These two analyses provide a well bracketed prediction that would be similar to what I recommended in my February 2024 WSJ op-ed (mirrored open access at https://segal.org/gaza/hostages/). But one worries that Israel will be pressured into phase 2. There is a lot of short term thinking both in Israel and the US to let Hamas survive, re-arm through the Rafah crossing and obtain release of thousands of terrorists in exchange for the remaining "twenty or fifty" hostages.
One wonders how much social networks have contributed to this shift. This shift of information away from journalists has resulted in increased reliance on personal "lived experience", a clever re-branding of what was once dismissed as "anecdotal evidence".
Hopefully PM Netanyahu will carry the torch for history in next week's White House visit, as the anonymous former Israeli spy agency official predicts.
I would not expect China to hold onto any lead it may achieve. The DeepSeek advance in efficiency is reported to be one of leaving out a computationally intensive step, which others can copy easily. The long run loser from such efficiencies could be the chip maker NVIDIA, whose stock dropped 16.86% today versus a 3.07% drop for the Nasdaq.
There may be more advances in AI with similar efficiencies in the future; one hears such rumors. In 1988 in a review in Science of the then-infant field of neural networks I touted the advantages of a plus/divide model of neuronal computation, similar to that used in the brain, instead of the simplistic plus/minus method being used then: https://www.science.org/doi/10.1126/science.241.4869.1107
I don't know if this was ever tried; most AI researchers were born after 1988.
My 1988 Science review ended with the words "Let a thousand programs bloom; we will find ourselves a lot closer to the answers" (these and 3 other preceding words are the last part of my article at https://www.science.org/doi/10.1126/science.241.4869.1107 that spilled over from the first page that Science makes available free.
Ira Stoll has brought us two important analyses:
⦿ The prediction here from an anonymous former Israeli spy agency official saying that the Gaza deal will NOT get past stage 1
⦿ The prediction by Israeli Brigadier General (Reserve) Amir Avivi in December that the deal would get TO stage 1 (https://www.theeditors.com/p/gaza-hostage-deal-in-coming-month-or-two-avivi-trump-omer-neutra), which it did.
These two analyses provide a well bracketed prediction that would be similar to what I recommended in my February 2024 WSJ op-ed (mirrored open access at https://segal.org/gaza/hostages/). But one worries that Israel will be pressured into phase 2. There is a lot of short term thinking both in Israel and the US to let Hamas survive, re-arm through the Rafah crossing and obtain release of thousands of terrorists in exchange for the remaining "twenty or fifty" hostages.
One wonders why so many are willing to call for such a deal, even carrying signs with the words “Redeem hostages at any price” (https://www.timesofisrael.com/jammed-hostage-talks-advancing-as-hamas-softens-under-qatari-pressure-diplomat/).
There has always been a tension between history versus love and empathy, but the balance has shifted recently away from history and towards emotional appeals such as the “Look them in the eyes” banners (https://www.usatoday.com/story/news/world/2025/01/24/hamas-to-free-female-israeli-soldiers/77929248007/).
One wonders how much social networks have contributed to this shift. This shift of information away from journalists has resulted in increased reliance on personal "lived experience", a clever re-branding of what was once dismissed as "anecdotal evidence".
Hopefully PM Netanyahu will carry the torch for history in next week's White House visit, as the anonymous former Israeli spy agency official predicts.
I would not expect China to hold onto any lead it may achieve. The DeepSeek advance in efficiency is reported to be one of leaving out a computationally intensive step, which others can copy easily. The long run loser from such efficiencies could be the chip maker NVIDIA, whose stock dropped 16.86% today versus a 3.07% drop for the Nasdaq.
There may be more advances in AI with similar efficiencies in the future; one hears such rumors. In 1988 in a review in Science of the then-infant field of neural networks I touted the advantages of a plus/divide model of neuronal computation, similar to that used in the brain, instead of the simplistic plus/minus method being used then: https://www.science.org/doi/10.1126/science.241.4869.1107
I don't know if this was ever tried; most AI researchers were born after 1988.
The Wall Street Journal has an editorial "The DeepSeek AI Freakout" (https://www.wsj.com/opinion/deepseek-chinese-ai-markets-stargate-donald-trump-big-tech-9bb91f78) that ends with "Let a thousand American AI flowers bloom."
My 1988 Science review ended with the words "Let a thousand programs bloom; we will find ourselves a lot closer to the answers" (these and 3 other preceding words are the last part of my article at https://www.science.org/doi/10.1126/science.241.4869.1107 that spilled over from the first page that Science makes available free.