plopilop

I have seen https://pubsonline.informs.org/doi/10.1287/mnsc.2023.00314 as an example.

Despite the limitations in size and speed of today’s quantum computers, our algorithm provides quantifiable liquidity savings when applied to the Canadian HVPS using a 30-day sample of transaction data. By reordering batches of 70 payments, we achieve an average of Canadian (C) $240 million in daily liquidity savings, with a settlement delay of approximately 90 seconds

It has to be noted that the technology used (quantum annealing) is at best erm disputed, and that the company DWave has made very wild claims in the past. Also note that many "quantum speedups" have been de-quantumized, i.e. classical algorithms with equal or even better performance have been developed, sometimes by drawing inspiration from the quantum algorithm. Quantum supremacy is still quite unclear.

Now, could the problems we are talking about be efficiently solved on a classical computer? Maybe. But if nobody knows how to do it, we might just as well use the quantum computer.

I'm not even there for my employer. I have compiled a list of ~20 different inventory/CBOM solutions and I cannot even fathom how to move to the next step other than picking a few vendors at random and ask them for a demo, but the public info I found was not convincing.

While I can assume these tools do a decent job at crypto asset discovery (a `grep -r "-----BEGIN RSA PRIVATE KEY-----"` is not the hardest product to design), I have no idea what to do for code scanning. CBOMkit and friends do not scan C code, which we actually need.

Man, the CBOM is such a pain. There is no standardised format yet (let alone efficient tools for crypto discovery), nobody knew what it was one year ago but now every client is asking ours anyway.

There has been a lot of new stuff over the last few years.

For instance, breaking RSA or ECDSA is requiring much fewer logical qubits than previously thought, and thus fewer physical qubits as well. Progress in error codes, quantum processing etc. made it that in 2019, it was estimated we needed ~20 million noisy qubits to factor RSA 2048. In 2025, we know we need fewer than 1 million. [0]. Some other papers even claim the need of 1000 physical qubits but they rely on a very exotic architecture so I would not consider them feasible.

Progress on the hardware is also continuing, see [1]. Researchers managed to have functional-ish error correction for the first time last year, and experts in the topic are confident that a cryptographically relevant computer will appear in around 15 years.

I personally am less optimistic than the experts (admittedly I am not an expert either), but there is enough activity to get worried for critical infrastructure.

Regarding the factoring issue, as you point out factoring 15 and factoring 21 are two very different tasks. The first one can be used to show that your quantum computer is indeed doing quantum computation; the second will prove that you have a functional error correcting code. If you can factor 21, it is probably only a matter of months/maybe a few years until you factor RSA 2048. As Scott Aaronson said [3], "Once you understand quantum fault-tolerance, asking “so when are you going to factor 35 with Shor’s algorithm?” becomes sort of like asking the Manhattan Project physicists in 1943, “so when are you going to produce at least a small nuclear explosion?”"

[0] https://arxiv.org/abs/2505.15917

[1] https://sam-jaques.appspot.com/quantum_landscape

[2] https://globalriskinstitute.org/publication/quantum-threat-t...

[3] https://scottaaronson.blog/?p=9665#comment-2029013

Y2K devs had it easy, they knew the bug would arrive and when.

The problem with PQC is not that nobody knows when a CRQC (cryptographically relevant quantum computer) will appear, but that by the time it appears, you are already ~10 years too late for migrating (5 years of migration time and 5 years of your adversary silently storing all your classical crypto messages to decrypt them at a later time, the "harvest now decrypt later" attack).

Of course the HNDL attack is only relevant for the most critical pieces of infrastructures, 99% of companies are not a real target for that, especially given the storage cost of such an attack.

There is also the "trust now, forge later" attack, in which a CRQC could break a chain of trust (i.e. digital signatures), and that attack does not need any storage besides the logs of past messages. If you want to guarantee authenticity and unforgeability of your logs for, say, 20 years, you better hope that no CRQC appears by 2050 at least. Once again, it only concerns maybe 1% of companies.

But hey, these 1% companies are exactly the ones that are needing specialised crypto equipment so the move from ANSSI tracks.

