Valerio Lomanto

778 posts

Valerio Lomanto

Valerio Lomanto

@verglasz

Sporadically hopeful humanist.

Stockholm, Sweden Katılım Ekim 2010
270 Takip Edilen103 Takipçiler
Valerio Lomanto
Valerio Lomanto@verglasz·
@TimApplebj @neogoose_btw Yeah if you want to do mutations later etc things can get trickier. Though in this case, pushing onto a vec can reallocate the vec, and thus invalidate the references to its objects. C lets you shoot yourself in the foot in that case, (safe) Rust doesn't.
English
0
0
1
43
Tim Apple
Tim Apple@TimApplebj·
@verglasz @neogoose_btw this is a great snippet, I’ll use this in the future. one issue is that a vector is not actually an arena or arena-like in its semantics and so if you, for example, wanted to push more objects onto buffers after you’ve put the first object in, the borrow checker complains
English
2
0
0
43
Dmitriy Kovalenko
Dmitriy Kovalenko@neogoose_btw·
Not being able to express self referential structs in rust is actually one of the biggest limitation of rust. E.g. I want to pull out a buffer of strings into one continuous deduped arena and store it next to list of items referencing them so I can drop them together which is safe - but I can't express this in "idiomatic" rust like at all
English
15
2
70
9.4K
Valerio Lomanto
Valerio Lomanto@verglasz·
@TimApplebj @neogoose_btw (Also this uses Vec for simplicity of demonstration but it's totally possible to have the lists stored inline in the arena object)
English
0
0
1
29
Valerio Lomanto
Valerio Lomanto@verglasz·
@TimApplebj @neogoose_btw In some more complex cases you may need to write non-idiomatic Rust with raw pointers and unsafe etc, but that means more C-like Rust, which seems a pointless critique if the alternative is C... and you can still encapsulate that in a safe API, which is not an option elsewhere
English
0
0
1
35
Valerio Lomanto
Valerio Lomanto@verglasz·
@keysmashbandit Ah, also people don't understand that "for cheap" is largely downstream of "with little labour input".
English
0
0
0
17
Valerio Lomanto
Valerio Lomanto@verglasz·
@keysmashbandit Because people don't understand that "I get to have things for cheap" and "i get to keep the same job with the same relative wages" are in tension when universalised.
English
1
0
1
147
Valerio Lomanto
Valerio Lomanto@verglasz·
@Aella_Girl Some halo effect is probably going on but another possible explanation(/factor) may be that people don't fundamentally view you as more X than you are but feel allegiance in a way that compels them to vote for you over others in those context (seems likely when other is unknown)
English
0
0
5
1.5K
Aella
Aella@Aella_Girl·
On glosso, you can rate your friends against each other in wars (e.g. most autistic, most 'mommy', kiki/bouba, etc), and they get an elo. My elo is very high in several flattering war types I don't think I should be high on, and I think this is probably because I'm high status in my local group. This freaks me out though - I suspect people are falsely viewing me as better along some axis because of some type of... power/popularity halo effect going on, and I did not realize the extent to which this was affecting perceptions of me
English
16
2
250
44.3K
tess
tess@xsphi·
x-risk, p-doom, h-bomb, d-list, n-word. nothing good ever matches \w-\w{4}.
English
11
3
75
2.1K
Valerio Lomanto retweetledi
Katja Grace 🔍
Katja Grace 🔍@KatjaGrace·
"The solution — this may shock people — must begin with the two A.I. superpowers, the U.S. and China. It is now urgent that they learn to collaborate to prevent bad actors from gaining access to this next level of cyber capability." Yes, but it's even harder than you think because AI systems themselves may be relevant bad actors if they are smart and agentic enough. Not empowering bad actors plausibly means not building them.
Thomas L. Friedman@tomfriedman

My column: Anthropic’s Restraint Is a Terrifying Warning Sign nytimes.com/2026/04/07/opi…

