XDA: “3 reasons people are moving from Windows 11 back to Windows 10”

Recently, StatCounter posted a chart showing that Windows 10’s market share is growing, while Windows 11’s is shrinking. This flies in the face of what Microsoft has planned, especially given how Windows 10’s end-of-life date is looming in the distance. As such, I wanted to do some research into the topic; did StatCounter’s statistics get things wrong? Or are people downgrading from Windows 11 to Windows 10? While there’s no way to state without a doubt that people are downgrading, I did find people with pain points with Windows 11 that might be contributing to the system’s downfall.

---

Again, this is speculation on my part, but I can’t help but notice that the Windows 11 adoption rate began reversing in January 2024, when Microsoft began introducing and announcing new Copilot features. There’s a chance that people are turning back to Windows 10 to dodge the 24H2 update, either as individuals who dislike the tools or as businesses that don’t want to entrust their private data with artificial intelligence.

Simon Batt

Pretty embarrassing for Microsoft to be losing market share on their most recent OS – then again, Windows 11 has been a pretty embarrassing upgrade from Windows 10. In the three years since its release things have barely improved, while Microsoft seems completely oblivious to these challenges faced by consumers and focused on chasing wild AI dreams instead of polishing the OS and expanding hardware support. I expect adoption problems will only worsen as Microsoft adds more AI features, as these will require hardware upgrades many will be reluctant to perform – including corporate customers.

Semafor: “The incredible shrinking podcast industry”

Apple has quietly tightened its reporting of how many people listen to podcasts, sending shock waves through an embattled audio industry still reeling from the end of the COVID-era production bubble.

The shift, Apple wrote in a blog post, was technical: The dominant podcasting platform had begun switching off automatic downloads for users who haven’t listened to five episodes of a show in the last two weeks.

But while few users noticed the shift, some of the biggest podcasts in the world saw their official listener numbers drop dramatically. Long-running shows that publish frequently were hit particularly hard. A user who listened to a show like The New York Times’ The Daily a few times, subscribed, but stopped listening would continue to count as a download indefinitely. Even better under the old rules: For people who listened to a show, dropped off for a while, but started listening again later, Apple would automatically download every show in between. The arrangement drove big download numbers, a crucial metric for ad sales and a sign of the vast reach of podcasts as a medium.

Max Tani

The title is somewhat misleading, as podcasting hasn’t suddenly shrunk overnight, but was instead revealed to have been smaller than previously reported. Not exactly a novel issue, as I have written before on this blog, but a reflection of the constant hyping and inflated expectations that is associated with basically every new trend in tech circles.

Ethan Zuckerman: “How Big is YouTube?”

Here’s how this works: YouTube URLs look like this: https://www.youtube.com/ watch?v=vXPJVwwEmiM

That bit after “watch?v=” is an 11 digit string. The first ten digits can be a-z,A-Z,0-9 and _-. The last digit is special, and can only be one of 16 values. Turns out there are 2^64 possible YouTube addresses, an enormous number: 18.4 quintillion. There are lots of YouTube videos, but not that many. Let’s guess for a moment that there are 1 billion YouTube videos – if you picked URLs at random, you’d only get a valid address roughly once every 18.4 billion tries.

We refer to this method as “drunk dialing”, as it’s basically as sophisticated as taking swigs from a bottle of bourbon and mashing digits on a telephone, hoping to find a human being to speak to. Jason found a couple of cheats that makes the method roughly 32,000 times as efficient, meaning our “phone call” connects lots more often. Kevin Zheng wrote a whole bunch of scripts to do the dialing, and over the course of several months, we collected more than 10,000 truly random YouTube videos.

---

Once you’re collecting these random videos, other statistics are easy to calculate. We can look at how old our random videos are and calculate how fast YouTube is growing: we estimate that over 4 billion videos were posted to YouTube just in 2023. We can calculate the mean and median views per video, and show just how long the “long tail” is – videos with 10,000 or more views are roughly 4% of our data set, though they represent the lion’s share of views of the YouTube platform.

