rchowe

The reasoning behind this made a lot more sense to me once I started to do a back-of-the-envelope calculation.

There can't have been that many end users who had Windows 10 Pro and went into group policy to turn off the store. So you're looking at small businesses who were using PCs with Win10 Pro on them (likely that came with the PC) that were turning off access to the store but can't any more. The IT admins for these companies are the people Microsoft wants to upsell.

Lets say that there are 500 businesses who care about this feature each with an average of about 20 PCs (probably a high estimate for PCs, low estimate for number of businesses). That's 10000 PCs that Microsoft could potentially convince to upgrade, at (a quick guess based on Google) $120/PC, to Win10 Enterprise, or a potential $1.2M more in revenue that doesn't cannibalize one of their other businesses (assuming more changes to differentiate Pro vs Enterprise). Probably the people who will upgrade are people with factory computers running Win10 Pro.

And for people that don't upgrade, they get to promote their app store. Win-win.

It probably just means more i3/i5-based tablets, which will probably end up being a good thing anyway, since the push to optimize for weak processors that there was on the iOS/Android side wasn't really there on the Windows side, since a lot of testing happened on developer workstations.

For most of the (new) equipment procurement the MBTA does now, they actually own the specifications for the equipment, which means that if they want to make more, they put out an RFP for companies to manufacture new equipment based on their spec. The downside to this is that the T likes to over-customize their orders, which drives costs up compared to buying more standard "off-the-shelf" equipment, and if no responses to the RFP come back, they have to relax some of the requirements and put it out again.

There's actually a very good reason. Windows XP has been officially unsupported by Microsoft for about two years now unless you are willing to pay them a quarter million dollars a year. Often the computers spec'd for those systems have pitiful amounts of CPU power and memory (i.e. much less than necessary to run Windows 10), and the cost of a re-spec for PCs in certain critical systems is basically the cost of developing a new platform (not to mention that some of them can't fit a microATX-sized PC or have low power requirements) makes it potentially cost-effective to use SBCs, especially a popular one which someone like Newark/Element14 can customize for you and potentially support.

I wonder if ReactOS could target the same APIs as the upcoming Nano Server [1] in Windows Server 2016. They'd have to make it entirely 64-bit and implement some closed Microsoft protocols like WMI and DISM, but it could be a pretty cool drop-in replacement for any server apps that people decide to target to Nano Server.

[1] https://blogs.technet.microsoft.com/windowsserver/2015/04/08...

I agree that it is way too difficult to manage a bunch of monthly subscriptions, and also that having to put my credit card information into a website (and have it bill me every month) is basically as bad if not worse as putting up with the advertisements on a website. However, I think that micropayments/a flattr-style fund for reading articles is doomed to fail because of the psychological shift between "I'm reading this article for free" and "I'm reading this article for $0.01, I only have 1000 of these left this month, maybe this isn't the one I really want to spend this on." My opinion is that a Spotify/Netflix-style subscription model is the best: collect subscription money, then pay per-user that accesses an article. However, the issue with this model is that the "network" of things that you are allowed to read will be really small at first.

That said, here is my proposed business model. Create an app that allows users to load money into the app monthly and issues each user a unique credit card number[1] for subscriptions to other websites, giving a central location to manage subscriptions. Once you build up a number of users on a single site, go to that site and say "Hey, you can save the 1% of the subscription cost from our users that's eaten up by credit card fees[2] by going through us to manage subscriptions. Oh, and it will also increase your conversion rate among our users since it will only take two clicks to subscribe instead of having to enter a bunch of credit card information. Once you have a critical mass, you go around to companies and now create the Netflix-style subscription service. If you're clever, you probably can create article-level bundles, e.g. "every article in our network that was linked to off of Hacker News," but NOW you have the pay-to-read subscription with a critical mass of information that people want to read, instead of "we're launching with these twenty sites from these four companies and we hope to add more."

[1] Yes, there are a bunch of problems with issuing credit card numbers. Yes, I know I complained about having to enter credit card numbers in the first paragraph. This is a big picture thing.

[2] This is kind of wrong, since a credit card fee will be charged to load money into the app, as well as for disbursements. At this point, the startup would probably have to eat that fee or heavily push ACH/lower fee methods of payment (even though I hate companies that push ACH for this reason).

