RSS Feeds

«Как думаете, вы проработаете в айти до пенсии? Если ещё не работаете, сможете ли найти работу в айти и проработать в этой сфере до пенсии?»

Опрос с таким текстом я опубликовал в своём телеграм-канале. Из 3713 проголосовавших (спасибо дружеским каналам за репосты и охваты) только 21% выбрали вариант ответа «Скорее да». 66% — не уверены. 13% — не в счёт, так как не из айти и переходить в эту сферу не планируют.

Как вышло, что многие айтишники сейчас сомневаются в том, что продолжат работать в айти? Как выделиться на общем фоне, когда вакансий сильно меньше, чем соискателей, рынок зрелый, а исполнители взаимозаменяемы (а также заменяемы ИИ)? Как сохранить себя, если ИИ всё-таки отжал у вас работу в айти?

Songwriter of I Will Always Love You, Jolene and 9 to 5 also invested millions in child literacy, Covid-19 vaccine development and other causes

• Alexis Petridis on Dolly Parton’s peerless career which turned country music cliches inside out

• Dolly Parton – a life in pictures

• Country music legend Dolly Parton dies aged 80 – latest updates

• Share your tributes and memories of Dolly Parton

Dolly Parton, one of the greatest singer-songwriters in country music who was also much cherished for her philanthropy, has died aged 80.

Her death was announced by nephew Brian Seaver on Instagram. “I have imagined the heaviness of this moment but haven’t truly felt it until now,” he said. “It is an honour, an honour that is mixed with absolute pride and great sadness.”

The Republican Senate candidate announced the plan with a graphic flagged as ‘Made with AI’

• US politics live – latest updates

Ken Paxton, a Republican candidate for the US Senate in Texas, has unveiled plans to tackle the “negative impact” of datacenters amid a growing backlash across the political spectrum.

Paxton, endorsed by Donald Trump but facing a close contest against Democrat James Talarico, posted a “Texas first” datacenter plan on X with a graphic that the social media network flagged as “Made with AI”.

Songwriter of I Will Always Love You, Jolene and 9 to 5 also invested millions in child literacy, Covid-19 vaccine development and other causes

• Alexis Petridis on Dolly Parton’s peerless career which turned country music cliches inside out

• Dolly Parton – a life in pictures

• Country music legend Dolly Parton dies aged 80 – latest updates

• Share your tributes and memories of Dolly Parton

Dolly Parton, one of the greatest singer-songwriters in country music who was also much cherished for her philanthropy, has died aged 80.

Her death was announced by nephew Brian Seaver on Instagram. “I have imagined the heaviness of this moment but haven’t truly felt it until now,” he said. “It is an honour, an honour that is mixed with absolute pride and great sadness.”

When the NFL season kicks off next month, millions of Americans will gather around their televisions to watch athletes of comic book proportions crash into each other at breathtaking speeds. Inevitably, those players’ heads will smash into each other, leaving them dazed on the field. And according to a major new study, these repeated head-on collisions may eventually lead to a degenerative brain disease for many — if not most — of the players long after they’ve left the field.

For years now, America has been reckoning with the tension between its love for football and the growing evidence of its dangers. Those concerns can be boiled down to three letters: CTE, or chronic traumatic encephalopathy, which is the progressive brain disease that can lead to dementia, depression, suicide, and other serious mental health problems.

One major question over the past 20 years has been: Exactly how often does playing football lead to CTE? What are the risks for peewee players versus pros? How quickly does the brain deteriorate once the disease takes hold? And new research, published this week in BMJ, has provided us with some staggering answers. After analyzing the death records for hundreds of former NFL players, the co-authors concluded that at least 25 percent of the former players who died between 2016 and 2021 had CTE at the time of their death. 

And the actual prevalence could be much higher, but more players’ brains would have to be given over to scientists for a CTE assessment for us to know for sure. Because among the 338 players whose brains were donated and studied, the CTE rate was an astonishing 93 percent. The question is, according to the researchers: Is that percentage skewed because the players exhibiting CTE symptoms were more likely to offer up their brain for research? Or, as the authors suggest is possible, is the real CTE rate closer to 90 percent? Only further research will be able to tell us, but even the latest estimate — one in four — is cause for concern.