Ethan Zuckerman

Fascinating method – and fascinating results! If these are anywhere near accurate, in 2023 there was a new YouTube video for every other person living on the planet! The stats also highlight the extreme inequality in traffic between a tiny minority of popular uploads and the typical YouTube video, which can’t even top 10,000 views.

Goldman Sachs Research: “The Risks of a Higher Rate Regime”

The Risk from Unprofitable Firms

In the corporate sector, investors might hesitate to continue financing unprofitable companies that they hope will pay off well down the road now that the opportunity cost has risen. The number of unprofitable firms has risen in recent decades, reaching almost 50% of all publicly-listed companies in 2022 (Exhibit 7, left)^[3]. The share of business activity that they account for is much smaller but still an economically meaningful 10% of total business revenues (Exhibit 7, right).

Unprofitable firms account for closer to 13% of capital spending and employment, and even just the smaller group of persistently unprofitable firms account for about 5% of employment (Exhibit 8).

Higher funding costs could force some of these companies to cut labor costs or even close. In previous research we found that unprofitable firms tend to cut capital spending more aggressively when faced with margin pressure, and we find they also cut labor costs more aggressively when hit with interest rate shocks (Exhibit 9, left). The larger risk is that some firms might simply have to close if their path to profitability is too distant. The exit rate of unprofitable firms is currently low by historical standards and has actually declined since the start of the pandemic, leaving it ample room to rise from here (Exhibit 9, right).

David Mericle & Ronnie Walker

Fascinating figures about the US economy: almost half of all listed companies are now unprofitable! Presumably an outcome of the near-zero interest rate regime that has dominated over the past decade, these companies can be expected to face increasing pressures to become profitable through comprehensive restructuring, or… go bankrupt or close down, contributing to higher unemployment and decreases in economic activity.

The Economist: “The lessons from America’s astonishing economic record”

If there is one thing that Americans of all political stripes can agree on, it is that the economy is broken. Donald Trump, who saw trade as a rip-off and his country in decline, came into office promising to make America great again. President Joe Biden is spending $2trn remaking the economy, hoping to build it back better. Americans are worried. Nearly four-fifths tell pollsters that their children will be worse off than they are, the most since the survey began in 1990, when only about two-fifths were as gloomy. The last time so many thought the economy was in such terrible shape, it was in the throes of the global financial crisis.

Yet the anxiety obscures a stunning success story—one of enduring but underappreciated outperformance. America remains the world’s richest, most productive and most innovative big economy. By an impressive number of measures, it is leaving its peers ever further in the dust.

Start with the familiar measure of economic success: gdp. In 1990 America accounted for a quarter of the world’s output, at market exchange rates. Thirty years on, that share is almost unchanged, even as China has gained economic clout. America’s dominance of the rich world is startling. Today it accounts for 58% of the G7’s gdp, compared with 40% in 1990. Adjusted for purchasing power, only those in über-rich petrostates and financial hubs enjoy a higher income per person. Average incomes have grown much faster than in western Europe or Japan. Also adjusted for purchasing power, they exceed $50,000 in Mississippi, America’s poorest state—higher than in France.

The Economist

Far from being convincing about ‘America’s astonishing economic record’, this article illustrates how easily you can draw flawed conclusions by cherry-picking data and relying on broad, but imprecise indicators. A higher income per person doesn’t mean much in a country with rising income inequality; higher efficiency and more patents doesn’t mean much if the gains from this productivity and innovation are amassed by the top 1% of earners and by corporations that are essentially monopolies with ballooning cash reserves.

