Pinned Tweet
DeLong🖖@delong
30 Nov 2023
Replying to @delong
What say the remnant denizens of this dumpster fire? Is it time to leave Twitter for good?
50
2
84
49,418
DeLong🖖@delong
12h
Noah Smith has a bull case for Anthropic's valuation. To justify a $2 trillion IPO, Anthropic cannot just win the AI market; it has to win every market. This involves buying a firm in an industry, feeding its data to super-claude, and outcompeting rivals. (1/4)
5
8
166
47,149
DeLong🖖@delong
12h
Replying to @delong
I am a relative bear. There is a low-probability chance of enough future supergrowth to justify the risks in buying in at anything like the IPO price. It depends on whether the secret sauce is that its models are smarter in a useful-general-intelligence sense. (3/4)
2
1
32
6,226
DeLong🖖@delong
12h
Replying to @delong
I cannot evaluate that, since I do not have a good understanding of what that secret sauce is. Read the full post here: braddelong.substack.com/p/crosspost-no… (4/4)
1
2
16
5,648
DeLong🖖@delong
16h
**Not a Good Use of My, or Your, Time, But: CHART OF THE DAY**
Democrats leading in Senate races in Michigan, Texas, Maine, Ohio, Kansas, and Iowa, with a D +11 generic ballot behind them...
braddelong.substack.com/p/not-a-good-u…
1
0
2
3,723
DeLong🖖@delong
19h
x.com/delong/status/…
DeLong🖖@delong
19h
@pruett_jordan @binarybits @dab_chick _Cf._: deepmind/alphazero. _Cf._: deepmind/alphago. The point that R[einforcement ]L[earning with ]V[erifiable ]R[ewards] is different at scale and cost and that that quantitative difference means that RLVR has powerful qualitative emergent effects is, I believe, very smart... 1/
1
0
0
3,212
DeLong🖖 retweeted
Martha Gimbel
@marthagimbel
Nov 1
“My feelings now are much more complex, and if there were a button I could press to undo the last year or two of AI advances—and freeze the technology there for the next several decades—I would be seriously tempted to press it.”
econvitals
@econvitals
Nov 1
✍️ New from Jason Furman: AI Could Do My First Job. Not My Last.

Why I still think AI could do most of what the Council of Economic Advisers did, and what that means for the people who learn by doing that work.

econvitals.org/posts/2026-11-…
0
7
49
4,455
DeLong🖖@delong
Nov 1
x.com/delong/status/…

Clever Hans in the House of Intellect...
DeLong🖖@delong
Nov 1
@MeisMorgan83094 touché... there is something great and powerful to be written about self-play in a restricted domain not just exploring but openng doors to new previously unthought-of rooms in the House of Intellect. Want to write it?
2
0
2
3,204
DeLong🖖@delong
Oct 31
Patrick Stewart died this week at 86. The world is mourning the actor, but this also marks a moment to reflect on the character he made real: Captain Jean-Luc Picard of the USS Enterprise NCC-1701-D. (1/4)
3
15
170
19,285
DeLong🖖@delong
Oct 31
Replying to @delong
Captain Jean-Luc Picard was a commander who would rather understand than fire, who used intelligence to build coexistence, and who was not obsessed with so much winning. Star Trek: TNG is a myth of who we should pretend to be. (3/4)
2
3
58
2,924
DeLong🖖@delong
Oct 31
Replying to @delong
The last quarter century has been brutal to the hopes of those raised on Star Trek: The Next Generation. It is about time to pick ourselves back up and start the journey again. braddelong.substack.com/p/crosspost-no… (4/4)
3
7
95
2,488
DeLong🖖@delong
Oct 31
Barack Obama joked that Donald Trump has the White House looking like Saddam Hussein opened a Cheesecake Factory. Trump was not mad about the Saddam Hussein comparison, but he was very upset about the Cheesecake Factory comments. (1/4)
1
0
21
8,001
DeLong🖖@delong
Oct 31
Replying to @delong
The Wall Street Journal Editorial Board claims it is mean and insulting to the working man of Middle America to point out that the White House looks like a Cheesecake Factory. However, Trump himself was offended by the comparison. (3/4)
1
0
8
1,748
DeLong🖖@delong
Oct 31
Replying to @delong
The Wall Street Journal Editorial Board pretends to think the Cheesecake Factory look is appropriate for public spaces. Read more here: braddelong.substack.com/p/trump-and-co… (4/4)
0
0
3
1,360
DeLong🖖@delong
Oct 31
If your large language model reminds you of a brain, it’s because you’re projecting—not because it’s thinking. It’s not reasoning, it’s interpolation. Anthropomorphizing the algorithm doesn't make it smarter—it makes you dumber. (1/4)
3
16
116
8,947
DeLong🖖@delong
Oct 31
Replying to @delong
A kernel smoother plus a verifier plus iteration is a different kind of object. Claude Code is that function called hundreds of times in a loop. Between calls, a compiler, a test suite, a shell, or a browser tells it whether the last step worked. (3/4)
1
0
9
1,688
DeLong🖖@delong
Oct 31
Replying to @delong
These things have an ability to reason built from interpolation plus outside checking, which is unlike a human mind and yet still powerful. braddelong.substack.com/p/chatgpt-clau… (4/4)
1
0
11
1,379
Load more