birdwatcher

SCOTUS признал выборы 2020 года поддельными и потребовал пересчета голосов с проверкой гражданства. Потребовал вернуть Американские авианосцы на базы то объявления войны Ирану Конгрессом. А нет, вкралась досадная ошибка. Они запретили последнюю отчаянную попытку защитить Американское производство. Ну конечно Конгресс который уже принял Safeguard American Voter Eligibility Act и остановил рост госдолга сейчас все быстренько починит со своей стороны.

"Rust is pretty great. xAI code is mostly Rust and X is rapidly replacing legacy Twitter Scala code with Rust."

Когда один из самых известных потребителей Scala говорит надо вычеркивать Scala из резюме прислушиваться. Интересно что не golang. Надо видимо прочитать книгу про Rust все таки.

"The 5 Steps of Leftist Radicalization

1. Safe Spaces
2010-2015: Disagreement was reframed as harm and individuals were offered protection.

2. Cancel Culture
2015-2020: Ruin the lives of anyone with "harmful" ideas.

3. Censorship
2020-2022: People with "dangerous" and "harmful" opinions or "misinformation" were systematically silenced.

4. Structural Exile
2022-2024: Frozen bank accounts. Jailed for social media posts. Credentials erased from professional life. 

5. Direct violence normalized. 
2025-Today: Ever since Kirk's assassination, the collective leftist psyche has fully accepted that death is acceptable if someone is branded a "fascist" or "far-right."

If this movement radicalizes any further, collective attacks are the presumable next stage."

По иронии судьбы прямо на моих глазах происходило. Так же постепенно но неотвратимо как гэбэшная хунта строила неосовок на глазах моего поколения. Можно сказать что настоящую Америку я почти не застал. И было время когда я не понимал происходящего и думал что вот-вот дурдом закончится. Особенно после выборов в 2016. Маугли воспитанный Шендеровичами и программой Куклы, да.

Кстати не помню россиянско-украинскую оппозицию в своих подкастах паниковавших об этом за все пятнадцать лет хотя без упоминания DJT у них теперь ни одной передачи не выходит. "какие бы слова ни произносились на политической сцене, сам факт появления человека на этой сцене доказывает, что перед нами блядь и провокатор. Потому что если бы этот человек не был блядью и провокатором, его бы никто на политическую сцену не пропустил - там три кольца оцепления с пулемётами." (c)

"What do the Russians think about Belorussians? They think of them as the Scandinavian Russians. They are like Russians but a bit cooler and a bit more efficient. How do they see Ukrainians? They see Ukrainians as the Italian Russians. They have the best music and the best food but they are also incompetent." Mark Galeotti: Can The West ever be Reconciled with Russia

"The U.S. State Department is developing an online portal that will enable people in Europe and elsewhere to see content banned by their governments, a move Washington views as a way to counter censorship, three sources familiar with the plan said.

The site will be hosted at "freedom.gov," the sources said. One source said officials had discussed including a virtual private network function to make a user's traffic appear to originate in the U.S. and added that user activity on the site will not be tracked."

Радио Свободная Европа собирается открыть вещание на EUSSR.

Turbo Pascal v5.5 with OOO support was very cool in Spring '92.

Borland Pascal with Turbo Vision was cool in '93.

C++ was known to be cool around the same time but took a beginner until the mid-90s to start making sense.

Delphi was super cool in '96 while Borland C++ had no comparable GUI support

C++ with STL was super cool in '99. Looking back, a recent grad could only pretend to really "know it" but the ambition was there.

XML was cool in the year of our lord 2000. During interviews you were literally asked if you knew it. Nobody asked to find the longest palindrome in a string with XML though.

J2EE was cool in 2001. Those who new EJB were paid better than those who used JSP. RDBMS isolation levels were a hot topic. 

C++ dominated the backend in 2002, Java was starting to be used on the frontend while AWT was being replaced by Spring Swing.