The latter seems more accurate. Although 23andme can't provide you with a health analysis, they still provide you with their entire panel in raw data, which you can upload to other services such as Enlis and get effectively the same health report.

If Enlis grows, they may get a similar letter to 23andme, because they would be offering a medical analysis not approved by the FDA.

I think that the most important part of the article is the following quote in the introduction:

> probabilities may not be well defined if the mechanism or method that produces the random variable is not clearly defined.

So comparing the answer of 1/2, where the chords are generated by moving along the radius in increments of dr, with the answer of 1/3, where the chords are generated based on the angle from the starting point.

The problem doesn't specify how to pick random points, though I suspect that if you were to choose two random points on the radius of a circle by picking a reference "going around the edge" of the circle clockwise by some amount in [0, 2pi), we get the angular situation that produces the 1/3 answer.

I would be suspicious of this too, since I believe a lot of New England gets its power from Hydro-Québec (from their HVDC lines) and nuclear sources. Even normalized per capita, Vermont should have one of the lowest carbon footprints, which makes me suspicious of some of their sampling methods, though I haven't read the paper.

It would be really cool if you could add SSL encryption by default once some service like Lets Encrypt [1] which issues SSL certificates if you can prove your domain name launches. Not sure how this jives with their terms of use though.

[1] https://letsencrypt.org

I would also assume that mail fraud is easier to investigate than telephone fraud.

This was exactly what Microsoft tried to do with the Surface Pro and Surface RT. The Surface RT hardware was very well made, however it only ran Metro-style Windows apps, so the only piece of big-name software for it ended up being Office.

Apple does have the scale on its app store to force developers to change over to ARM (say, if they announce ARM computers a year before they are for sale), but at the PPC -> x86 switch, Intel processors were more powerful than equivalent PowerPCs, allowing Rosetta to bridge the gap. This is more difficult (but not impossible) with ARM processors.

Also, if all of the computers are ARM, you lose out on virtualization, which is a biggie for many business users.

I believe the intent of the parent poster was actually correct. SSL client certificates [1] are a way to securely authenticate (i.e. login) by using a certificate which you control the key to in addition to the normal SSL certificate for the website.

An over-simplified example: if Alice (A), a client with web browser on a compromised computer, wants to access the BigCorp website (B) she sends a challenge, such as a random string, to B, and B sends back the challenge encrypted with the BigCorp SSL certificate. Alice's browser verifies that the signature corresponds to a valid certificate, which it does because BigCorp's certificate is legitimate and signed by a legitimate CA, and then Alice's web browser and the web server negotiate a session key to use for SSL.

If we introduce an attacker, Eve (E), to this, who has the key to a trusted certificate on Alice's machine, Eve can intercept Alice's original challenge and send back an answer to that using the trusted certificate, which Alice thinks is genuine because the certificate is valid in her browser. This is where certificate pinning helps, since if Alice has visited the site before, the certificate does not match and she gets an error. After the negotiation, Eve then has a session key to both A and B, and any data that A sends, Eve decrypts and re-encrypts with the session key for B, passing it seamlessly onto B, and vice versa for traffic going the other way.

SSL client certificates break for a MITM attack for a different reason. When Alice is using an SSL client certificate to authenticate herself, she has her own certificate on her machine which she knows the private key to (it can even be self-signed). The web server knows Alice's public key (stored like a password hash in a database), and can therefore perform a similar challenge for Alice. When Eve intercepts this challenge, she cannot send a valid response back to the server, as the only valid key for Alice's account is stored on Alice's computer. Therefore, sites that let you sign in with an SSL client certificate often just stop working when you are being MITM'd.

[1] http://www.symantec.com/connect/blogs/client-certificates-vs...

If you don't actually check the fingerprint of the key matches the fingerprint you were given (in some other secure way, such as walking over to the server), then yes, a first-visit MITM is possible.

It's probably still valuable for Microsoft to have an east coast presence, especially if it allows them to recruit from Boston-area tech schools, MIT especially.

Would it be possible to read a value from the array and do something with it (e.g. send it to /dev/null)? Even a JIT shouldn't be able to optimize the value out if you're actually using the value, and as long as you're not zeroing secure memory all the time, the performance hit shouldn't be that large.

This might cause side effects if /dev/null does not exist or is not the null device.