It’s a stark reminder of football’s risks, delivered right on the cusp of a new season for pro, college, and high school teams around the country. And according to the Aspen Institute’s annual State of Play report, there are signs that the game’s future may actually be at risk. 

More and more American parents are steering their kids away from tackle football. How long can an elite league like the NFL last when the pool of players is getting smaller and smaller — at the same time the scientific evidence of its dangers is becoming clearer and clearer?

The youth football pipeline is drying up

Despite the latest research, the NFL is going to be fine in the short term. Professional football is the biggest TV show in the country: Last year, 83 of the top 100 rated TV shows were NFL games, up from 72 in 2024. The next Super Bowl will almost certainly attract more than 100 million viewers and end up the single most-watched TV program in the US for the whole year. Football is one thing that transcends politics in an increasingly polarized America.

But with each new piece of evidence about the physical and mental costs of playing football, football’s long-term dominance looks less and less certain. Already, if you dig into the trend lines for youth participation, they reveal a real threat to the sport’s future.

Let’s start with the good news for football enthusiasts: Flag football has become more and more popular among kids, supported by investments and promotion by the NFL itself and expert consensus that it’s much safer than tackle, especially when it comes to head injuries. According to the Aspen Institute’s State of Play survey, flag football participation among kids ages 6 to 12 grew by 13.6 percent from 2019 to 2024, while tackle football play dropped 7 percent among the same age group over that time.

But it’s participation among the next age group that may be most revealing. According to the Aspen Institute, tackle football participation rates among Americans ages 13 to 17 — the prime age for middle school and high school teams — decreased 9 percent from 2019 to 2024. It has even fallen behind tennis among the teenage cohort. Many of the tweens who play flag football may never strap on a helmet to play tackle.

Even if fewer kids play tackle football year over year, the sport may still subsist on its promises of upward mobility for people who come from poorer backgrounds. With the recent introduction of compensation for college players, elite athletes can make millions of dollars in income before they even reach the NFL. It may be no surprise, then, that the young people still playing tackle football are more likely to be lower-income and live in conservative states. 

Another study found that Black and Hispanic parents, who on average earn less money than their white counterparts, are more likely to see football as a path to college scholarship or even a professional career; the Washington Post reported in 2023 that the proportion of white players at the college level is falling while the percentage of Black players is rising. Football players are often thought of as modern gladiators: participants disproportionately from lower classes aspiring to riches and glory in a violent game for the viewing pleasure of the general public. That’s becoming more and more true as we learn more about its dangers. 

As the annual hoopla around the NFL kickoff makes clear, football’s cultural dominance remains unrivaled among professional sports. But this new study has left even me, a lifelong football fan, wondering how much longer it can, and should, last.

Есть бизнес-процессы, которые на схеме выглядят элементарно. Например, оформление пропуска.

Есть сотрудник. Есть автомобиль. Есть объект заказчика. Нужно проверить документы и оформить допуск.

Что здесь вообще автоматизировать?

Примерно так задача выглядела до тех пор, пока мы не начали разбираться, как этот процесс реально устроен в транспортно-строительной компании на Ямале.

Gang leader threatens to kill hostages after attack in Kenscoff in which homes were also set alight

• ‘A massacre’: Haiti gangs carrying out mass killings

At least 47 people were killed and more than 50 others kidnapped when armed men attacked a once peaceful community near Haiti’s capital at the weekend, the UN has said.

A gang leader has threatened to kill the hostages if authorities kill any gang members as they try to secure the community of Kenscoff, nestled in the hills above Port-au-Prince.

Gang leader threatens to kill hostages after attack in Kenscoff in which homes were also set alight

• ‘A massacre’: Haiti gangs carrying out mass killings

At least 47 people were killed and more than 50 others kidnapped when armed men attacked a once peaceful community near Haiti’s capital at the weekend, the UN has said.

A gang leader has threatened to kill the hostages if authorities kill any gang members as they try to secure the community of Kenscoff, nestled in the hills above Port-au-Prince.

ВНИМАНИЕ! LLM не вовлечен, опечатки и прочие ошибки сохранены.