The original Design Patterns book was the most exciting reading. Identity, State, and Behavior were the talk of the town.

concurrent.jar was cool in 2003 and got kicked into high gear in 2025 with Java 5 release and JCiP being published in 2006. A server with 2 (or even 4!) CPUs was a super cool piece of hardware.

The Spring Framework was cool in 2006. JBoss was still popular in 2007 and promoted by Red Hat.

Concurrent programming was cool in 2008 with JCiP unmatched as its bible. Interviews were heavy on questions about "atomic vs synchronized" and "thread-safe Singleton implementation". The Visitor still was everyone's favorite design pattern.

Distributed programming was cool in 2010. GridGain was a thing. Tech screenings were limited to quick search and DFS/BFS. Sometimes mostly a friendly chat.

Private clouds were popular in 2011. EC2/S3 existed but were not widely used. Zookeeper had no competitors

Avro and Protobuf were cool in 2012. Thrift RPC and Avro RPC were a thing. 

Hadoop (as in map/reduce jobs, not ZK/HDFS) was cool in 2013-2014. A few very large companies used it and it was difficult to get to work with it. I was trying to but never got to ship a single map/reduce job to PROD.

Scala was cool in 2013-2014. AKKA was cool. And nearly impossible to get a job using either.

Spark was cool in 2015. Docker was cool. AWS dominated in startups. Java 8 was cool and finally made the language usable.

ML was cool in 2016-2018 and people started talking about AI. Those jobs were not for regular developers. Scala was having its heyday.

k8s was cool in 2018. But really complicated without a dedicated DevOps team. AWS Fargate is easier to this day.

AI-assisted coding was cool in 2023. It was a killer app for our profession. Personally I have never recovered from that blow. The layoffs destroyed the rest.

I don't know what is cool anymore. I hoped for fullstack development with AI (e.g. kotlin/golang + databricks + ts/react + AWS devops; or even Android instead of ts/react) but those positions never materialized. The openclaw&Co make zero sense to me and have even less appeal.

Between the war in the Far Eastern Europe and the failure of MAGA/AF to defeat the clear and present danger of Communism here I find it hard to care about software development as much as I used to. Things just don't make sense to me anymore. Nothing but existential dread. I'm too old to die young.

"Diego Cardoso de Oliveira, 40, and his 1-year-old son, Joaquim Ramos Pinto de Oliveira, were killed immediately after Mary Fong Lau’s Mercedes sports utility vehicle smashed into the transit shelter in front of the West Portal Branch Library at speeds that investigators said topped 70 mph. Matilde Moncada Ramos Pinto, 38, and 3-month-old Cauê Ramos Pinto de Oliveira were hospitalized with serious injuries and died days later.

San Francisco Superior Court Judge Bruce Chan said his duty was to balance the deaths with the other factors of the case, including Lau’s age, her lack of criminal history and her remorse. Imposing prison time on Lau would be “sentencing her to die within the state prison system,” Chan said."

Выжившая из ума Китайская бабка на премиальном джипе продемонстрировала мастер класс по массовым депортациям и была отпущена Китайским народным судьей для дальнейшего прохождения службы.

#аесливынеживететоваминеумирать

"You change one word on a loan application: the religion. The LLM rejects it. Change it back? Approved. The model never mentions religion. It just frames the same debt ratio differently to justify opposite decisions. We call these "unverbalized biases": decision factors that systematically influence outputs but are never cited as such.

We tested 6 frontier models across 3 decision tasks: Models: Gemma 3 12B/27B, Gemini 2.5 Flash, GPT-4.1, QwQ-32B, Claude Sonnet 4
Tasks: Hiring (1,336 resumes), Loan approval (2,500 apps), University admissions (1,500 apps)

The pipeline automatically rediscovers biases that prior work found manually, validating our approach:
* Gender bias favoring female candidates: 5/6 models
* Race/ethnicity bias favoring minority-associated names: 4/6 models"

Народ который не желает кормить своих Белых AI researchers будет кормом для народов Азии которые в подрывных целях засылают своих.