Where does it say that they haven't commented on earlier versions?
Posts by Michael Wojcik
12271 publicly visible posts • joined 21 Dec 2007
Page:
- ← Prev
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
- 42
- 43
- 44
- 45
- 46
- 47
- 48
- 49
- 50
- 51
- 52
- 53
- 54
- 55
- 56
- 57
- 58
- 59
- 60
- 61
- 62
- 63
- 64
- 65
- 66
- 67
- 68
- 69
- 70
- 71
- 72
- 73
- 74
- 75
- 76
- 77
- 78
- 79
- 80
- 81
- 82
- 83
- 84
- 85
- 86
- 87
- 88
- 89
- 90
- 91
- 92
- 93
- 94
- 95
- 96
- 97
- 98
- 99
- 100
- 101
- 102
- 103
- 104
- 105
- 106
- 107
- 108
- 109
- 110
- 111
- 112
- 113
- 114
- 115
- 116
- 117
- 118
- 119
- 120
- 121
- 122
- 123
- 124
- 125
- 126
- 127
- 128
- 129
- 130
- 131
- 132
- 133
- 134
- 135
- 136
- 137
- 138
- 139
- 140
- 141
- 142
- 143
- 144
- 145
- 146
- 147
- 148
- 149
- 150
- 151
- 152
- 153
- 154
- 155
- 156
- 157
- 158
- 159
- 160
- 161
- 162
- 163
- 164
- 165
- 166
- 167
- 168
- 169
- 170
- 171
- 172
- 173
- 174
- 175
- 176
- 177
- 178
- 179
- 180
- 181
- 182
- 183
- 184
- 185
- 186
- 187
- 188
- 189
- 190
- 191
- 192
- 193
- 194
- 195
- 196
- 197
- 198
- 199
- 200
- 201
- 202
- 203
- 204
- 205
- 206
- 207
- 208
- 209
- 210
- 211
- 212
- 213
- 214
- 215
- 216
- 217
- 218
- 219
- 220
- 221
- 222
- 223
- 224
- 225
- 226
- 227
- 228
- 229
- 230
- 231
- 232
- 233
- 234
- 235
- 236
- 237
- 238
- 239
- 240
- 241
- 242
- 243
- 244
- 245
- 246
- Next →
Microsoft ain't happy with Russia-led UN cybercrime treaty
FreeBSD can now boot in 25 milliseconds
Re: Hold Your Horses
Lightweight VMs have a number of advantages over containers, such as kernel separation.
Re: The bubble has burst
For small N the Standard C qsort() is not going to perform well, because comparisons are expensive (call through a function pointer for each one). Generally, for small N, you don't care; but if you know N is going to be small, and performance is important anyway, then inlining some other sort – even one that's not O(lg N) in time complexity – is better.
Re: The bubble has burst
Quicksort can be (and often is) implemented iteratively rather than recursively, if there's a ceiling on N (which there is in all real-world implementations). It will still need O(lg N) auxiliary space but the coefficients are small and if lg N is always less than, say, 32, it's unlikely to be a problem.
Even with recursive Quicksort, if you're careful, you can keep your levels of recursion to O(lg N).
And even in a heavily-constrained environment, gnome sort would beat bubble sort, because it would use fewer registers.
Re: Pretty impressive
Shellsort is an interesting case because its theoretical performance is difficult to analyze.
Quicksort, of course, has bad worst-case performance in the naive implementation. There are various tricks you can use to avoid that, and there's Introsort, which detects adverse performance and switches to a different sort.
Radix sorts are even faster than comparison sorts like Quicksort under the right conditions; the problem is that those conditions don't fit most cases.
These days, state-of-the-art sorts seem to be things like the widely-used Timsort (a mergesort / insertion sort hybrid that employs a number of techniques for improving real-world performance and has the advantage of being stable, i.e. not reordering items with the same key) and Learned Sort (which builds a model of the input's distribution and uses that to approximate a radix sort).
