asuffield

> Then things like the NSA leaks happen

So I don't think that's the cause here, because:

https://googleblog.blogspot.co.uk/2013/06/what.html

"Press reports that suggest that Google is providing open-ended access to our users’ data are false, period."

"Any suggestion that Google is disclosing information about our users’ Internet activity on such a scale is completely false."

And you have chosen not to believe a statement made in the strongest possible terms. If you won't believe this, then I do not think there are any words that you would believe, so there's no point in trying to get more published.

> What can you offer me other than root access to your servers?

Indeed, and we can't do that because it would invalidate the very security that you want.

> If up until today, Google was a good company and wasn't abusing anything in any way, and suddenly tomorrow they turned evil, why would they erase my data first? They still have it, and I have no control over whether or not Tomorrow-Google will keep it.

The easy solution here is to not keep any of that data in de-anonymised form for longer than is necessary, but then people don't want to believe the privacy policies which say this is happening...

(Tedious disclaimer: not speaking for anybody else, my opinion only, etc. I'm an SRE at Google. I can't respond to most of the things in this thread, so don't bother asking.)

"If Google were to provide real transparency into the information they track about me, that would be fantastic."

"In no way do I believe that is all the data Google tracks about me."

The second comment here is why there's no point in doing the first one. It doesn't matter how much information we release about this, when people are determined not to believe it.

It is a frustrating experience for the people who work on them to pour time and effort into making sure a privacy policy is really precisely accurate about what is happening, and then see threads like this where people will go looking for loopholes that aren't there, and because English is fairly ambiguous they'll eventually find a way to misinterpret the words to support what they wanted to believe all along.

Here's an interesting hypothetical question: if the primary effect of the conspiracy theorists is to throw bricks at the people who are trying really hard to make sure the bad things never happen, then which side are the conspiracy theorists really on? (I make no claims about whether this is what's happening, because I can offer no evidence, I just think it's an interesting question)

(Tedious disclaimer: not speaking for anybody else, my opinion only, etc. I'm an SRE at Google.)

> All phones comes preloaded with Google Chrome

That's not true. There's still a fair bit of Android Browser out there - it's up to the OEM whether they ship Chrome or their own customised version of Android Browser.

> (apart from sending your every keystroke to the mothership)

I don't believe you. Here's the source, show me where it does this: https://chromium.googlesource.com/chromium/src.git

Here's the privacy policy, which explains what it actually sends: https://www.google.co.uk/chrome/browser/privacy/

(Perhaps you're thinking of the instant search feature? That's quite a long way from "every keystroke", and the page above explains how to turn it off if you don't like it.)

(Tedious disclaimer: not speaking for anybody else, my opinion only, etc. I'm an SRE at Google.)

I'm only going to respond to a couple of parts of this. I more or less disagree with everything you said and even some of the whitespace, but there are only some parts that I can discuss publicly so I'll limit myself to those.

> to figure out what any competent human being would think of as blindingly obvious.

No: to take the things which you thought were obvious, and collect data proving which ones were true and which ones were false. This is no longer "obvious", it is now a proven fact.

Yes, science is a lot more expensive than guessing.

> HR people can and do make those kinds of assumptions all the time. They do it when they are looking at resumes; they do it when they are setting up phone screens

> There's a deeply seated problem happening when you allow people with no subject matter experience be the gatekeepers of who does and doesn't get into a company.

I can state unequivocally that peopleops at Google is not involved in these functions, except to monitor their performance.

> Perhaps every pod at Google doesn't need its own HR rep, but there is a level of management that would make sense to have one HR person to handle hiring/recruiting/necessary paperwork.

We do assign people in something roughly like this fashion, although they aren't really involved in hiring, recruiting, or paperwork.

> your credit is pulled multiple times (making it harder to get a mortgage)

That's a myth. All mortgage credit queries within a 14 day window count as a single query. Lenders have not biased the system to penalise you for shopping for a good deal - they want to make offers to people who show good financial sense by doing that.

Leaking the MAC address of the user's machine is also inconsequential.

(Tedious disclaimer: not speaking for anybody else, my opinion only, etc. I'm an SRE at Google.)

> Many major companies like Google or Netflix use FreeBSD in production.

Erm. Well, the company's so huge that I'm sure there's something with some freebsd in it somewhere, but... not really.

I've worked with freebsd extensively in a past job, where it was used for licensing reasons. I don't agree with this article.

The correct thing to happen in this scenario is for the other founders to buy out your share based on today's valuation (which should be modest but accurate according to what you've been telling investors). Anything else will involve a lot of lawyers and grief.