English
4
10
70
4.7K
Valerio Lomanto retweetledi
Eliezer Yudkowsky
Eliezer Yudkowsky@allTheYud·
As a reminder, @robinhanson, if I recall and understood correctly, predicted that no AI company would get far enough ahead of other AI companies that an AI built by them would find enough vulnerabilities to eg steal a bunch of money, before other companies had toughened and secured their code using earlier AIs from other companies that had very nearly all the same capabilities. I replied to Hanson that (1) qualitative step changes in intelligence might produce a lot of newly discoverable vulnerabilities after a step, and (2) I worried about whether all the Internet companies on Earth would actually get around scanning and fixing all those vulnerabilities with earlier AI tools; by way of defending my model in which a nascent ASI could get early fast access to amounts of money on the order of a million dollars by exploiting Internet security flaws. Anthropic is in this case proactively offering their much more advanced tool's scans to other companies, but taking their claims at face value, they could've basically pwned the Internet with Mythos if they had wanted to do that. (Eg, Mythos figured out user-to-admin escalation on Linux among other things.) This corresponds to the kind of capability step-change model that I suggested in the Yudkowsky-Hanson Debate, where it is, indeed, possible for an AI inside a frontier AI company to get ahead of the AI capabilities on offer publicly, and find significant security vulnerabilities not already hardered by other AIs. I predict this will not be the last time this happens, and that Mythos will fail to find seculrity vulnerabilities that future big-jump LLMs do find.
English
7
19
394
36.4K
Valerio Lomanto
Valerio Lomanto@verglasz·
@robinhanson In a way that compiling forms for the optimal disposition of widgets or making sure the sprocket pipelines are appropriately degreased just isn’t.
English
0
0
0
10
Valerio Lomanto
Valerio Lomanto@verglasz·
@robinhanson What about the satisfaction the worker draws from the work? Competing feels meaningful and exhalting, not to speak of victory…
English
1
0
1
66
Robin Hanson
Robin Hanson@robinhanson·
Workers in "soul crushing " jobs at big firms, whose actions are quite constrained by context, still have far more freedom & discretion than do most athletes. Yet few complain that sports is "soul crushing". Why?
English
52
1
213
49.8K
Valerio Lomanto
Valerio Lomanto@verglasz·
@MattiasInSpace This holds for some cases but not others: if you're already using most of the resources then this product may be unusable or perform poorly etc, why shouldn't there be marginal customers where the cumulative impact makes a difference?
English
1
0
1
20
Mattias in Space
Mattias in Space@MattiasInSpace·
@verglasz Consumers notice when an app fails to do what they want it to, but they don't necessarily notice when it marginally steps up the utilization of their machine.
English
1
0
0
28
Valerio Lomanto
Valerio Lomanto@verglasz·
Ok but _why_ is a cost that consumers pay externalised? Why don’t their user choose the cheaper-to-run systems which internalise that cost? Are there specific reasons we should expect this market to be inefficient?
Aakash Gupta@aakashgupta

The reason software eats RAM is the same reason factories used to dump chemicals in rivers. The cost is externalized. Every mass of inference compute shows up on an engineering manager's AWS bill, broken down to the cent, reviewed quarterly. Every mass of RAM consumed on YOUR machine shows up nowhere in anyone's budget. Chrome could cut memory usage by 60% tomorrow and Google's revenue wouldn't move a single basis point. Docker's 2GB idle footprint costs Docker Inc. exactly $0. Electron's 500MB todo list costs the Electron team exactly $0. The user paid for the RAM. The user pays the electricity. The user deals with the fan noise. The company ships faster because they chose the laziest possible runtime. The token-optimization obsession makes this even clearer. Companies optimize inference cost because inference cost hits their margins. They'll spend six months shaving 200ms off a model response. They won't spend six days reducing a desktop client's memory footprint because that memory belongs to someone else's hardware. This is why the 16GB vs 32GB debate is a trap. You're asking consumers to buy more expensive hardware to subsidize the software industry's refusal to optimize for a resource they never have to pay for. The market will never fix this on its own. The people writing the checks and the people running out of RAM are on opposite sides of the transaction.

English
2
0
2
103
hyper(bolic) disco(unting) girl
hyper(bolic) disco(unting) girl@hyperdiscogirl·
i just used a can opener that removes the whole lid instead of slicing thru the inside rim, leaving no sharp edges anywhere wtf, why aren't they all like this! never seen it before today
English
8
0
50
2.7K
Valerio Lomanto
Valerio Lomanto@verglasz·
@rikardhjort [citation needed] Sure engagement farming from afar is crappy but i do like my cosmopolitan online communities
English
1
0
1
38
Valerio Lomanto
Valerio Lomanto@verglasz·
@rikardhjort what i had in mind is that you can read these values from a density map probably, but not quite urbanisation, because small cities aren't very good at public transit either; it's a measure of how split into nothingness vs large metro areas with subways etc is the country
English
1
0
1
87

Keşfet

@TimApplebj @neogoose_btw @aliceisplaying @keysmashbandit @Aella_Girl @xsphi @robinhanson @MattiasInSpace