I personally do not believe a CRQC will appear before 2050 either. I am willing to bet some money on it, despite researchers in quantum computers being quite confident it will appear in the next 15 years, but I am not willing to bet the entirety of Internet security on it.

Don't quote me on this but I once heard that in the EU you always need to confirm purchases, ie. you cannot have the Amazon 1-click system.

But in general, yes, having a confirmation dialog on important buttons is good UX; similarly as to setting minimum/maximum saturation will increase accessibility by visually impaired people, epileptic people etc.

Pretty sure this is wrong. Code pénal, art. 434-25, translated with deepl.

"Any attempt to publicly discredit a judicial act or decision—through actions, words, writings, or images of any kind—under circumstances likely to undermine the authority or independence of the judiciary is punishable by six months’ imprisonment and a fine of 7,500 euros.

The provisions of the preceding paragraph do not apply to technical comments or to acts, statements, writings, or images of any kind intended to seek the reversal, annulment, or review of a decision."

You are free to disagree with the ruling, but you cannot say that the trial is a parody of justice, the judge biased and the whole thing a conspiracy (cough cough Sarkozy)

My point is that flashy presentations already were an issue before the rise of LLMs. The evaluation of a hackathon relies on presentations and subjective opinions rather than pure benchmark of technical assets, there is nothing new under the sun here.

You can now win with 20 skills.md files now, you could win with "it would be great to use this sexy tech" 10 years ago.

10 years ago I won the only hackathon I participated in (not by choice). The jury was especially impressed in our report with the AI section. That part was a bunch of technobabble that I wrote, more or less saying "in the future the system should do this and that", quoting some popular algos from that time. None of it was implemented in our demo in any way, shape or form. They checked that I knew what I was talking about, and talking with confidence was all it took.

We did not even try to win the hackathon, just to get a passing grade.

Seems like the blog succumbed to the HN hug of death (`Actioning this file would cause "jbkempf.com//blog/2026/dav2d/" to exceed the per-day file actions limit of 160000 actions, try again later`), is there a copy available somewhere?

You usually do secret sharing in a finite field because computers don't like real numbers. The size of your share is a point (x, y), x can be small (typically log n in case of n participants), y is a random point in the field.

Since Shamir Secret Sharing is information-theoretically secure (if you do not know k points from the k-out-of-n secret then all secrets are equally plausible even when bruteforcing), the bitsize of your field can be whatever you want (but obviously bigger than the bitsize of your secret, you can't hide 100 bits in a finite field of 5 elements).

Usually, you don't want an attacker to be able to bruteforce your secret (while the scheme is ITS, your secret typically isn't, e.g. the seed of your wallet), hence randomness can be added to your secret and the bitsize of the field is taken big enough to thwart these attacks.

Depending on your attack model, an 80-bits or 128-bits field is more than secure enough, hence a share bitsize slightly above 80 or 128 bits.

And regarding quantum computer, since the scheme is ITS no attacks can exist.

What I write down is usually a quite literal dump of my brain. I have a problem, and rather than keeping it in my head, I write it down, and force myself to continue writing about the topic to force myself trying to find solutions, rather than be obsessed with the question.

Example: "I need to solve problem A. Problem A can be formulated in this way. This way is similar to a project I did a few years ago, if I remember correctly I had done B and C. However B would not work in the current situation, but would it not though? The issue is that it clashes with component X and Y. What about C? Hmm maybe but I needed approval from Z." etc. All of these thoughts are written down, without filter.

Forcing me to write down has two effects. The first one, slow down my thoughts, because discarding idea B after only 0.1 second of consideration is not productive if you do not explicitly think about why it is a bad idea, and consider the bad idea anyways. The second one is that writing down (especially manual writing and not keyboard typing, for reasons I cannot explain) allows you to think more deeply about your ideas, to envision it in different ways, not only the first way that popped to your mind. I think that keyboard writing requires too much of my brainpower compared to handwriting.

Moreover, in these sessions, having the possibility to look back to a previous idea immediately is extremely useful, and cannot be attained if you use an erasable surface rather than a notebook.

I have to say though that I very rarely look back to what I wrote after the session took place, unless I need to get back to the exact same problem.