For sorting large amounts of data ("external sorting"), there are various partition / parallel sort / merge approaches, some of which are fairly ornate (such as the clasic polyphase mergesort). There's a tradeoff between partitioning and merging: you can partition so none of the subfiles have overlapping keys, in which case merging is just concatenation; or you can partition randomly, and then during merge you compare the keys of the next record from each subfile. You could even do external Shell sorting where you use the Shell stride to do the partitioning and merging.
But even for small N, bubble sort is pretty terrible. It's not that easy to code or understand; I'd say naive mergesort is easier to write and read if the language has decent support for recursion and allocation of auxiliary space. And gnome sort is simpler than bubble sort, with the same performance characteristics. Selection sort is even easier to understand; you can describe that to an novice in two sentences. So even if you expect N to remain small, it's hard to see how bubble sort is ever the right choice – except when it's the only sort you know and you can't be bothered to do any research.
America's financial cops say Impact Theory's NFTs were unregistered securities
Re: We've done this before
While I am certainly not a fan of NFTs, I don't see a clear bright-line test that makes them in the general case any different from various traditional art products, except in not having a tangible realization (like a physical painting). Since the latter also applies to, for example, an ebook that's never published in print form, it doesn't seem like a useful test.
I do agree with jmch's evaluation here that these particular NFTs were being marketed as securities, however. Peirce and Uyeda have a point: just declaring NFTs to be securities in general seems problematic.
(Those two commissioners have interesting history. They're the only two current SEC commissioners affiliated with the Republican party, but Uyeda was nominated by Biden, and Peirce was originally nominated by Obama. That nomination was blocked by Democrats in the Senate, and Trump re-nominated her.)
University cuts itself off from internet after mystery security snafu
Re: Wrong School!
Which "other Michigan University" were you thinking of? Michigan State? Michigan Tech? Eastern Michigan? Central Michigan? Oakland? Western? Wayne State?
Those are just the public research universities.
(The wolverine is the mascot of U Michigan, because there probably aren't any wolverines in Michigan any more. The mascot of Michigan State is the Spartan, because there have never been Spartans in Michigan. This par for US universities in general, of course, though U Nebraska gets some credit for using "cornhusker", which at least used to be a real profession, and before that "bugeater", which is both possible and for US mainstream culture fairly hardcore. And U Kansas for "jayhawk", which is not a real bird but originated as a term to describe anti-slavery activists; that's kind of nice.)
Apple security boss faces iPads-for-gun-permits bribery charge... again
I've browsed through the linked articles and some of the court findings. The evidence doesn't appear to be overwhelming either way. The iPads could have been specific quid pro quo for the CCW licenses, or just a general goodwill donation arranged by a high-ranking Apple employee for an outside organization, which would not constitute bribery. Both the dismissal and the reinstatement by the District Court appear to hinge on presumptions about Moyer's likely state of mind and whether there was mens rea.
Basically, the grand jury returned an indictment because they found there was enough evidence to suggest there should be a trial. Indicting grand juries return an indictment in the vast majority of cases.
The trial court dismissed the case on the grounds that insufficient evidence for mens rea existed, specifically because the judge thought the preponderance of evidence suggested that Moyer thought the CCW permits were already approved (so no quid pro quo was needed). The district court disagreed on the grounds that Moyer's actions could be seen as sufficient evidence to justify a trial – that is, that the evidence needed to be evaluated by a jury (or a judge if Moyer did not request a jury trial).
So the question is whether there's enough evidence supporting a finding of an intent to bribe that there's a reasonable expectation that a jury could find Moyer guilty beyond a reasonable doubt. That's still a very obscure situation. This is basically judges quibbling over whether the prosecution has any (reasonable) hope of proving their case.
Perhaps AI is going to take away coding jobs – of those who trust this tech too much
Re: Useful but heavy pinch of salt required
Certainly it's a good way to avoid learning anything, improving your own skills, or serendipitously encountering something else of interest.
Delegating your work to a machine has essentially the same advantages and disadvantages as delegating it to a stranger of unknown capabilities and motives.
Brain-computer interface and AI helps stroke victim speak through avatar
Re: "The [NYT] updated its terms of service"
"Blah blah I can't be bothered to even think about this topic, but I'll post anyway."