Предыстория

С самого начала как я попал в enterprise я знал — карьера идет не туда. Кнопки, круды, общаться людьми. Все не то, общаться с машиной интереснее. Но все не было времени к этому подойти. Непонимание как выглядит индустрия, недостаток опыта, неуверенность в себе, переезды.

Когда жизнь улеглась у меня появилось свободное время и скука. Так я начал программировать в свободное время. Очень много, иногда больше чем в рабочее время.

Началось летом 2022.

Children were among victims of Sunday shooting in Billings in case police have characterized as a family disturbance

Multiple people – including children – were killed on Sunday evening in a shooting at a home in Billings, Montana, that was also set on fire, according to authorities and local media reports.

Police have characterized the case as a family disturbance, according to Montana news outlet KULR.

Oliver Blume faces workforce at German HQ as carmaker struggles against Chinese competition and US tariffs

The chief executive of Volkswagen has faced boos and whistles of protest at the company’s headquarters after telling thousands of workers that a comprehensive reorganisation, which could include job losses and factory closures, is vital for its future.

As Oliver Blume urged more than 10,000 workers on Tuesday to “pull together”, saying drastic restructuring measures at the struggling manufacturer were essential, they held up banners and shouted slogans, including: “Our jobs are not your balance sheet adjustments”.

Two years ago, progressive parties allied to see off the far right in a parliamentary election. Ahead of a seismic presidential election, that unity is missing

In the summer of 2024, as polls suggested that Marine Le Pen’s far-right Rassemblement National party would win a snap legislative election rashly called by Emmanuel Macron, the perennially fragmented French left united to spectacular effect. The New Popular Front, whose name evoked Léon Blum’s anti-fascist alliance in the 1930s, came first in a poll that delivered a hung parliament.

Eight months before a presidential contest in which Ms Le Pen is the clear frontrunner, the prospect of a repeat performance appears close to zero. Jean-Luc Mélenchon, the leader of the hard-left France Unbowed, and Olivier Faure, the leader of the Socialist party (PS), have both ruled out the idea of backing a single candidate. Efforts by the Greens to broker talks between progressive parties have fallen on deaf ears.

Do you have an opinion on the issues raised in this article? If you would like to submit a response of up to 300 words by email to be considered for publication in our letters section, please click here.

Tariffs on Canada, threats against Iran’s partners and bond market intervention show a president weaponising US power while making Americans poorer at home

Three moves in the last week have exposed the limits of Donald Trump’s economic bullying. The US slapped 50% tariffs on another $20bn of Canadian goods – then threatened the same rate on the cars, trucks, parts and steel that America Inc depends on. On Monday, the Treasury secretary, Scott Bessent, menaced Iran’s trading partners with exclusion from the US financial network, hoping that coercion can deliver what months of war have not. And as the costs of war, tariffs and pro-billionaire tax cuts helped send US bond yields higher, the Treasury stepped in to tame long-term rates. These are not contradictory policies but a gamble: that US economic power can be repeatedly weaponised without reducing others’ willingness to depend on it.

The Trump administration’s “D-day” sanctions are potentially the most consequential because they expose the scaffolding of American financial power. While markets shrugged off the immediate threat, that calculation might change were the US to target a major Chinese refinery or bank. Such an escalation could spark a trade war with Beijing. The US has leverage. But China has pressure points – especially farmers in Republican states – that make using it costly. In the long run, there is a price to pay. Nations hold one another’s currencies partly because they trust the political relationship. Mr Trump is testing what happens when they don’t.

Люблю родной русский язык. Но слово Вайбкодинг вообще не лезет на скрепные русские щи.

Идея пришла, потому что я вижу реальное смещение в некоторых подходах к разработке проектов. Это все еще можно назвать "запилить что-то" уже нельзя назвать "закодить", так как по факту ты не кодишь. Хотелось придумать веселый и легко произносимый в разных вариациях термин. В результате я пришел к этому варианту.

Короче. Отныне и навсегда предлагаю всем профессиональным сообществом называть это Вибрированием. Вот список возможных вариаций в обиходе.

Навибрировать — создать код (накодить).

Повибрируем — предложение сделать код (покодить).

Навибрировал — создал код (накодил).

Довибрирую — скоро закончу код

Довибрировать — закончить начатый код.