I mean, the example you link is probably an engineer doing their job of signalling to hierarchy that something went deeply wrong. Of course, the lack of action of Facebook afterwards is a proof that they did not care, but not as much as a smoking gun.

A smoking gun would be, for instance, Facebook observing that most of their ads are scam, that the cost of fixing this exceeds by far "the cost of any regulatory settlement involving scam ads.", and to conclude that the company’s leadership decided to act only in response to impending regulatory action.

https://www.reuters.com/investigations/meta-is-earning-fortu...

A long time ago I read "the good Brahmin" from Voltaire: https://www.online-literature.com/voltaire/4411/

Basically, the story goes that the good brahmin, for all his wealth and intelligence, is miserable, whereas the stupid beggar down the street is very happy. While the brahmin accepts that the beggar is objectively happier than him, he would never swap places with her.

It made me realise that the quest for intelligence is fundamentally different from the quest for happiness, and even to this day I still take the story in consideration when making life choices. I do not believe that intelligence forbids happiness, simply that if you spend too much time trying to be right, you don't spend enough trying to be happy. Of course trying to be right can make you happy, but in the general case you always need to remember to take a step back.

You can defend a lot of atrocities by arguing "for the greater good" and comparing to uchronic hypotheticals. I could as well argue that without Rome, the greek democracies would have been much more prevalent, and lead to modern democracies much sooner. Or that a world leader would have emerged, leading the ancient world to endless peace and prosperity.

Agree. I tried the first 3 examples:

* "Rubber bouncy at Heathrow removal" on Google had 3 links, including the one about SFO from which chatGPT took a tangent. While ChatGPT provided evidence for the latest removal date being of 2024, none was provided for the lower bound. I saw no date online either. Was this a hallucination?

* A reverse image lookup of the building gave me the blog entry, but also an Alamy picture of the Blade (admittedly this result can have been biased by the fact the author already identified the building as the blade)

* The starbucks pop Google search led me to https://starbuckmenu.uk/starbucks-cake-pop-prices/. I will add that the author bitching to ChatGPT about ChatGPT hidden prompts in the transcript is hilarious.

I get why people prefer ChatGPT. It will do all the boring work of curating the internet for you, to privde you with a single answer. It will also hallucinate every now and then but that seems to be a price people are willing to pay and ignore, just like the added cost compared to a single Google search. Now I am not sure how this will evolve.

Back in the days, people would tell you to be weary of the Internet and that Wikipedia thing, and that you could get all the info you need from a much more reliable source at the library anyways, for a fraction of the cost. I guess that if LLMs continue to evolve, we will face the same paradigm shift.

None of these were in your initial law. Furthermore, karma is also addictive.

You have banned Google, Reddit, and HN.

Oh, I know that strong emotions increase engagement, outrage being a prime candidate. I have also no issue believing that FB/TikTok/X etc aggressively engage in such tactics, e.g. [0]. But I am not aware of FB publicly acknowledging that they deliberately tune the algorithm to this effect, even though they carried some research on the effects of emotions on engagement (I would love to be proven wrong though).

But admitting FB did publicly say they manipulate their users' emotions for engagement, and a law is passed preventing that. How do you assess that the new FB algorithm is not manipulating emotions for engagement? How do you enforce your law? If you are not allowed to create outrage, are you allowed to promote posts that expose politicians corruption? Where is the limit?

Once again, I hate these algorithms. But we cannot regulate by saying "stop being evil", we need specific metrics, targets, objectives. A law too broad will ban Google as much as Facebook, and a law too narrow can be circumvented in many ways.

[0] https://www.wsj.com/tech/facebook-algorithm-change-zuckerber...

Sooo... Should we ban Google too? It is also ordering the contents of its research results with algorithms. Similarly, HN and reddit order the contents of their front page with some algorithms, and in the case of Google and Reddit, the algorithm is personalized with the user's preferences.

Or do we only ban websites that design their algorithms to trigger strong emotional emotions? How do you define that? Even Musk doesn't go around saying that the algorithm is modified to promote alt right, instead he pretends it is all about "bringing balance back". Furthermore, I would argue that systems based on votes such as Reddit or HN are much more likely than other systems to push such content. We could issue a regulation to ban specific platforms or websites (TikTok, X...) by naming them individually, but that would probably go against many rules of free competition, and would be quite easily circumvented.

