HNHacker News
TopNewBestAskShowJobs

unholiness

1,306 karma · joined August 20, 2013

submissionscomments
unholiness··on Algorithm edits every sales demo into a highlight-reel video
The target first user is a sales rep, not a CEO. It doesn't seem far off the mark.
unholiness··on Algorithm edits every sales demo into a highlight-reel video
For American football you can actually see every play back-to-back in ~40 minutes for a 3 hour game. NFL Gamepass releases these cuts 24 hours after.
unholiness··on Algorithm edits every sales demo into a highlight-reel video
To learn usability testing you'll probably want to jump straight to "Rocket Surgery Made Easy", by the same author of "Don't make me think"

Steve Krug is a former colleague and brilliant guy, but at this stage learning how to test and hear from your users in practice is time 10x better spent than just learning the theory.

unholiness··on AI Generated Seinfeld runs 24/7 on Twitch
You've been watching the comedy show this whole time
unholiness··on Google to reduce workforce by 12k
Possibly naive question:

Everyone's talking about how these companies are preparing for a global recession. But it's striking how specific to large tech companies. Is it possible these companies are just seeing the new remote-first paradigm as an opportunity to replace US devs with cheaper, foreign developers?

Obvious this is absurdly difficult, but it's likely more realistic now than it's ever been before.

unholiness··on MacBook Pro featuring M2 Pro and M2 Max
I used a 2013 MBP for work, and when work upgraded me to a 2019 MBP I kept the 2013 for myself.

Since then, the 2019 macbook has been having tons of problems, crashing, thermal issues, software issues, keyboard issues. But the 2013 has been working like a dream. My coworkers have many similar experiences.

Mid-2010s was definitely a peak for macbooks. Hopefully M1s/M2s will be another peak.

unholiness··on “I’m selling data of 400M Twitter users that was scraped via a vulnerability”
Mandated SHAKEN/STIR seems to have been huge. I also used to get about 1/day, now it's 1/week and it doesn't even ring - my Pixel just gives a notification that I missed a call from a known shady number.

I know no news site will ever get clicks from "Competent Government Helps Solve Real Problem", and I know we're not 100% there, but it was genuinely a leap forward and people should talk about it more.

unholiness··on The Simple Geometry Behind Brownie Bake Offs and Equal Areas
Yeah, talk about burying the lede! I had no idea so much progress had been made on this problem.
unholiness··on On CAD
Onshape is a decent middle ground here. The BREP kernel (parasolid) is obviously closed source, but the standard features built on it (extrude, loft, etc.) are open-source and written in their own bog-standard c-family language of FeatureScript[0]. You can write your own custom features pretty easily, and the feature inputs and outputs are actual BREP. It's easy to mix and match manual modeling with standard features and automated modeling with custom features.

[1] https://cad.onshape.com/FsDoc/

unholiness··on On CAD
Yep. Here's a video of a CAD competition, with experienced Onshape user modeling simple rectilinear parts[0]. Video is sped up, but live modeling took under 3 minutes per part[1]

[0] https://www.youtube.com/watch?v=TzyI7xnGoPI [1] https://www.youtube.com/watch?v=Whw9RrZz6XI&t=2073s

unholiness··on Resolving the great undo-redo quandary
That's how every other undo works. The whole GRUQ is caused by undos not themselves forming a part of history...
unholiness··on Resolving the great undo-redo quandary
Interesting. So basically the last chain of undo doesn't enter the history until you do something else.

I often use undo as a faster delete when making edits, so I think I would still be annoyed by this (why am I seeing this crap I undid reappearing?) Now after a long undo chain I'm afraid to type for a new reason: it will pollute my undo stack. But it might be reasonable if my brain was used to it. Can't knock something I've never tried.

unholiness··on Resolving the great undo-redo quandary
Feels like I'm missing something...

If I press ctrl-z ctrl-z, I expect the last 2 things typed to be undone. Based on what he's saying, the first ctrl-z undoes one step, and the second ctrl-z undoes the undo i.e. puts me back where I started, with no way to get back further.

Is there a special case for multiple undo's in a row? If so it seems unclear where to draw the line. If not it sounds nonfunctional.

unholiness··on Show HN: RankedVote – SurveyMonkey but focused on ranked-choice voting
Every voting system has some flaws (see: Arrow's impossibility theorem), and Instant Runoff Voting definitely saw its worst-case scenario rear its ugly head in Alaska: A Condorcet winner existed (Nick Begich), he lost, and voters whose first choice was Plain and second choice was Begich could theoretically, retrospectivally, all changed their first votes from Palin to Begich to have him win rather than Mary Peltola.

It's important that I say retrospectively. This was a house race with national coverage, tons of high-quality polling from pollsters with nothing better to do, and it was still absolutely not clear from that polling that Begich would do better than Palin in round 2, or that Palin was likely to beat Begich in round 1. If I had the opinions of those hypothetically strategic voters, and I were completely strategic, and completely tuned into that polling, I still would have voted honestly. The strategy was only apparent after the fact.

