Pranavvv retweeted
Being 21 is strange. A few months ago, most of my problems were assignments, exams, and placements.
Now I’m thinking about salaries, taxes, career growth, health, and how to make time for things I actually enjoy. Nothing particularly dramatic happened.
Life just quietly changed the syllabus without asking.
Being 21 is strange. A few months ago, most of my problems were assignments, exams, and placements.
Now I’m thinking about salaries, taxes, career growth, health, and how to make time for things I actually enjoy. Nothing particularly dramatic happened.
Life just quietly changed the syllabus without asking.
Pranavvv retweeted
Reddit recently reported that referrals from search engines declined by more than a million per week on average during October. Now, the company is planning to experiment with a primarily logged-in experience in selected countries, partly to prevent its content from being scraped.
And this raises a much bigger question about the future of the internet.
Think about it. For years, millions of Reddit users have been sharing experiences, answering questions, reviewing products, and solving problems. That collective knowledge is what makes Reddit valuable.
But now, AI search engines can take information from those discussions, summarize it, and deliver answers directly to users. People get what they need without necessarily visiting the original website.
The AI platform captures the attention, while the community that produced the knowledge may receive little in return.
Of course, AI isn't necessarily responsible for every decline in search traffic, and some AI companies do pay to license content. But the underlying incentive problem is becoming harder to ignore.
If websites stop receiving traffic, what motivates them to keep their content publicly accessible?
And if more platforms start requiring logins or restricting access to protect their data, we could end up with an increasingly closed internet.
The irony is that AI was supposed to make information more accessible, but it might also be giving websites a reason to make their information less accessible.
That's a pretty interesting contradiction in the future of the open web.
Reddit recently reported that referrals from search engines declined by more than a million per week on average during October. Now, the company is planning to experiment with a primarily logged-in experience in selected countries, partly to prevent its content from being scraped.
And this raises a much bigger question about the future of the internet.
Think about it. For years, millions of Reddit users have been sharing experiences, answering questions, reviewing products, and solving problems. That collective knowledge is what makes Reddit valuable.
But now, AI search engines can take information from those discussions, summarize it, and deliver answers directly to users. People get what they need without necessarily visiting the original website.
The AI platform captures the attention, while the community that produced the knowledge may receive little in return.
Of course, AI isn't necessarily responsible for every decline in search traffic, and some AI companies do pay to license content. But the underlying incentive problem is becoming harder to ignore.
If websites stop receiving traffic, what motivates them to keep their content publicly accessible?
And if more platforms start requiring logins or restricting access to protect their data, we could end up with an increasingly closed internet.
The irony is that AI was supposed to make information more accessible, but it might also be giving websites a reason to make their information less accessible.
That's a pretty interesting contradiction in the future of the open web.
AI has made contributing to open source easier. But reviewing those contributions is getting harder.
A recent Linux Foundation study estimates that 25% of incoming pull requests are low-quality.
And AI-generated contributions are adding to the problem.
🧵
A pull request isn’t valuable just because it contains working code.
It’s valuable when it solves a real problem without creating three new ones for someone else.
More code doesn’t necessarily mean more progress.
every abstraction is beautiful until something breaks and you have to understand the seven abstractions underneath it.
Load balancers, caching, databases, distributed systems, scalability.
Everything that sounds intimidating when a senior engineer starts talking.
A massive free resource for understanding how large-scale systems work.
github.com/donnemartin/syste…
AI can write 10,000 lines of code in minutes.
Unfortunately, someone still has to understand why none of them work in production.
One of the most interesting things about AI right now is knowledge distillation.
A smaller model can learn from a much larger model and become surprisingly capable without needing the same amount of computing power.