It would be trivial – trivial – to ensure one of OpenAI's GPT models is including ToS statements published on websites in its input, and then to have GPTbot query that model before crawling the site. That's a well-defined problem. You don't even need an LLM to do it; a decent HMM or MEMM would almost certainly show very high F1 in a task like that.
Try criticizing LLMs and AI hype for real reasons. You're just muddying the waters.
Silicon Valley billionaires secretly buy up land for new California city
Re: They keep trying to build in deserts
Though in California, "prime agricultural land" often means "land where large agricultural conglomerates can bribe Reclamation into ignoring their rampant violations of the law while draining the Colorado River and pumping ancient aquifers dry".
There's no question California has a housing crisis, but it also has a water disaster which is only going to get much, much worse (unless they turn it into an energy disaster by trying to ramp up large-scale desalinization). Congress could in theory alleviate this somewhat by purging Reclamation and re-staffing it with people actually willing to enforce the law, which would greatly cut down on how much water is taken by the ag conglomerates, but in reality California has far too much political power for that to happen, and, hey, we want cheap almonds now rather than water tomorrow, right?
(Though eventually all that irrigation will ruin all the arable soil anyway. Keep flushing those salts out of the high country into your fields, kids. We've known since Babylon that eventually the dropping osmotic pressure and defloccination of soil will make it impossible for you to grow useful crops. This has happened to every irrigating culture in history, except the ones like Egypt (prior to building dams) that used natural flooding, which deposits fresh, low-salt soil. In the not so distant future California's agricultural sector is screwed, and it's their own damn fault.)
Foxconn founder Terry Gou to run for Taiwan's presidency
Foxconn never received those tax breaks
And OP never claimed they had, so what's your point? The text you quoted said that there were tax breaks, not that Foxconn profited from them.
Foxconn promised large developments. They did not deliver.
And, of course, Foxconn did not engage in these projects out of the goodness of Gou's heart. Some of it may have been hedging – entering into the venture to see whether it would turn out to be profitable – and some of it may have been quid pro quo, which might have been anything and everything from kickbacks to political capital (though Trump is wildly unlikely to repay such debts, and Walker's political worth is now minimal).
Re: Separation of business and state
"Military-industrial complex" is generally said to have been coined by Eisenhower in his 1961 farewell address. So 62 years; your recollection was right on the money.
Remember when Republican presidents were like Eisenhower? It'll be a long time before we see that again from the GOP. They've done an excellent job of turning the party into an authoritarian dumpster fire. (Lindsey Graham crying like a televangelist asking people to donate to Trump's legal defense fund is a splendid example: complete ideological bankruptcy, utter lack of shame. I don't know if the remaining handful of Republican politicians who still have any backbone are enough to rescue the party in their lifetime.)
US Republican party's spam filter lawsuit against Google dimissed
Hope for nerds! ChatGPT's still a below-average math student
Re: " throw in a few trick questions"
we analyze correlations between a lot of things, not just tokens
Care to cite any evidence from, say, methodologically-sound neuroscience research, or even psychology, to support this claim, for some useful definition of "token"?
As I've noted before, I'm not particularly impressed by LLMs. I'm even less impressed by handwaving appeals to res cogitans to try to claim human cognition is magically different.
Re: Just wondering
Due to the large size and dimensionality of the parameter space in contemporary LLMs, and the large context windows, it often is possible to elicit large passages of text verbatim from a source. It's not hard to find gradients in parameter space in these models to recover that sort of thing, particularly if you keep the temperature low enough to avoid introducing noise or annealing the output into some "better" (in objective-function terms) result.
Re: Just wondering
It's also worth noting the economics (sorry, Rikki) of classroom instruction and evaluation. Low-level courses tend to have large class sizes, often in the hundreds of students for some disciplines. Evaluating that many students quickly becomes very expensive; it's not feasible unless you use methods that permit quick, preferably automated evaluation. By definition, such responses will have quite low information entropy, and so LLMs will tend to perform well on them.
