, and government institutions) for financial gain and substantial losses to victims.”

According to the announcement of the Metropolitan Police, between June 2021 and July 2022, iSpoof was used to make 10 million fraudulent calls worldwide.

Europol reports that iSpoof caused approximately $120m in losses, with the service’s operators raking in estimated profits of $3.85m in the last 16 months.


It’s taken forever for the police to get around to this. These sorts of scams have been going on for what feels like a decade – certainly five years – and has been written about extensively in the papers. Police work might be slow sometimes, but it feels like this was left on a back burner.

The fix that’s still needed is to prevent such spoofing. Another iSpoof site was up within minutes.

unique link to this extract

Study: AirPods Pro are *this* close to being full-fledged hearing aids • Ars Technica

Kevin Purdy:


A study in the journal iScience suggests that, in some noise situations, AirPods, particularly the Pro model, can work just as well as far pricier prescription-only models.

AirPods are not sold or approved by the Food and Drug Administration as devices for those with mild to moderate hearing loss. But with cheaper, over-the-counter hearing aids now available at common retailers, there’s a renewed interest in non-medical companies moving into the space to help people who don’t need expert care—including from Apple itself.

Researchers from the Taipei Veterans General Hospital, Taiwan’s National Yang Ming Chiao Tung University, and other entities conducted what they believe is the first comparison of smartphone-oriented earphones with medically prescribed hearing aids. The study had a very small sample size of 21 people between 26 and 60 years old and was conducted in a lab setting with a single source of sound. Still, the results are intriguing, especially considering how many people already have access to iPhones, AirPods, and their audio-enhancing features.

The researchers tested AirPods with their Live Listen feature activated against five standards for a personal sound amplification product (PSAP) under ANSI CTA 2051-2017:

• Frequency response smoothness
• Frequency response bandwidth (range)
• Maximum output sound pressure level (OSPL) at 90 decibels input
• Total harmonic distortion (THD)
• Equivalent input (or internal) noise level (EIN)

AirPods 2 only met two of the standards, bandwidth and THD, while AirPods Pro met all of them except EIN, registering 37 decibel sound pressure levels (dB SPL), when the standard calls for 32 or less.


Quite possible there are other TW (true wireless) noise-reducing headphones out there which can do the same or a better job. Apple’s brand still leads on this stuff, just as the iPod became the only music player, and the iPhone the only smartphone, and the iPad the only tablet – even though they weren’t, or aren’t.
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: none notified

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.