Catch up on stories from the past week (and beyond) at the Slashdot story archive


Forgot your password?
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×

Comment The issue isn't (just) speed - it's (also) range. (Score 1) 43

LTE is already pretty darn fast, so losing a little performance isn't going to make that big of a deal. It's not as if you can torrent to your hearts content without killing your cell phone bill.

The issue isn't just speed. It's also range.

At any given speed, the Qualcom can support it at substantially lower signal levels. 6ish dB in a lot of cases, a bit less in some, enormously more in others.

Look at the graphs in TFA. In addition to some specific pathologies that penalize the Intel chip farther, the bulk of the graph has the drop off looking similar but with the Qualcom shfited 5 or 6 dB to the right. (Those squares are 5 dB wide.)

6 dB is four times the effective signal strength, which corresponds to twice the range. That maps into four times the area served at that speed from a single cell tower (important in sparsely-served areas), deeper penetration into buildings and the like (in more heavily-covered areas). It can also map into more data pushed before a given area and channel allocation's bandwidth is saturated. 3 dB corresponds to twice the effective signal strength, 1.4ish times the radius, twice the area served.

If the modems were equivalent and the problem just the layout of the board and antenna, you'd expect the two curves to be the same shape but just offset. The shape is substantially different, so (board issues or not) something else is going on.

Comment REALLY? (Score 5, Insightful) 209

The crime is making orders with the intent to cancel before being fulfilled. ... The intent to cancel, in order to create a false market perception, is the crime. ... a pattern of cancelled-while-unfulfilled orders, combined with other orders that profit from the market perception that the unfulfilled orders create, is a very clear establishment of such intent.

Is it also an establishment of intent if you (as a large financial firm) deploy, in actual trading on real markets with real money, an algorithm that exhibits such behavior? If, in addition, you KEEP it deployed even after its behavior is noticed and complained about in public media of the sort likely to be read by trading professionals?

And it is something that the traders at Goldman Sachs can make a fortune without doing.

But it's something that they can make a BIGGER fortune by DOING. And something that can count toward the rise of individuals and groups through the corporate ladder and pay scale.

While don't recall if G.S. was specifically one of the organizations complained about (and am not going to spend the time right now digging through archives to check), I DO recall com"plaints about high-speed traders taking advantage of the cancellation features of the online market engines in just this way.

One of the advantages of shaving milliseconds off the communication delays and algorithms that was specifically mentioned (once the pattern was observed) was the ability to send an order and a cancellation in rapid enough succession that it could not be pounced on (and thus didn't really risk money), sending price signals that tricked competing, slightly less high-speed or well-tuned, algorithms into making other bad trades from which their operators lost and the perpetrators gained.

Comment Re: Has Wikileaks jumped the shark? (Score 1) 269

The only poll that showed that was the LA Times poll, ...

Which "that" are you referring to?

- If it's who's ahead, you have some point - though the USC Dornsife/Los Angeles Times "Daybreak" poll is NOT the only one showing Trump ahead by at least a small margin. (In fact it shows them neck-and-neck, crossing back-and-forth, and has Hillary up by 0.3% just now.)

- If it's how the musdslinging is changing the voters' preferences, it's not fully over that poll's one-to-two-week report delay + smoothing yet.

In particular (for both cases) I note that this morning's Rasmussen daily tracking poll has Trump slightly ahead AND is starting to have any dings to Trump from the spate of allegations within horizon. Yet they see no such dings (and comment about their absence in the accompanying analysis).

[USC/L.A. Times] is a bit notorious because it polls the same people at each iteration.

It's also notorious for being far more accurate when it comes to predicting both the winner and the margin. (The latter can turn into the former if the race is very close, as this one is.) That repeated sampling of varying subsets of the same group is a part of the methodology that they credit for avoiding certain distortions that affect other polls.

The rationalle is explained on their web site. Give it a look.

Comment Re:systemd (Score 3, Interesting) 78

Honest question: in what use case does systemd bother you?

I'm in a startup, still on angel funding and strapped for resources, building a multi-layered platform. One of the four-or-more layers is implemented on a machine about the power of a smartphone/credit-card-computer in the raspberry/beaglebone/etc. class. That layer needs an O.S., and it's internet-facing, so it needs to be secure - and auditable.

Posix-compatible OSes, such as Linux, should be ideal. But there's that little matter of being reasonably sure that they're not full of security holes or reliability issues, and doing so on a shoestring, using a handfull of people who have a LOT of OTHER stuff to do in order to get through the market window before the wolf gets to our door.

