Software Engineer. Alumni of Arizona State University. Interested in life stuff, AI, science, and video games
United States
Joined April 2009
- Tweets49.8K
- Following481
- Followers1.2K
- Likes376K
sukru tikves retweeted
Exodus
Coming to Xbox Series Consoles 2027
Mass Effect fans, if you somehow have never heard of this game you need to.
Former BioWare and other Industry vets are working on this.
🪐
Even then what you get is a "code point", not an actual character, let along a glyph
Two code points can make a single character, or a later one can change the prior
It is so much fun! And makes text processing inherently O(N^2) for simple operations.
en.wikipedia.org/wiki/Code_p…
I was thinking the Academia in Computer Science was bad.
But then I read some highly cited articles from other disciplines with 6-8 "prestigious authors"
If they received an actual audit, not only article would be retracted, their phd jury would be stripped of professorship.
Glenna did an unusual thing that people typically don't do: she checked a citation in a law review article to see whether the citation said what the authors claimed it did.
She found that the citation said the opposite of what the authors claimed it did.
She then asked the editor of the law review for a correction.
The editor did not reply to the request.
She then asked the authors for comment on the incorrect citation.
The authors acknowledged that the citation was erroneous and regretted the error.
The authors then insisted that the citation did not matter to their actual argument because nothing in their argument depended on the citation being right.
Glenna then wrote a post explaining why the erroneous citation was in fact important to their actual argument, and why the erroneous citation also exposed a wider ignorance of the subject on which the authors were writing. (She claims, convincingly, that it's important to know what sex is in order to understand what sex discrimination is, and that being able to define what sex is is an inextricable aspect of knowing what sex is.)
I consulted with some lawyer friends in a chat group who helpfully explained to me that no one ever reads footnotes, no one checks the underlying references, and no one cares if they are correct or not because law review articles like the one Murray and Shaw published are merely political advocacy masquerading as legal scholarship that no one reads, and therefore not to expect the Supreme Court Review to issue a correction for a citation that the authors are on the record admitting was erroneous.
Do I have all that right, @williambaude?
Who would think a professor from Stanford does cannot properly separate correlation and causation
"We have discounted these 3 trivial variables, so it cannot be correlation, and a causal relationship is proven" (paraphrasing)
That is not how it works.
Here is a simple breakdown on why ME:5 unfortunately would not work financially at this point
$5 million is just the money they would pay to game directors (+benefit +overhead)
There is a Turkish saying
"He either does not know how to count, or did not receive spanking before"
Polygon’s editor thinks they can finish Mass Effect for $5 million, because that is what salaries would cost for 50 people at $50k per year for 2 years.
First of all, the team would have to scale up considerably to build a game of this scale — the credits for “Dragon Age: The Veilguard” has 3,671 people listed.
Second, most of these people make a lot more than $50k per year.
Third, this game was a lot more than two years from shipping.
Fourth, even if you imagine a “Mass Effect” game can be made by 50 people who all make less than an assistant manager at a Chipotle, the $5 million covers salaries, but not benefits, payroll taxes, office space, equipment or any of the other costs associated with running a company.
The cost of finishing the “Mass Effect” game is in the neighborhood of $200 million. If they could finish it for $5 million, they would have done it.
The Atlantic is finally writing about the failure of Communism in Cuba. It signals the things are really bad.
(Most so called "intellectuals" and "journalists" do not mention failures until very late. What was it? "No true Socialism")
theatlantic.com/magazine/202…
We could have said the same thing 30-40 years ago. Some others would point to the first day Communists took over.
It never ends well, but the signal is delayed
(There is only one Communist country that did not collapse, GDR, which voted to join actual democratic Germany)
"Yapay zekanın en hızlı değiştireceği meslekler" dese daha doğru olur.
Ilk sıraya psikolog yazilmis. Daha Chat GPT "bebe" iken (3.5) insanlar onu kisisel psikolog olarak kullanıyordu.
Klinik karar alma?
"Expert system" diye birsey var. Doktorlar ne kullanıyor zannediyorsunuz?
Hakim?
Afet uzmanı?
Neyse... Anlaşılan o ki insanlar yapay zekanın ne yapabilecegi, hatta onlarca yıldır neler yaptıgı konusunda bilgi sahibi degil.
"Entry level" denilen, giris seviyesindeki *tum* mesleklerde degisim bekleyin. Uzmanlık farklı bir durum, insanlik toplu olarak yukarı doğru itilecek.
While catchy, this is incorrect at almost every level.
"Open Weight" implies Open Source, but many so called Open models are just "Shared Source" / "Source-Available" (look, but don't use with heavy restrictions)
(None of Moonshot AI models are truly open for example. ZAi is roughly 50%.)
Also for their credit xAI had 2 models in the past (Anrophic is still zero)
OpenAI has a very large catalog, but it is "old heavy", everything from CLIP to GPT2 are there.
Google is similar.
Meta? 3% Really?
Aren't they the ones that started the open models revolution with Llama?
(I might have missed a few things, late night, feel free to correct)
How it started vs how it's going
Readers added context they thought people might want to know
xAI has published the weights of two of its models, Grok-1 and Grok 2, on Hugging Face, so the 0% figure shown for xAI is incorrect. huggingface.co/xai-org/grok-1 huggingface.co/xai-org/grok-2
This is a big step, but I would want to focus on the response from CS community on this news
(Dis)proving a major hypothesis in the literature is welcomed, and brings excitement
Very different that other disciplines that outright reject AI usage in solving age old questions
So, AI has produced a disproof of the Exponential Time Hypothesis.
Is it just me, or is everybody on the planet who knows anything about computational complexity wondering the same thing tonight: could this be the universe's way of whispering that P=NP?
Nah, I don't think it's just me.
I still think it's more likely that P≠NP. But my odds on that have shortened dramatically in the last 4 hours.
While this is great news for good employees...
I find "buyouts" strange. As anyone who already had a job lined up, or can get interviews easily is more inlined to take advantage of it
But a lower performer will choose to stay
(There is a wide range in between, I know)
Microsoft’s new Voluntary Exit Offer
Eligible employees at level 70 and below (i.e., roughly below VP) on participating teams in the US, Canada, the UK, and India are being offered financial support and career-planning resources if they choose to leave voluntarily.
Reports put the pool at about 60,000 people, roughly a quarter of the company. Not every organization is included; it varies by business priorities, customer commitments, and workforce needs.
Microsoft is clear that this is a voluntary program, not a buyout, and not a replacement for layoffs (which remain an option depending on conditions).
This comes after roughly 15,000 cuts in 2025 as the company continues heavy AI/infrastructure spending.
mynorthwest.com/money/micros…