Analytics at Meta: “Notifications: why less is more”

We had a hunch that such a trade-off should be avoidable: if users are more satisfied with their experience, we believe this would be reflected in usage gains too. We decided to keep the experiment of sending only a few notifications running for a year. And lo and behold: little by little Facebook usage started inching back up! After a year we saw that in the fewer notifications experience, users were using Facebook more — it just took a long time for user behavior to shift and less disruption led to high organic usage, which increased both user satisfaction and app usage. We wrote this blogpost to socialize this finding as we believe other data science teams outside Facebook could benefit from the same lessons we learnt:

1. Experiments in the very long-run can show different results than in the short run.
2. Sending only a few notifications and the most relevant ones (e.g. those predicted to receive a 5/5 rating in the survey mentioned above) can improve user sentiment and usage too in the long run.

Weijun C., Yan Q., Yuwen Z., Christina B., Akos L.

Seems a rather obvious insight that people become frustrated with high volumes of notifications, and after a short period of increased engagement they start to pay less attention to them. This is especially true as people tend to add more friends and follow more sources over time, and Facebook crams new features into the feed, both contributing to even more notifications. But it’s nice to see this confirmed with actual research and, as far as I understand from this post, implemented into real product changes at Facebook.

MKBHD 2022: “Blind Smartphone Camera Test”

The Best Smartphone Camera 2022!

It’s time to find out once with SCIENCE: What’s the best smartphone camera? https://vote.MKBHD.com

Marques Brownlee

Cool experiment! One of the issue with smartphone reviews are that you are getting a limited viewpoint, that of the reviewer, colored by their hidden preferences and subjective views. With a blind test though people don’t know the brand beforehand, so the results should be less influenced by preconceived expectations.

World Health Organization: “14.9 million excess deaths associated with the COVID-19 pandemic in 2020 and 2021”

Excess mortality is calculated as the difference between the number of deaths that have occurred and the number that would be expected in the absence of the pandemic based on data from earlier years.

Excess mortality includes deaths associated with COVID-19 directly (due to the disease) or indirectly (due to the pandemic’s impact on health systems and society). Deaths linked indirectly to COVID-19 are attributable to other health conditions for which people were unable to access prevention and treatment because health systems were overburdened by the pandemic. The estimated number of excess deaths can be influenced also by deaths averted during the pandemic due to lower risks of certain events, like motor-vehicle accidents or occupational injuries.

Most of the excess deaths (84%) are concentrated in South-East Asia, Europe, and the Americas. Some 68% of excess deaths are concentrated in just 10 countries globally. Middle-income countries account for 81% of the 14.9 million excess deaths (53% in lower-middle-income countries and 28% in upper-middle-income countries) over the 24-month period, with high-income and low-income countries each accounting for 15% and 4%, respectively.

The estimates for a 24-month period (2020 and 2021) include a breakdown of excess mortality by age and sex. They confirm that the global death toll was higher for men than for women (57% male, 43% female) and higher among older adults. The absolute count of the excess deaths is affected by the population size. The number of excess deaths per 100,000 gives a more objective picture of the pandemic than reported COVID-19 mortality data.

Technical Advisory Group on COVID-19 Mortality Assessment

Tracking excess mortality as a consequence of the coronavirus pandemic is an important tool for assessing various health measures and their effectiveness. This study however arrives at some questionable conclusions and has been discussed critically on Twitter.

Coronavirus in Romania: skydiving without a parachute

While the attention of the public and authorities has been thoroughly captured by the Russian invasion of Ukraine over the past weeks, and despite massive efforts to diminish the issue in public eyes, the two-year-old pandemic is far from being a thing of the past. As I feared in my previous article, case numbers have skyrocketed here in Romania throughout January, reaching an official daily high over 40,000 on February 1^st, a new record. Like other Omicron waves around the world, the sharp increase was followed by a quick retreat, with new cases declining at a rate of about 35% each week. Currently, in the second week of March, the case numbers hover above 3,000 daily on average.

