A useful project related to Apple's Core Audio is qaac - it wraps iTunes Windows DLL's in a standalone encoding tool with a CLI interface. I believe it even works under Wine on Linux: https://web.archive.org/web/20250814194428/https://www.andre... So you don't need a Mac or even a full iTunes installation to get high quality AAC encoding.
Either way you look at it, stablecoins are both better rails and asset than commercial bank deposits - they are cheaper, faster, global, 24/7, and remove unnecessary friction. And if you look at, say, USDC reserves (mostly treasuries or overnight repos) they are structurally better than what your commercial bank has (illiquid mortgages and similar). That said, you don't get the FDIC $250k backstop and for some that might flip the entire conversation (in particular, narrow banking/reserve quality argument doesn't apply under that cap - you are simply fully covered).
>You'd probably have to 4X to 5X the current housing supply to make a dent in prices.
Not at all. The important thing here is that price is set on the margin. Relatively modest supply increases can have outsized price effects. As an example, consider rents in Austin - something like 30% housing stock increase led to 16% drop in median rent, all while Austin simultaneously had massive in-migration, i.e., rents fell despite demand surging.
As another example of margin behavior: vacancy rates matter a lot, a modest vacancy rate increase can crater rents, which we see in the CRE market.
Uber/Lyft takeover had little to do with price (though, yes, they were cheaper) and everything to do with reliability and overall quality of service. Even though ride sharing industry lost money in subsidy arms race and side bets it was fundamentally sound in major metros since early on (similar to how Amazon was fundamentally sound from early on, despite not recognizing profit for a long time). Popular "analyses" kept equating Uber/Lyft with firms losing money on every sale with no path to fix it but the demand was always there as riders had already left taxis and transit on reliability and convenience grounds.
Exactly right. Even though ride sharing industry lost money in subsidy arms race and side bets it was likewise fundamentally sound in major metros since early on. Popular "analyses" kept equating Uber/Lyft with firms losing money on every sale with no path to fix it but the demand was always there as riders had already left taxis and transit on reliability and convenience grounds.
Similar experiences in Boston area. Hailing a taxi at a taxi stand (e.g., at Prudential or Logan) - good experience to this day. Calling dispatcher - half of the time they don't show up (esp. so for scheduled airport rides) or show up late or arrive in a smoke-filled car. Hackney carriage medallions might have been bad investments for some cabbies, but Uber/Lyft are simply a much better service for the customer. Uber/Lyft takeover had little to do with price (though, yes, they were cheaper) and everything to do with reliability and overall quality of service.
It works fine in Chrome 148. Earlier Chrome versions removed chrome://flags versions of the above even with "Temporarily unexpire M147 flags" (and similar), but command line invocation continues to work.
Nice work! FWIW, you can still use Manifest V2 extensions, like uMatrix, uBlock Origin, or Violentmonkey, in Chrome by passing command line flags. For example, on macOS:
open -b com.google.Chrome --new --args --disable-features=ExtensionManifestV2Unsupported,ExtensionManifestV2Disabled
When Google finally nerfs that, it is past time to move to Firefox or Brave, the latter of which has explicitly announced uMatrix support.
Speaking as a European who did his PhD at MIT: that's destructive zero-sum thinking and "outsiders benefit" is backwards.
Knowledge spillover benefits everyone - would there be ASML (Dutch) without DARPA's monumental fundamental research investment in EUV? BioNTech (German) without NIH-funded mRNA research? Without American investment this research likely wouldn't have happened or would have come a decade later.
Phrasing idiosyncrasies are publicly observable and anyone can note - as external observers did in Kaczynski or Hanssen cases - that a particular phrasing is quaint. It is probably true that Twitter is unlikely to run a browser fingerprinting query to de-anonymize someone tweeting spoilers from a softcore porn show. But a potential leaker has to ask: "how sure am I of that?"
There are so many more ways one could screw up, and you only need to screw up once. For example, does X do browser fingerprinting and did you ever use similar setup to use a more identifiable Twitter account? Are you using unique phrasings or behavioral patterns? These things can be solved to a satisfactory degree, but I don't think "it's not hard" captures it - for an average user it _is_ hard.
A major problem with the article is the author's inability to weigh the evidence: actual evidence, like presence/absence pattern, is buried whereas p-hacking stylometry (let me try another expert, this one didn't give me what I wanted! let me feed him the Satoshi/Adam Back tells that I'm already in love with!) is majority of the article. It also includes absolute garbage like the vistomail spoof email during the block size wars. And, oh by the way, both Satoshi and Adam Back knew C++. Theranos evidence was binary (machines either work or they don't) but it is not so here and the author is simply out of his depth here.
It is sad - but entirely unsurprising - that NYT decided to paint a big target on someone's back just for clicks. Judith Miller-tier all over again. Miller too had real evidence and junk evidence, couldn't distinguish between the two, and editors wanted a flashy headline. Carreyrou has exactly the same problem here: NYT editors need multimedia events (like junk stylometry filtering - watch the number shrink from 34,000 to 562 to 114 to 56 to 8 to 1!!!) because that's what its audience-product relationship demands. I think it not unfair to say that modern Times' editorial culture has no mechanism for distinguishing rigorous inference from merely compelling narrative. Open the front page on a random day: how often do you see the Times staking credibility on a causal claim "A causes B" vs simply "X happened. Then Y came." vibes/parataxis.
Indeed, for any particular outcome you could design a custom protocol. In contrast, Radius gives you a generic execution environment. Going custom you lose EVM tooling, audited standard contracts, and every bespoke protocol needs its own security analysis, its own edge case handling, and its own integration work.
That said, the most important loss is composability: an agent can't take your Chaumian coupon escrow and snap it together with someone else's DvP protocol and a third party's rate-limiting wallet in a single atomic transaction. The power of a shared execution environment is that these interaction patterns don't need to be anticipated and designed around in advance. Agents aren't going to negotiate which custom protocol to use for every interaction (the necessary protocol might not even exist!), they will use shared infrastructure where composability of various primitives is the effortless default.
eCash is an answer to "can you move value privately?" (yes, with concessions). Radius is solving "can you execute a million conditional state transitions per second at sub-penny costs?" Blind signatures are still just value transfers and offer no expressivity. If your agent wants DvP with conditional logic, or trustless escrow that releases on programmatic conditions, Chaumian eCash doesn't help. LN gets you close to some forms of atomic exchange, but close-to-atomic isn't atomic, and neither gives you composability. You can move balances, but you can't orchestrate multi-step workflows that can't unwind in the middle.
There is only one type of "like" on X. Since June 2024, all likes (both historical and new) are hidden from profiles, but they aren't fully anonymous: post authors can still see who liked their content (unless the "liker" has a protected account the author doesn't follow).
Bookmarks are the only truly private engagement—no one, including the author, can see who bookmarked a post, though the public count still increases.
A retweet actively redistributes content to your followers; a like signals approval (the author will normally see it) and influences the algorithm without that same direct amplification. Prior to the June 2024 update, your feed also had likes from people you follow.