Thinking about "defense in depth" against this scenario, we'd be remiss not to mention Condorcet ranked choice voting methods, which always elect a Condorcet winner, and only resort to (game-able) elimination procedures if there isn't one. Not only would that have resulted in a better outcome in Alaska, it would have made the possibility of strategic voting much, much more remote, both because a strategic scenario would be much less likely, and because that strategic scenario would be much less visible before the election.

With score voting, strategies are always available and almost always obvious: just exaggerate your score differences between the most likely candidates. The fact that strategic dishonesty would have helped in Alaska is a fair criticism of IRV, but the real commonness of strategic scenarios in STAR and score are a more primary feature.

unholiness··on Show HN: RankedVote – SurveyMonkey but focused on ranked-choice voting
Absolutely. Naive exaggeration is a fantastic way to make your vote count more than others in score voting. It also encourages more negative campaign tactics, since convincing a voter to rank your opponent 0 rather than 4 is literally 4 times as powerful as convincing them to rank you a 5 rather than a 4.
unholiness··on The anti-inflation pivot of 2022
Macroeconomics novice here. What's the argument that interest rates need to be close to inflation rates to counteract it?

It seems like there should be an intrinsic time-value of money (e.g. 1 stuff today is worth 1.05 stuff next year) which is tradable (supply of future-stuff will equal demand for future-stuff at some rate like 1.05). And that single rate of stuff-interest would exist if the central bank does nothing. If that's true, then in a long-run equilibrium, shouldn't there be zero inflation if the interest rate in dollars (i.e. the fed rate) matches that intrinsic rate of stuff-interest?

Any explanation or links to the same would be helpful! I'm just past the point of grasping the fact that interest rates going down causes inflation to go up, but finding it hard to find sources on anything that give me intuition on answering e.g. how much things would go up if you held one policy forever, and what factors that answer would depend on.

unholiness··on On Corpspeak
Honestly "ask" as a noun, despite the hate, was one of the most useful corpwords they mentioned.

If I say "what's the ask here?" I mean the concrete part. I mean the "request" just described (by a middle-man) was so fluffy in its language I couldn't figure out what actual action we need to take to make the asker happy. It's a common situation in a corporate world, and imo deserves its own word.

unholiness··on The Google Pixel 6a highlights the problem with the U.S. phone market
Worth noting the vendor lock-in offers are quite significant. My wife and I both had Pixel 3as on Google Fi, both got offers of $300 to trade them in to buy a Pixel 6a. The obviously gimmicky trade in price was only for buying a 6a specifically (buy another phone and the trade in drops to $30)

Gimmicky but worth it, $150 each to get brand new phones that check all the boxes.

unholiness··on Political betting site PredictIt to shut down after CFTC withdraws approval
From the fees. 90% of a $1.1111 profit is $1.

Notably, you don't need to post $N to make this bet. PredictIt calculates your worst case scenario and holds only that money. A set of "No" shares worth N-1.12, if it existed, would be literally free to hold. Fees are included in that calculation, but also fees are based on the difference between that cost to hold and the return. So yes, profits mean different things in different parts of that calculation in this degenerate case.

A full set of Nos worth N-1 would be a guaranteed loss of $0.10 per share from fees. PredictIt would hold that 10¢ and you would never see it again.

unholiness··on Political betting site PredictIt to shut down after CFTC withdraws approval
This is true about arbitraging between markets on PredictIt, but I do think the impact of fees on market prices (and on ability to make money on PredictIt) was usually overstated.

PredictIt has two fees: A 10% fee on winnings, and a 5% fee on withdrawals.

The 10% fee on winnings is only on winnings. So when buying a share for 90¢, your $1 won will earn 99¢ and pay 1¢ in fees.

This means in a market with multiple candidates, betting "No" on all of N candidates will always earn you $(N - 1) without fees. That's is a winning bet iff all the "No" shares add up to $(N - 1.1111). In practice, the bias toward "Yes" shares alone is strong enough that buying all "No" shares, even at current asking prices, usually cost between $1.07 and $1.10 less than $N.

So, if you're only buying "No" shares in markets with multiple outcomes, the prices are set close to a point that entirely negates that 10% fee.

And of course the 5% fee on withdrawals is only on withdrawals. If you send the money straight back into another trade over and over, the fee is diluted enough to be irrelevant to any individual trading decisions.

unholiness··on Political betting site PredictIt to shut down after CFTC withdraws approval
Betting on PredictIt has been a side-hobby of mine for the last 5 years, in which time I turned my fun $10 deposit into (just) over $1000. For me, it's a satisfying way of interacting with the news, motivating clear-headedness and accuracy over simplicity and filter bubbles.

