Karthik Puthraya

@_krthk

Aspiring rationalist

Mountain View, CA

Joined March 2009

233 Following

175 Followers

225 Posts

Karthik Puthraya @_krthk

9 days ago

@dollarsanddata I enjoy your writing, Nick, but the ad sprawl on your blog makes the reading experience less than ideal. Do you really need the ad money at the expense of your reader's experience?

_krthk retweeted

Sridhar Vembu

@svembu

over 1 year ago

I am not alarmed about the rising talent migration from India. Here is why. First, historic patterns. Talent migration from Korea/Taiwan to the US peaked in the 1970s and 80s (Japanese migration to the US peaked much earlier), talent migration from China peaked about 10 years ago. That was the time of rapid development and transformation of those nations. When a country is extremely underdeveloped, talent migration is very low. When a country is very wealthy, talent migration is also very low. Talent migration peaks during the period of rapid development. This is observable in silicon valley where migrants from India now well outpace those from China, Hong Kong and Taiwan. Why does this happen? Rapid economic development is preceded by and happens due to rising aspiration levels; and talent migration also happens due to rising aspiration levels. Our aspirations necessarily race ahead of reality and it is precisely during the period of rapid growth that we have the most restless people, and it is the restless spirits who tend to migrate. We can observe that restlessness in India, particularly in urban tech centres. Eventually when the full economic potential is realised, talent migration slows down or stops entirely. This pattern has been seen repeatedly. As late as the 1980s, Japanese still refused to believe Japan had achieved parity with the West. Same with Korea in the late 1990s. I predict that even in 2035 with a per capita GDP upwards of $10K, Indians will likely lament how far behind we are. The last stage of catch up happens quickly over a decade (witness China is the last decade) but the foundations have to be laid over many previous decades and those previous decades feel like a slog. We are slogging through that now in India. To summarise, rising talent migration from India is not a symptom of hopelessness, it is the opposite. Be an optimist! 🙏

315

830

887

492K

_krthk retweeted

steve hsu

@hsu_steve

over 1 year ago

Re: H1B MAGA debate I've helped numerous physicists and engineers get O1 (alien of extraordinary ability) and H1B visas. The former tend to be PhDs (eg Russian theoretical physicist from top university) and the latter tend to be engineers with more applied backgrounds. I also helped @Dominic2306 Dominic Cummings and No 10 Downing St. reform skilled immigration to the UK. The following are all true statements: 1. O1 threshold is high. People who get this are usually PhDs from top programs who have published significant research and are internationally recognized. Stealing these people from other countries is great for the US economy, bad for their home countries, and has a negative effect on job prospects for really talented Americans (let's not kid around). 2. H1B: There are many super talented people that US companies would like to bring here, and the only way to do it is to enter the horrible, unpredictable H1B lottery. (Example: graduating CMU MS AI/ML students from India that we interviewed at Superfocus.) These people generally won't qualify for O1 but are still super valuable. Similar impact on economy and jobs as #1 above. 3. There are lots of NOT TALENTED H1B workers brought in by US companies simply to reduce costs - bc they can do the work more cheaply, are less likely to leave for a better job, etc. Any mid-level IT worker in the US is familiar with this situation - they are not making it up. H1B could be converted from a crap lottery system to a merit-based system more along O1 lines. This would require more resources, for specialized review of applications, but would be worth it to the US economy. OF COURSE immigration hurts US workers that have to compete with the new workers. It increases supply and lowers the local value of labor. This is true for both high- and low-skill work. (Don't believe crap studies that claim otherwise - ie by "economists" based on weak data and influenced by ideological priors.) But high-skill immigrants (categories 1&2) make the US economy (eg SV) more competitive and able to WIN important tech competitions. Some of these competitions lead to WINNER TAKE ALL MONOPOLIES with returns to scale (eg FAANG). Or, they let the US dominate a specific industry, set standards, leap ahead in R&D, etc. PRC has as much talent as the Rest of the World combined - significantly more than the US produces domestically - SEE BELOW. If US tech companies want to compete with Chinese counterparts, they NEED to import top talent in categories #1 and #2 or they have NO HOPE OF WINNING. Take AI as an example. Probably ~50% of all top AI researchers in the US come from China, and the remainder are drawn from all over the world. Only a minority are US-born. (Source: Global AI Talent Tracker.)

hsu_steve's tweet photo. Re: H1B MAGA debate

I've helped numerous physicists and engineers get O1 (alien of extraordinary ability) and H1B visas. The former tend to be PhDs (eg Russian theoretical physicist from top university) and the latter tend to be engineers with more applied backgrounds.

I also helped @Dominic2306 Dominic Cummings and No 10 Downing St. reform skilled immigration to the UK.

The following are all true statements:

1. O1 threshold is high. People who get this are usually PhDs from top programs who have published significant research and are internationally recognized. Stealing these people from other countries is great for the US economy, bad for their home countries, and has a negative effect on job prospects for really talented Americans (let's not kid around).

2. H1B: There are many super talented people that US companies would like to bring here, and the only way to do it is to enter the horrible, unpredictable H1B lottery. (Example: graduating CMU MS AI/ML students from India that we interviewed at Superfocus.) These people generally won't qualify for O1 but are still super valuable. Similar impact on economy and jobs as #1 above.

3. There are lots of NOT TALENTED H1B workers brought in by US companies simply to reduce costs - bc they can do the work more cheaply, are less likely to leave for a better job, etc. Any mid-level IT worker in the US is familiar with this situation - they are not making it up.

H1B could be converted from a crap lottery system to a merit-based system more along O1 lines. This would require more resources, for specialized review of applications, but would be worth it to the US economy.

OF COURSE immigration hurts US workers that have to compete with the new workers. It increases supply and lowers the local value of labor. This is true for both high- and low-skill work. (Don't believe crap studies that claim otherwise - ie by "economists" based on weak data and influenced by ideological priors.)

But high-skill immigrants (categories 1&2) make the US economy (eg SV) more competitive and able to WIN important tech competitions. Some of these competitions lead to WINNER TAKE ALL MONOPOLIES with returns to scale (eg FAANG). Or, they let the US dominate a specific industry, set standards, leap ahead in R&D, etc.

PRC has as much talent as the Rest of the World combined - significantly more than the US produces domestically - SEE BELOW.

If US tech companies want to compete with Chinese counterparts, they NEED to import top talent in categories #1 and #2 or they have NO HOPE OF WINNING.

Take AI as an example. Probably ~50% of all top AI researchers in the US come from China, and the remainder are drawn from all over the world. Only a minority are US-born. (Source: Global AI Talent Tracker.)

295

610

_krthk retweeted

Sudhish @sudhishkr

almost 4 years ago

I am volunteering at @codepath [501c3 non-profit] this summer - to help CS students entering industry. If you can spare a little time this month (1-2 hrs/week) then please signup and help out. https://t.co/5IEOtLrAr6.