Also, at least in US higher education, introductory courses are asked to serve multiple competing aims. They are supposed to introduce students to a subject area so they'll be prepared for more advanced study. They're generally asked to perform as general-education courses, i.e. to help students "catch up" to the base level expected of undergraduates. Administrators want them to serve as gatekeepers, discouraging students who think they might be interested in a discipline but turn out not to be from continuing. Students use them for easy (or easier) credit hours to fulfill requirements. Departments staff them with the cheapest faculty – fixed-term, contingent, and graduate-student workers – and run many sections, to make their budgets. Students take different approaches to the courses depending on which of these aims align with their own motives, and many are content to simply squeak through.
Meta lets Code Llama run riot under almost-open terms
Carr's Observation, applied to programming
Yes, by all means, let's employ more and more tools which spoon-feed developers answers (correct or not) rather than forcing them to consult authoritative sources and perhaps learn something.
I have a CS degree (plus some graduate study in the field, as part of my other degrees), and more than three years in the industry; but a great deal of what I know arrived serendipitously. Some of that was from obvious sources like reading CS journals and papers, but quite a lot of it was stumbling across APIs and other technical information while browsing documentation, or looking at other people's code, or reading newsgroups like comp.lang.c and sci.crypt, or trial and error.
LLMs are making us – well, many of us – stupid. Easier programming is often not better for software development.
Tools such as compilers mostly improve software quality because they abstract away details that are usually rote and irrelevant to quality. That's not broadly true of things like algorithm and API choices, where understanding the available alternatives can make a big difference.
Uncle Sam accuses SpaceX of not considering asylees and refugees for employment
It makes them a particular type of resident. Not all residents are asylees.
Asylee means "someone who has sought asylum and been granted it by a sovereign nation". While it is not, to my mind, a euphonious term, I don't know of another which concisely represents that precise concept. "Asylum grantee" (in parallel with "asylum seeker") is the closest, and it's hardly a deft flick of the pen either.
Zoom CEO reportedly tells staff: Workers can't build trust or collaborate... on Zoom
Two teens were among those behind the Lapsus$ cyber-crime spree, jury finds
Re: computer intrusion, blackmail, and fraud
Lapsus$ didn't do anything sophisticated or novel. Their attacks were notable primarily because they used a relatively expensive approach1 and were a small, hands-on group rather than using a franchise / multi-level structure like most IT extortion groups. That made targeted attacks more economically feasible.
They don't have anything to teach security experts, and what they have to teach ordinary users has already been known for decades. Ordinary users just aren't capable of maintaining that level of vigilance. It's not something people are good at.
1Contacting victims in person is much more time- and labor-intensive than blasting out a million 419 emails, or scanning a million systems for a known vulnerability.
Re: Hold on...
Autism is a whole bunch of things, which is why people refer to it as a "spectrum". I'm suspicious of the autism defense too, but without a lot more information I'm not going to comment on it.
Social-engineering attacks are often carried out using scripts, which can be obtained from various malware vendors. Social engineering is sometimes customized for a specific target, but often attackers just use generic approaches against a number of targets, knowing there's a good probability that at least one attack will succeed.
Jessica Clark's demonstration for Kevin Roose at DEFCON – you can find the video on YouTube – is a well-known example of a generic soc-eng attack. Clark's good, but even someone with poor social skills could learn to do it from resources like that video and some practice.
Want tech cred? Learn how to email like a pro
Hey, let's argue about "original"!
USENET news, the original pre-web social network
Um ... USENET was launched in 1980. BBSes arrived on the scene at least seven years prior to that, with Community Memory. USENET has many advantages over BBSes, but BBSes were definitely multi-user public-messaging services that were used for a wide range of communication. I don't think you can reasonably argue they were qualitatively different, as social networks, from today's Facebook and the like. Just better.
No doubt someone can make an argument for an online (in some sense) "social network" earlier than that.
Re: Next story: Proper use of CR, CR/LF & LF (and other whitespace) ...