Довибрировал — закончил код.

Перевибрировал — намудрил что-то с кодом.

Завибрировать — начать кодить.

Завибрировал — закодил проект.

Подвибрировать — слегка поправить или дополнить код.

Вибрёжка — процесс молотилки кода. (например: жесткая Вибрежка)

Больше мне сказать нечего. Если есть варианты лучше для удобного обиходного названия - добро пожаловать в обсуждение. Всем удачи!

Home in Billings was also set on fire, with victims representing four generations of a blended family

A man killed eight of his relatives, including four children, on Sunday evening in a shooting at a home in Billings, Montana, that was also set on fire, according to authorities and local media reports.

Police have characterized the case as a family disturbance, according to Montana news outlet KULR.

Research shows young adults often feel isolated. So alongside embracing ‘grandma hobbies’, they want to get involved with the places they live and be good neighbours

Name: Community-maxxing.

Age: The portmanteau is recent, but Homo sapiens have always been social creatures.

Project Lighthouse — Part 3: Introducing project-lighthouse-anonymize

The data in Project Lighthouse is powered by privacy-preserving anonymization code. We’ve put this code into open source, and published two new technical papers detailing the scalable algorithms and data quality frameworks behind it.

By: Adam Bloomston

Introduction

In 2020, we launched Project Lighthouse, which we developed in partnership with leading civil rights and privacy organizations. As our 2020 announcement details, Project Lighthouse enables us to measure potential disparities in user experiences. This work uses perceived race data that is never linked to individual accounts; we only use this data for measuring potential disparities, and users who want to opt-out can do so by turning off the data use settings in their account’s Privacy page. Our results, shared in 2024, demonstrate how we use these analyses to measure our progress in mitigating those disparities.

Earlier this year, we open-sourced project-lighthouse-anonymize, the Python library that powers Project Lighthouse’s anonymization process. To provide the full technical foundation for this work, we also published two new papers on arXiv alongside the code release. Together with our original 2020 paper, these three papers form a complete story: the foundational methodology, the scalable implementation, and the quality validation framework.

The foundational methodology (2020)

Our original 2020 paper established the privacy-by-design approach for Project Lighthouse and provides the rationale for choosing k-anonymity as the technical privacy model to prevent sensitive attribute disclosure at scale. For an introduction to this paper, see our first blog post on p-sensitive k-anonymity and our second blog post on measurement with anonymized data.

Core Mondrian: Scalable partition-based anonymization (2025)

The first of our new papers, Core Mondrian: Basic Mondrian beyond k-anonymity, presents the k-anonymity algorithm at the heart of the open source library. Core Mondrian extends the classic Mondrian algorithm with:

• Extensible architecture, using the Strategy Pattern to support k-anonymity and future privacy model extensions
• Parallel processing, with a hybrid recursive-queue execution model (combining immediate recursive processing for small partitions with queue-based parallel processing for large partitions)
• Other enhancements, including NaN-pattern pre-partitioning (accommodating missing values) and dynamic suppression budget management

This algorithm enables scalable anonymization of large datasets while preserving the ability to use the underlying data for statistical analyses.

Measuring Data Quality for Project Lighthouse (2025)

The second of the new papers, Measuring Data Quality for Project Lighthouse, addresses a critical question: how do you know if your anonymized data are “good enough” for your analysis?

The paper introduces a comprehensive framework for measuring data quality under anonymization, including:

Three primary metrics:

• Pearson correlation (preserving linear relationships between original and anonymized values)
• Revised Information Loss Metric or RILM (measuring how well the “shape” or geometric size of data is preserved — higher scores mean less distortion)
• Normalized Mutual Information v1 or NMIv1 (measuring entropy preservation, essentially how much information content is retained)

Empirical validation methodology: We reframe data quality assessment as a machine learning classification problem, using synthetic datasets to validate that our metrics and thresholds successfully predict when anonymized data will produce statistically valid results

Default thresholds: The paper and library include the specific threshold values we use for Project Lighthouse, which may serve as useful starting points for others implementing similar systems

This framework enables analysts without deep anonymization expertise to confidently assess whether their anonymized data supports valid statistical conclusions.