Mortality has also progressed on a trajectory similar to other Omicron waves: deaths have increased after the typical two-week delay, but at a much lower rate than cases – on average 35% per week compared to almost 90% for cases. The number of daily deaths surpassed 200 on two occasions in the second half of February, considerably lower than the mortality of the autumn Delta wave, which reached as high as 600 a day. Deaths are also currently in decline, although, again, at a lower rate than cases, around 20% per week. Our experience seems to support conclusions from other countries that the Omicron mutation causes less severe disease, at least in populations with sufficient prior immunity, either through infection or vaccinations. But during this wave, medical personnel also had access to new antiviral medication for severe COVID cases, so this might have also contributed to keeping the death rate lower than before.

The Atlantic: “Is Old Music Killing New Music?”

Old songs now represent 70 percent of the U.S. music market, according to the latest numbers from MRC Data, a music-analytics firm. Those who make a living from new music—especially that endangered species known as the working musician—should look at these figures with fear and trembling. But the news gets worse: The new-music market is actually shrinking. All the growth in the market is coming from old songs.

The 200 most popular new tracks now regularly account for less than 5 percent of total streams. That rate was twice as high just three years ago. The mix of songs actually purchased by consumers is even more tilted toward older music. The current list of most-downloaded tracks on iTunes is filled with the names of bands from the previous century, such as Creedence Clearwater Revival and The Police.

Ted Gioia

Interesting observation, but I fear the author is drawing exaggerated conclusions in support of his own preconceived ideas, as he later launches into a tireless rant against the music industry. First of all, the definition of this supposedly ‘old’ music is tracks released more than 18 months ago – the standard delineation in the music business in fact. That can hardly be considered ancient music snuffing out rising talent. The original report also makes clear that this decline happened for the first time since MRC Data began tracking streaming music in 2008 – not exactly a long-standing trend that should be looked upon with fear and trembling.

Coronavirus in Romania: brace for the Big, Bad O.

As expected in my last update, infections soared in Romania over the month of October, reaching as high as around 19.000 a day in the middle of the month, and a total of almost 415.000 for the entire month. Following a similar pattern with other Delta waves around the globe, new cases started decreasing afterwards almost as fast as they were rising before, despite few extra restrictions – you may attribute this pattern to some inherent property of Delta or to the complicated transmission dynamics in a networked population. Personally, I continue to think that school closures contributed significantly to reducing transmission.

The decline continued until the week between Christmas and New Year, when cases started rising again, almost explosively: we went from a weekly average of 650 before Christmas to 1.220 after, an almost 90% increase, and this week the trend accelerated, surpassing 6.000 cases yesterday. This could be an effect of winter holidays, combining less testing with more social mingling, but I fear that it’s in fact the start of our Omicron wave. At this growth rate, we may well reach and surpass the previous record by the end of January – or sooner… Officially, the number of confirmed Omicron infections in Romania remains low, with 91 cases confirmed on January 5th (for a total of 183 so far), but worryingly over half of these 91 cases (51) are in vaccinated people.

Coronavirus in Romania: tsunami warning

After a quiet summer, in early autumn the coronavirus has returned to Romania with a vengeance. In the first half of July, the case numbers had dropped under 50 daily, and the total for the entire month was just shy of 2400, the lowest case number since the beginning in March 2020. For a good portion of July, Romania was recording daily deaths caused by coronavirus in the lower single digits, and the monthly total was slightly under 500.

Since then, the situation has changed drastically. Beginning with the second week of July, cases have climbed at a remarkably steady pace, on average 57% each week – except for last week, when the percent increase from the previous week was even higher, almost 90% – bringing us back to a daily count of 7000 cases. Compared to the earlier stages of the pandemic, this growth rate is unprecedented, likely attributable to the more contagious Delta variant, coupled with the lack of restrictions over the summer and the low vaccination rate. Between people returning from holidays, from countries like Greece with higher infections rates, recent large festivals with high attendance, and children returning to school in person last week, the outlook for the next months (even weeks!) is grim.

Coronavirus in Romania: an end in sight?