Not that I disagree on the effect of social medias on society, but regulating this is not as easy as "let's ban the algorithm".

Websites use ready-to be used cookie banners provider by their advertisers. Who have all the incentive to make the process as painful as possible unless you click "accept", and essentially followed the model that Facebook pioneered.

And since most people click on accept, websites don't really care either.

Cookie consent popovers were the deliberate decisions of company to create the worst possible compliance. A much simpler one could have been to stop tracking users especially when it is not their primary business.

Newer regulations also mandate that "reject all cookies" should be a one click action but surprisingly compliance is low. Once again, the enemy of the customer here is the company, not the eu regulation.

As I understand it right now, HME is a weaker form of encryption, but perhaps still strong enough to be a worthwhile tradeoff for the use cases being discussed.

Exactly. Homomorphism was first seen as a weakness in encryption, since it implies malleability. For instance, in the one-time pad encryption where you XOR your message with the secret key, flipping a bit in the ciphertext will result in same bit being flipped in the decryption. The attacker does not know what the end result is, but knows that the bit has been flipped, hence OTP encryption is malleable. This is enough for some attacks. With FHE encryption you have a bit of the same, from Enc(a) and Enc(b) it is easy to create Enc(a+b), hence is malleable too.

Cryptography uses several security levels. The top one for encryption is NM-CCA2 (non-malleability under chosen ciphertext attack). For instance, RSA-OAEP is NM-CCA2 secure. Since FHE schemes are malleable, they are not NM-CCA2 secure. However, a slightly lower security notion is IND-CPA (indistinguishability under chosen plaintext attack). FHE schemes are IND-CPA secure. Furthermore, IND-CPA security is shown to be equivalent to semantic security, which means that given a ciphertext the attacker cannot know any bit of information about the underlying cleartext.

Hence, FHE schemes guarantee that for all the ciphertexts they receive, the attacker cannot know anything about the underlying cleartexts. You can run a ton of operations on the ciphertexts, let's say run a homomorphic LLM, the attacker will still have no idea about what the final output is. Hence, in the model where you consider that the attacker has full control over the LLM, will behave honestly but will try to learn your secrets, you are fine. However, in the model where an attacker runs a MITM and just wants to disrupt the numbers you get back from the LLM, then you are not fine, since this encryption is malleable (in theory we could add some verifiable execution proofs but that is another topic).

As you say, everything is a tradeoff.

I mean it seems clear that Meta did not pirate the content to watch/read it. However, I guess according to the ruling you could pirate anything you want (but no seeding), produce a shitty haiku based on what you pirated and then claim fair use.

As the article mentions, fully homomorphic encryption is insanely slow and inefficient. But I have to say that it is a relatively new field (the first FHE scheme was discovered in 2009), and that the field has immensely progressed over the last decade and a half.

The first FHE scheme required keys of several TB/PB, bootstrapping (an operation that is pivotal in FHE schemes, when too many multiplications are computed) would take thousands of hours. We are now down to keys of "only" 30 MB, and bootstrapping in less than 0.1 second.

Hopefully progress will continue and FHE will become more practical.

How is that any different than people programming for fun?

The hegemony of software only accepting . has de facto pushed the standard everywhere for computers, but here in France I still write with a comma, but type with a dot.

A few years ago Excel and some other softwares started to be locale dependent and I never wanted to burn my computer this much

When have gun laws ever stopped a shooting?

Have you heard of Australia? https://www.sydney.edu.au/news-opinion/news/2018/03/13/gun-l...

The cost of rent is 30€ per square meter per month in Paris.

Only for big enough apartments. 20m² appartements do not go for under 850€, ie around 42€/m². Granted these flats are too small to even have a laundry machine, let alone a laundry folding robot.

Switching one uppercase letter in an otherwise lowercase password adds exactly 1 bit of entropy (per switched letter).

On the other hand, any extra lowercase letter will increase the entropy by 4.7 bits (assuming a password on [a-z]).

Given that most passwords have at best 2 uppercase letters, I would argue it is safer to force longer passwords than passwords with at least one uppercase letter.