Even if systemd were solid as a rock and the best thing in init systems since pre-slicing was applied to bread, it's an extra complication - with its fingers in a lot of pies. That makes security auditing much harder and more time consuming. And THAT makes it "more expensive than money" for us - to the point that the current move of Linux versions to systemd may drive us to abandon Linux entirely for something else. (OpenBSD would be one contender. A plethora of other, stripped-down-to-minimal-functionality, OSes also come to mind.) (The main reason we haven't done so already is that we can't afford that effort, either, until our concept's proven and we must bite the security bullet in order to ship.)

One of the great things about pre-systemd Unix and unix-like systems was the design philosophy, which explicitly drove strong modularity, with simple modules that did single jobs and were easy to check - or encapsulate. (This was one of its big advantages over things like Windows, where all the apps were in bed with each other and any security hole in one became a security hole in many.) Systemd violates that philosophy.

Comment Re: Has Wikileaks jumped the shark? (Score 2) 269

Perhaps if Trump hadn't publicly bragged about doing ...

The main thing I got out of it was: He said
  - he'd made a pass at a married woman,
  - she'd turned him down,
  - and he took "no" for an answer.

What a pity the Clintons don't seem to do that. ... what he is now accused of,

It's the last four weeks of the campaign. There is no longer time for the truth of accusations to come out before election day. Now is the time when, historically, dishonest politicians and/or their supporters have a track record of making up believable lies and broadcasting them.

So now is the time that I don't believe smear stories - that don't have solid, PRE-EXISTING, evidence to support them - about candidates in a tough race.

Maybe he did something bad, maybe not. But the spate of accusations certainly seem to be nicely tuned to a narrative based on the recording. The many accusations allegedly took place years ago. If true, they could have been published months ago. Why are they only coming out now, just after the classic cut-in time for fake-story smears? The timing certainly makes them LOOK like they are lies, constructed and carefully tuned for use in the last weeks of the campaign.

his campaign wouldn't be crashing and burning.

The last polls I saw (from just after the recent debate, and substantially after the release of the recording, but before the series of accusations) had him pulling ahead by a substantial margin - like by almost the total of the "undecideds".

My impression is that the only crashing and burning going on is on the part of the Republican politicians who took this opportunity to break their agreements and oppose Trump - only to find he was the one with the bigger army of supporters.

Comment Re:Has Wikileaks jumped the shark? (Score 4, Interesting) 269

The one-sided nature of the leaks suggests that either Wikieaks has an agenda, or it is the willing accomplice of someone who has an agenda.

The murder of Seth Rich, along with the rumor that he was Wikileak's source for the DNC leaks, puts Assange in a tough position. Whether Seth WAS his source or not, public perception that Wikileaks sources may be killed and their leaked information thus suppressed could make his sources dry up.

To counter this he has to create the perception that, if you try to plug the leak by plugging the source, it will result in the leaks being more extreme and damaging, rather than less.

One way to do that is to publish more than would normally be published, and redact it less.

Perhaps that is what is going on: As with rule 804, where hearsay from a deceased person becomes admissible, insuring that leaks from (or perceived to be from) a murder victim are extreme might help make murder of leakers less likely.

Comment Re:systemd (Score 3, Interesting) 78

ubuntu comes with a no systemd option.

Good to know.
  - How does one use it?
  - How do you KNOW no systemd hair is still tangled in your system?
  - Do all the components work correctly when you opt out of systemd? Nothing breaks or performs substantially more poorly?
  - Are they all supported as well in both environments? No obscure "gotcha"s?

Comment Re:Of course (Score 2) 78

There is no free market when the government 'gets' to license spectrum.

And doubly so then they consider it "competition" when there are TWO suppliers:
  - With ONE supplier (monopoly) the incentive is to charge as much as possible - until you're driving away more in customer revenue from people who do without or with less than you gain from higher prices.
  - With TWO suppliers (duopoly) the market forces lead rational players to adjust prices to divide the market roughly evenly and continue to charge monopolistic prices. Price signals are enough, no forbidden collusion required.
  - With THREE suppliers you're starting to get instability. If the big two make it too hard on the little guy, he may cut prices to try to grab market share, leading to a price war that drives prices down toward cost-of-goods-plus-modest-margin. Or they may manage a balancing act and prices stay high.
  - With FOUR OR MORE suppliers the high-priced equilibrium is almost certain to collapse, and it gets progressively more unstable with more players.

So government communications regulatiory policy - at least in the US - is explicitly encouraging duopoly markets and monopolistic pricing structures.

Slashdot Top Deals

Brain off-line, please wait.