In the two months since my last update, the situation in Romania has improved significantly. Week after week, cases have declined at a steady pace of 20–30%, down to only 100–200 daily, a level not seen since late May and early June of last year. The number of tests has remained relatively stable, so we can be reasonably certain these case numbers reflect reality. In parallel, the number of patients in ICU is considerably lower as well, confirming this trend: it declined under 900 on May 9^th for the first time in six months, and is now already under 300. Authorities have lifted several restrictions on May 15^th, starting with the curfew in large cities and mask mandates in open spaces, but masks are still required inside stores, public transportation, and crowded places like bus stations.

As expected, the reported deaths have continued to increase in the first half of April, reaching a record of 237 on April 20^th, and then steadily declined at a rate of around 20% each week. On the other hand, the new increases in June, topping 200 daily deaths again on 8^th and 9^th, is only a reporting artifact. Looking into the detailed daily accounts reveals that many of these deaths occurred in previous months and are just now being added to the official statistic for COVID-related deaths. As an example, on June 8^th the official death count was 277, but 256 took place in prior months, going back to June 2020, so the actual death toll for the day was 21. This follows the discovery of inconsistencies between deaths reported by individual hospitals and central government, so I guess this is the solution chosen by authorities to update the number of deceased because of the pandemic. It is a rather odd way of correcting statistics, as it skews both the old reports and the current ones for June and makes it difficult to correlate current mortality with the case numbers.

Reading insights in the Kindle Android app

As the pandemic has disrupted lives and economies across the world, I have felt its effects in an unlikely area of my life: my reading habits. I started working from home in the second half of March 2020 and ever since I have almost completely stopped reading! Part of the explanation is that I did most of my reading on my Kindle during my daily commute on the subway – no more commute, no more reading…

Travel insights from Google Maps Timeline

Speaking of personal tracking for the purpose of creating better products, one of my favorite examples is Google Maps’ Timeline feature. I have discovered its true potential only after switching to Android, as I think on iOS the Google Maps app does not have permanent access to the device location. In a sense, the Timeline is a nicer, user-friendly representation of Location History, which Google happily collects in the background from every Android user.

You may of course turn this feature off or selectively delete portions of this history. Personally though, I love to be able to scroll back in time and see where I was months or years ago (well, three years ago at the most, as little was saved while I was using an iPhone): the place, when I arrived and left, if I was walking, in public transportation or in a car. It is a very convenient way to log my travels and recount them later, in case I would like to revisit those places. For a good portion of time, I also used this to fill my timesheet at the office, as my company’s badge did not correctly register the time I arrived and left the building. For photographers, daily tracks can be exported in a standardized format, which you can then use in Lightroom to add location to photos captured on that day. I use a dedicated app for that, but it’s good to have a backup option in case the app fails or gets discontinued at some point in the future.

Coronavirus in Romania: troubled times

As I have feared when writing my previous update, cases of coronavirus disease in Romania started climbing again following the partial reopening of schools and restaurants. Since the last week of February, cases have increased by 20–25% each week, reaching a peak of around 6650 on March 25^th. Naturally, new restrictions were put in place, stricter curfews in the larger cities during weekends, followed by rounds of street protests and online complaints, and schools will again be closed for the duration of April – officially an extended Easter break. Fortunately, we are already seeing some positive signs, with cases declining slightly last week, by 4%.

This smaller wave has brought with it increased death counts as well, climbing again above 100 deaths daily to a peak of 174 on March 23^rd and 196 on April 6^th. While numbers have remained slightly below the highs recorded during the autumn wave, the mortality rates look worse than before. I am using a rough estimation comparing deaths to cases from three weeks ago, as their evolution appears to track cases with a three-week delay. Calculated this way, mortality in December and January based on official coronavirus deaths dropped to 1.8–2%, while in March 2021 it is back up in the 3.3–3.4% range. Not entirely surprising, considering the increased load on the medical system (the number of ICU patients is constantly climbing and has recently surpassed 1450, above the previous record of almost 1300 in November), and the spread of the new, more lethal UK mutation of the virus.