If I recall correctly, most dealerships won't actually charge you that much more to custom-order a car, because the car won't be sitting on their lot for very long compared to the cars they have to keep in stock and hope they sell.

I'm not too versed in this argument, so could someone please explain why Netflix's activities aren't covered by the normal peering agreements that make the "internet backbone" work? Doesn't Netflix pay it's ISP appropriately which gets disseminated by the other ISPs the traffic gets routed through?

From just looking at their website, it seems that they're going to run afoul of banking/money services regulations in nearly every country, if it is not shut down by VISA first.

1. I don't know much about countries other than the US, but in the US you have to be at least 16 with an adult as the joint owner of the card to get a debit card.

2. (Correct me if I'm wrong) I haven't seen a case where you can get a debit card without an ID. No withdrawal limits, no ID verification, and moving it through bitcoin seems like a very easy way to launder money.

To me, it seems useful in cases where the style is a part of the meaning of the symbol. This mostly comes up in mathematics, where a letter represented in fraktur or blackboard bold has some semantically different meaning, and this meaning can be part of the file instead of part of the foramtting of the file.

The practical part of me agrees with patio11, and that the knowledge gain of having these semantics inherent in the file is offset many times by possibly having to treat different bytes as the same character semantically.

Honestly I think that, like version control, time estimation can be a very useful feature for people working on a project alone if they use something like Joel Spolsky's evidence-based scheduling[1]. It would be nice if GitHub provided a way to keep track of that. I agree with you that there's a point where there get to be too many features, but if they're building a project for teams to work on things time estimation seems too useful to pass up.

[1] http://www.joelonsoftware.com/items/2007/10/26.html

I am a self-proclaimed introvert. I tend to think about small "mistakes" I made socially and how I could have done better. However, I enjoy being around people, just a small group of people all of whom I know. I like to be engaged in a conversation, and if put in a situation with a large number of people (such as a party or "social gathering") i will spend most if not all of the time with the same people. It's not that I don't like people, it's just that to connect with them I have to have a conversation with them of greater length than happens at such social gatherings - and it just so happens that the people who enjoy these conversations are other introverts, creating the appearance of the introverts hanging out at the edge of the crowd and talking with other introverts.

AirDrop is great at college: most people have macs where I go to school and don't know what fileshares are. "go to this icon in your sidebar and press accept" is much easier than putting it on a fileshare.

Remember that as time goes on drive space prices fall as well. If you build backblaze storage pods[1] a petabyte over 3 years costs you a little under 100k. If we assume the average email user uses around half a gigabyte[2], we can accommodate around 2.1 million accounts on the petabyte (more if you compress the data). Throw in another two petabytes for compressed backups and you've got a million email accounts for 150k, or $0.15 an account (over three years). If you show users ads that average more than 15 cents per user over the course of three years you're making money (less paying employees that don't operate the storage, etc.). The actual numbers may even be less. Combined with the fact that the price of storage seems to still be going down at a rate that most likely is faster than user inbox growth, and you'll never hit this particular singularity.

[1] http://blog.backblaze.com/2011/07/20/petabytes-on-a-budget-v...

[2] 200MB more than the amount currently in my gmail; I have email going back six years and I never delete emails, but I'm not a heavy email user.

EDIT: Formatting

And conveniently it means that mac app store apps can't be java apps, which leads to them being better integrated and (slightly) higher quality.

I find it interesting that this is a reversal from the days of the spatial finder[1] in OS 9: Apple is implementing the spatial metaphor and people don't want it compared to when they took the metaphor away and people complained.

[1] I can't find a great link to describe this. Suffice to say the paradigm was that each folder had a new window and the windows stayed where you put them.

> Source: Green Onion News Network

Combined with the fact that no state was mentioned, i think this is a work of parody.

What I think part of the problem people have with it is that it doesn't necessarily do what a seasoned git user would expect when syncing remote repositories, such as the fact that it pushes and pulls at the same time, and the pulls aren't normal pulls, they're rebase pulls.

But there are rainbow tables, or tables of all of the MD5s/SHA1s/<insert favorite hash algorithm> for arbitrary strings. So the time's already sunk in.

8 days for one password is a very short amount of time comparatively (tiny for a botnet). If you use bcrypt, which you can force a certain complexity on, you can get that amount of time up much higher.

Is twos compliment behavior guaranteed in C#?