Getting started

The library is available on PyPI and GitHub at github.com/airbnb/project-lighthouse-anonymize. With this library, you can successively enforce both technical privacy models from our 2020 paper:

p, k = 2, 5
# First technical privacy model: k-anonymity
anon_df, dq_metrics, disclosure_metrics = k_anonymize(logger, input_df, qids, k, {}, "row_id")
# Second technical privacy model: p-sensitive k-anonymity via perturbation
sensitized_df, _, _ = p_sensitize(logger, anon_df, qids, "race", p, k, sens_attr_value_to_prob)

The algorithm for enforcing k-anonymity is described in Core Mondrian: Basic Mondrian beyond k-anonymity. And the data quality metrics and thresholds for k-anonymity are described in Measuring Data Quality for Project Lighthouse:

minimum_dq_met, minimum_dq_met_reasons = check_dq_meets_minimum_thresholds(dq_metrics)
assert minimum_dq_met, str(minimum_dq_met_reasons)

The getting started guide builds on the code snippets above and provides a complete, runnable example using the UCI Adult dataset.

Conclusion

We continue to invest in trying to combat potential discrimination and bias users may face when using Airbnb, and in taking steps to enable everyone in our global community to use and enjoy Airbnb. We believe that doing so requires transparency in our methodologies, both to build trust with our users and to encourage other companies to do the same.

If this type of work interests you, check out some of our related positions!

Acknowledgements

The Airbnb Anti-discrimination & Equity team is Adam Bloomston, Elizabeth Burke, Megan Cacace, Anne Diaz, Wren Dougherty, Matthew Gonzalez, Remington A. Gregg, Yeliz Güngör, Eeway Hsu, Heesoo Kim, Sara Kwasnick, Joanne Lacsina, Demma Rosa Rodriguez, Adam Schiller, Jessica Simon, Maggie Tang, Skyler Wharton, Marilyn Wilcken. I also want to thank Natalija Fijacko, Lauren Mackevich, Laura Rillos, Jessica Simon, Floyd Smith, and Lei Wei for their role in refining and improving this blog post.

The author acknowledges the use of Large Language Models (LLMs) for assistance with literature review, technical writing, and editing.

All product names, logos, and brands are property of their respective owners. All company, product, and service names used in this website are for identification purposes only. Use of these names, logos, and brands does not imply endorsement.

Project Lighthouse — Part 3: Introducing project-lighthouse-anonymize was originally published in The Airbnb Tech Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

Это пересказ моей скопусовской статьи. Оригинал на английском, там более строгие формулировки и немного подробнее. Для прикладных целей вполне достаточно этого обзора.

Конструкции типа look for, look after, depend on часто относят к фразовым глаголам, но в действительности это обычные глаголы с предлогом. Называть их фразовыми – это всё равно, что считать фразовыми глаголами русские сочетания «смотреть за» или «зависеть от».

Словари содержат фразовые глаголы (далее ФГ), состоящие из трёх компонентов. Например, close in on, put up with. Но, строго говоря, третий компонент в них – просто обычный предлог, поскольку ФГ, так же, как и обычные, имеют управление (используются с определёнными предлогами).

ФГ состоит из двух компонентов: собственно глагола (далее «базовый глагол») и образованной от предлога «частицы» (часто её называют «послелог») –  in, up, off и т.д. Частица либо передаёт буквальное пространственное перемещение (come in, walk out, spill over), либо несёт один из исторически более поздних переносных смыслов.

Рассмотрим несколько примеров, в которых обычные глаголы с предлогом путают с фразовыми.

The coin dropped out of my pocket не содержит ФГ: out of здесь – предлог, как и в Mary came out of the house. Но:

I lost the coin – it dropped out.
I am not a student anymore – I dropped out.

В этих двух предложениях drop out – полноценный ФГ с двумя самостоятельными значениями: «выпасть» [из кармана] и «вылететь» [из учебного заведения].

The stain won’t come off the surface не содержит ФГ: off the surface здесь – предложная фраза и off – предлог. Но:

The dress is ruined: the stain won’t come off содержит ФГ come off с самостоятельным значением, отражённым в словарях – «сходить» [о пятнах], а off здесь – частица-наречие.