Coronavirus in Romania: new year, new issues

Contrary to my concerns, the rate of infections in Romania has steadily dropped since my last update, except for a spike in the first week of January that may be a consequence of the Christmas celebrations. Indeed, a similar spike appears in the case evolution of several European countries I regularly check: Poland, Italy, Germany and the UK all reported increases on that week. The current weekly average of new cases is around 2500 here in Romania.

Some people speculate that this reduction is caused by people performing home tests – recently antigen tests have become available for individual purchase – and then not reporting positive results to the authorities to avoid isolation. A disturbing possibility, and a reason why rapid home testing is not a silver bullet against the pandemic – it does not matter if people know they are infected if they do not take proper measures to contain the spread. Nevertheless, daily deaths have also decreased these past two months, to a weekly average of about 80/day, half the average values recorded during November. The number of patients in critical condition has also declined from the November peak of nearly 1300, recently dipping below 1000, although for a single day on January 22.

Coronavirus in Romania: evaluating excess mortality

Since the early days of the pandemic, tracking excess mortality became a valuable tool to evaluate and compare the situation across countries. It should better reflect the overall death toll of the crisis, as strains on the health care system led to a larger number of deaths during this period from unrelated causes, people that would have otherwise received medical care and survived.

Finding the necessary data for Romania proved a bit challenging though. After some searching on the internet, I discovered the site of our National Statistics Institute, which publishes demographic data regularly. Unfortunately for my purposes, the available summary does not offer weekly data, as in many other countries, only monthly. The time series only goes back to 2015, meaning there is less historical data to compare 2020 with. There are delays in publication as well, data is made available around 5 to 6 weeks after the end of the month in question, for example now in December INS released preliminary demographic numbers for October.

Coronavirus in Romania: the third plateau

In stark contrast with the relatively quiet summer months here at home and in Europe, autumn has brought an avalanche of news, most of them bad, but some hopeful as well. The daily reported cases have skyrocketed in many parts of Europe, followed by increased hospitalizations and deaths, and by reports that in some cities, hospital staff are being asked to continue working despite testing positive. Governments have responded by gradually reintroducing restrictions, although not as harsh as the spring lockdowns.

In Romania the number of cases has also steadily increased throughout October, culminating in the first week of November, when almost each day another record was broken, and we surpassed 10.000 cases per day for the first time on November 6^th. The following weeks this upward trend, which started in the second half of September, was finally interrupted: on the second week of the month the number of cases declined slightly, almost 1% compared to the first week, and in the third week it rose again, by a little over 3%. The number of people admitted to ICU is also rising steadily, surpassing 1000 patients on November 4^th. The number of daily deaths peaked at 203 in the second week, on November 11^th, when the weekly death count also climbed over 1.000 for the first time, but at least week-on-week growth slowed last week from 30% to only 7%.

Financial Times: “Covid-19: The global crisis — in data”

Yet for all the patterns that appear in the data, the resurgence also demonstrates the outsized role seemingly played by random chance in the story of Covid-19. When Italy avoided a second wave in the late summer, many pointed to its high levels of mask-wearing as a contributory factor. And indeed, only one country has higher levels of mask-wearing than Italy. That country’s identity? Spain, home to the worst second wave of the continent.

But there is room for hope as the world attempts to stave off a bitter Covid-19 winter. Mortality rates from the virus have fallen over the course of the year as healthcare expertise and technology have improved. In March, somebody aged 70 or above had a 50:50 chance of survival if admitted to hospital with the virus. By August their chances had climbed to 74 per cent.

---

Leaders have vacillated between reopening to protect the economy and locking back down to protect citizens. Yet recent analysis suggests that this may well be a false trade-off.

FT Visual & Data Journalism team

Interesting and extensive series of visualizations about the coronavirus pandemic from the Financial Times. I found the graph below especially revealing: while the correlation is not particularly strong, it is clear that overall countries with larger death counts have also experienced heavier economic recessions. This indicates that protecting citizens from this disease and economic recovery are not mutually exclusive and should be treated as a common goal.