That's Jon Postel's Interoperability Principle, not "Law". And "interoperability" is in the name for good reason: it's excellent advice for interoperability, and terrible advice for security.
Like most things, it's a trade-off.
Re: Its all about *efficient* communication...
While I am not a fan of synchronous media, and very much not of interruptions, I do use phone calls (well, Teams calls, because that's the particular hell my employer has chosen this week) at work. I almost never initiate them, but often enough I'm on the receiving end. Generally it's to help investigate a technical problem that resists analysis with a quick back-and-forth by email or chat, where we want to screen-share logs or a debugger session or some such thing; or to discuss some other sort of complex technical issue.
I take Snake's point that there are use cases where the exchange efficiency of conversation offers significant benefit. And since I work thousands of miles from most of my co-workers, and hundreds from even the closest, some sort of telephone-equivalent technology is the only way to do that.
Most of our discussions do happen via asynchronous text of one sort or another (email, comments in the review system, chat messages that are treated like email, and so forth), because that alleviates timezone differences, avoids interruption, and provides a durable record – and that last point is hugely important. I keep a lot of old email, going back decades, and probably at least once a week I'm running searches against it and finding some obscure piece of information that someone sent years ago.
(Voice transcription technology Ain't There Yet, particularly for technical discussions where there are a lot of terms of art, jargon, abbreviations, internal cant, and so forth. And current employer forbids them anyway, possibly for legal reasons – they're hell for discovery, because they add a lot of volume and are very noisy and thus difficult to search.)
Those were the days.
I still have my Usenet archives somewhere. I should dig them out; I'm sure they'd be both entertaining and informative. And, of course, Back in the Day you'd encounter a lot of people whom you'd recognize as online personalities or of real-world interest, for one reason or another. Computing pioneers in alt.folkore.computers; well-known technical experts in comp.*; authors you'd read in rec.arts.books. Many, I think most, of us went by our real names back when I was a frequently Usenet contributor, or at least our real-world identities were pretty well-known.
I don't want to go completely GOML here, but Usenet, for me, was a great improvement over its successors. Maybe something similar can be found on some Mastodon servers these days. I haven't mustered the energy to go looking.
Re: Wrapping at column 78
It's trivial to do in vim, and not that hard to do in vi with a simple word-wrapping program. Writing the latter is an exercise for the reader. When I wrote mine, circa 1990, I spent maybe an hour on it, and that was mostly because I did it in C rather than use a scripting language.
(And, yes, I'm including handling the >s. Remove them, wrap at the appropriate length, add them back. Super easy, barely an inconvenience. And having to do it encouraged trimming.)
Tornado Cash 'laundered over $1B' in criminal crypto-coins
LibreOffice 7.6 arrives: Open source stalwart is showing its maturity
Re: "it is increasingly difficult to develop entirely new features"
Shrug. I haven't needed a spreadsheet to do anything Lotus 1-2-3 didn't do. No doubt some people need the various bells and whistles in Excel, but I'm willing to bet that in most cases most users would do just fine without them, and the UI/UX wouldn't be such a horrible fucking nightmare.
Re: I miss ClarisWorks
And how many people use a chisel as a screwdriver?
There is a vast amount of Excel abuse in the world, true. I'm puzzled by the suggestion that this is a Good Thing. Certainly the ill effects are well documented.
Re: long-form writers...
Veritably, the Mount Sunflower of Microsoft's efforts in the land of document creation.
Word's dominance is proof enough of the mediocracy that governs software.
Re: long-form writers...
And LaTeX was still prominent for submissions to academic journals in technical fields back when I paid attention to that sort of thing, which was not so many years ago.
In the humanities Word seemed to dominate, unfortunately, which is one reason why I rarely submitted to humanities journals and collections. Just not worth the grief, whether it's done with Microsoft Word or LibreOffice.
Netflix flinging out DVDs like frisbees as night comes for legacy business
Re: bizarre state of affairs
Hell, I don't care about Blu-ray;1 I don't even think I have a Blu-ray player.2 Nor do I care about commentary tracks and the like – I can't remember the last time I watched one.