The talk about "rational markets" in this thread is well-meaning, but I think it could better targeted toward BetFair and other uncapped foreign markets. PredictIt has a cap of $850 from any individual in any market, which means almost no market is dominated by "sharps" exploiting differences between prices and reality. Sure, PredictIt's prices are often more accurate than the average person's guess, but they still exhibit a lot of small and predictable biases:

- "Yes" positions are more popular

- Pro-Republican positions are more popular

- Cheap, improbable positions are more popular

- Positions confirming simple ideas are more popular

- Positions traders wish were true are more popular

- Things being discussed on the news are treated as more contentious than they are

Ultimately, almost all of my 100x growth just boiled down to finding these biases and maximizing my expected log(return) with them in mind.

I do believe PredictIt was good for discourse. Though it may have ultimately affected a small number of views, the "put-up-or-shut-up" mentality is much closer to the scientific method than media and its tendency to navel-gaze. The domain of PredictIt markets was small, but I often imagine a world which creates this kind of ecosystem for a broader array of scientific fields.

For me, though, this is the end of the line. The 2020 races (with Georgia runoffs) risk being unresolved by the 2/15 deadline, and it's impossible to predict what PredictIt will do then, much less if markets will properly price that in. It's been a fun run, and I'm glad it lasted as long as it did.

unholiness··on Future trains could provide carbon capture
Direct air capture is the holy grail of climate solutions. Its success would offer us a way to put the cat back in the bag, and see CO2 levels decrease.

The problem is how absurdly expensive these things are. In the next couple decades, even being optimistic, it's hard to forsee a scenario where we've plugged enough holes in our infrastructure that direct air capture will be better or cheaper than doing more capture inside some existing pollution site, or replacing more fossil fuels with clean energy.

But one day we will reach that point. And when we reach that point, anything to reduce the cost of carbon capture a little bit will help a lot.

So yes, regenerative breaking connection is sort of a distraction. But improvements to DAC efficiency are very welcome, and improving airflow by mounting the mechanism on a high-speed train seems like it offers a big improvement.

unholiness··on Magnus Carlsen to give up World Championship title
But that doesn't actually separate the players from their time. Opening theory and endgame knowledge is developed with computers now, and humans use and learn from them. Move accuracy for thousands and thousands of GMs and IMs will easily be higher than Kasparov in peak form back in his heyday. But Kasparov came up with those moves, learning only from lesser minds.
unholiness··on Explaining Code Using ASCII Art
Any visual information that needs to be understood by a human doesn't scale. ASCII UMLs might scale slightly less well than, say, powerpoint UMLs, but having them right there in the code comment in cases that do fit is splendid.
unholiness··on How Townscaper Works: A Story Four Games in the Making
It's rare to see such a well-written article, which manages to distill complex ideas into something approachable without becoming non-technical.

I've come to expect with articles like this that at a certain point, the author stops fully understanding, and switches to metaphors or analogies that ultimately fail to explain the central concept. While plenty of detail was skipped, the switch to the non-technical never came and I happily read it the whole way through.

unholiness··on Welcome to the M1 Windows project
Following this completely imaginary "never end a sentence with a preposition" rule is some bullshit up with which I will not put.
unholiness··on Web Audio Modem (2017)
A project I worked on in college was a "steganographic modem", transferring data in the audio spectrum by lacing a song being played with a signal, such that the signal in each frequency band would always be below the threshold that the human ear could tell the difference.

Even in a controlled setting the bitrate we achieved was very low, but for most songs it was enough to encode the lyrics. Seems quite useful until you remember both ends already need the original audio file in its entirety!

unholiness··on Show HN: A Reddit style site to discuss podcast episodes
The biggest thing missing for me getting started is a reddit-like "feed" showing where discussion is actually happening. Like the reddit logged-out homepage (/best, not /new).

The value of Podbabble is the intersection of podcast episodes that 1) people are talking about and 2) I have listened to. Right now the only way for me to find that intersection is to start from (2), searching for every podcast I can think of, looking at several of the latest episodes and seeing there's no comments. It would be great to start from (1), so I can see right away which episodes people are discussing and see which ones might be relevant to me.

Also, it seems like the only way to add a podcast to "my podcasts" is with an RSS feed link? Not from a link on the podcast's page I searched for? I'm unsure on what the desired setup workflow is.

unholiness··on Show HN: A Reddit style site to discuss podcast episodes
There's definitely reasons for it (to easily link to clips in comments) but there's no reason to block loading the comments on loading the episode or the player
unholiness··on Show HN: A Reddit style site to discuss podcast episodes
I got the same searching "mindscape" but tried again and it worked. I think it's a race condition in the search-as-you-type system (which also appears to have some flakiness, with searches from previous searches showing up after the latest search arrived).
← PreviousPage 4 of 11Next →