But DVDs are still superior to streaming for a whole host of reasons. I can watch them when Internet service is down or congested. (On a laptop, I might be able to watch for a while even if the power's out, though I doubt the battery would last long.) I can find what I want to watch in seconds, rather than having to search through streaming-service catalogs. I can find something serendipitously: Oh, here's something I haven't seen in ages! Oh, I'd forgotten I have that! I can loan one to a friend. The UI and remote for the DVD players I have, though certainly crap, is an order of magnitude less crap than Every Fucking UI for the streaming-service apps on our set-top box, and the box's remote is an exercise in irritation and unusability.3 I will always have my DVDs; I'm not beholden to the whims of streaming services, or of the loathsome content companies like Discovery or whatever they're calling themselves now.4
Streaming is certainly point-and-drool convenient, as long as you don't care what you watch.
For that matter, I'd much rather have to get up off the couch, find a DVD, and insert it into the player if I want to watch something. At least that's a bit of physical activity.
1The official spelling, as far as I can tell.
2I haven't bothered to confirm that my various DVD-playing devices aren't Blu-ray, because I don't care.
3Which isn't surprising, because it's an Apple device. My wife purchased it. To be fair, that's the only way we would have gotten one; I'd refuse to spend the money and build one myself, except I'd never get around to doing that, because I just don't care about TV and movies enough.
4I refer, of course, to the HBO Pogrom that Discovery embarked upon after the merger. If there were any moral order to the universe, Discovery execs would be frying in Hell.
ICANN warns UN may sideline tech community from future internet governance
Re: Not stupid IMHO
There are many things wrong with ICANN. There have been many things wrong with it in the past.
Removing it from Internet governance and substituting some vague UN blather will not improve the situation. A deeply flawed but still largely successful approach is better than one patently constructed from ignorance.
I'm not one of those people who believe the UN is useless – at the very least, it helps to have a ball pit where the toddlers who govern the world can go to pull one another's hair – but this is not something they can productively take over.
Page:
- ← Prev
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
- 42
- 43
- 44
- 45
- 46
- 47
- 48
- 49
- 50
- 51
- 52
- 53
- 54
- 55
- 56
- 57
- 58
- 59
- 60
- 61
- 62
- 63
- 64
- 65
- 66
- 67
- 68
- 69
- 70
- 71
- 72
- 73
- 74
- 75
- 76
- 77
- 78
- 79
- 80
- 81
- 82
- 83
- 84
- 85
- 86
- 87
- 88
- 89
- 90
- 91
- 92
- 93
- 94
- 95
- 96
- 97
- 98
- 99
- 100
- 101
- 102
- 103
- 104
- 105
- 106
- 107
- 108
- 109
- 110
- 111
- 112
- 113
- 114
- 115
- 116
- 117
- 118
- 119
- 120
- 121
- 122
- 123
- 124
- 125
- 126
- 127
- 128
- 129
- 130
- 131
- 132
- 133
- 134
- 135
- 136
- 137
- 138
- 139
- 140
- 141
- 142
- 143
- 144
- 145
- 146
- 147
- 148
- 149
- 150
- 151
- 152
- 153
- 154
- 155
- 156
- 157
- 158
- 159
- 160
- 161
- 162
- 163
- 164
- 165
- 166
- 167
- 168
- 169
- 170
- 171
- 172
- 173
- 174
- 175
- 176
- 177
- 178
- 179
- 180
- 181
- 182
- 183
- 184
- 185
- 186
- 187
- 188
- 189
- 190
- 191
- 192
- 193
- 194
- 195
- 196
- 197
- 198
- 199
- 200
- 201
- 202
- 203
- 204
- 205
- 206
- 207
- 208
- 209
- 210
- 211
- 212
- 213
- 214
- 215
- 216
- 217
- 218
- 219
- 220
- 221
- 222
- 223
- 224
- 225
- 226
- 227
- 228
- 229
- 230
- 231
- 232
- 233
- 234
- 235
- 236
- 237
- 238
- 239
- 240
- 241
- 242
- 243
- 244
- 245
- 246
- Next →