If they walk away and start their own company, they're handing control of this company to you along with all its investment and property, and that no-poaching clause in your contract says they can't hire away your team. I don't think they want to do that.

I think the primary answer to your question is "because that's exactly the same as the situation in their own home where they live for most of the year".

Is your home inspected and polished to the level of a hotel? If not, why is your concern the one week you spend in a flat that airbnb is connected with?

Bazel is blaze released to the public. Android never used blaze because it had to be built by people outside Google. Different worlds.

So the one thing that you might be missing is that in the engineering org, we don't hire into a job or team, we hire into a ladder. While individual job openings exist and drive the number of people that we're looking to hire at a site, during the recruiting process you aren't really being matched to those. If the job opening is for a SWE, then you get a generic SWE recruiting process and interview panel. If you pass that, then the site managers will figure out which SWE roles at that site you would best fit into - but because everything we do is secret, as an outsider you are basically unable to have any useful input into that process, you just cannot know enough about it before you're hired. The public versions of job postings are missing most of the important details, and the way we do things is too different. We're also far more interested in skills than experience.

Moving between teams is easy once you've been hired (because you've already passed the generic interview for your ladder), and at that point you'll have enough information to make an informed choice. People are expected to move around, and do. This has got a lot to do with why the hiring process works the way it does: you're being hired for all the future jobs you'll do, not the one you get assigned to on day one.

(If this was about roles outside the engineering org, I have no knowledge of how hiring works for those)

> On the positive side, all the Googlers who I've spoken with through the various interviews have been smart and pleasant.

This is in some sense both the goal of our hiring process and the evidence that it's working well.

(Tedious disclaimer: not speaking for anybody else, my opinion only, etc. I'm an SRE at Google.)

> I'm sure I'm just a number to make sure the hiring manager passes some internal threshold so that they can hire the person they want anyways

This is definitely not the case, because we don't have hiring managers (at least in engineering; there are parts of the company that this post doesn't apply to). The system has been designed to definitively prevent that sort of thing from happening. Nobody can say "I want that person" and hire them. You feed people into the hiring pipeline, they get interviewed by a series of calibrated peers, and then a separate committee of people who have never met the candidate reviews the interview feedback. There is no way for any single individual to get somebody hired. This largely removes personal biases from the outcome.

As for your own experiences, the key thing to understand here is that our hiring processes are designed to have a high false negative rate and a near-zero false positive rate. It is explicitly a goal to pass up on many capable people in exchange for not hiring any bad ones. It's not perfect, and there are ways in which people are working to improve the system, but for the most part it's doing what it's designed to do.

If you have the resources to do this then please do, and collect the data and write it up. I'll just note that you wouldn't be replicating this study, which was not randomised (they picked 13 people that the charity was already engaged with). This kind of larger scale randomised experiment is exactly what we need to prove the hypothesis.

That's an interesting experiment, although the sample is too small and uncontrolled to give much confidence in the results (I'm completely confident that if I could pick which people received the money, I could find a hundred people in London who would show the same pattern; to really prove the case we need a controlled random sample). Still, this experiment lends weight to the idea that somebody should run that study.

> It's hard to generalize without a larger-scale study, but the article has some evidence to the contrary

This is an interesting and subtle case: they gave out money, and did some amount of aid work (the article indicates £50k was spent, of which £39k was the handouts, and it's not clear what the rest was spent on), in a city where addiction treatment is free and readily available to anybody who's willing to take it. That combination of things worked, at least in these cases, and I think that's consistent with my expectations that we need combined financial, healthcare, and education support to get really good results. To get more insight we'd need a study which ran the same programme in many locations, including ones where these things are not readily available.

> But if by moderate you mean $100 then of course they'll be begging again tomorrow. It has to offer a new perspective.

And therein lies the tricky part: it's definitely true that if you give somebody enough money, it will solve all these problems for them. Unfortunately it's just unaffordable to raise everybody to the median income. Can we afford enough to make a difference at the population level, even if it's not enough to make all their problems go away? I don't know, and we should keep running more and larger experiments to find out.

If you're also providing them with a treatment programme, yes.

It's clearly nonsensical to withhold money from them that they would otherwise have received. It also doesn't sound like a good idea to expect them to pay for their own treatment from the money you gave them. We need both for this to work.

For long-term unemployed? They usually don't have marketable skills. Bachelors degrees are clearly not the right answer, since those don't usually teach you any.

The usual package in Europe is a combination of remedial math/english/science and apprenticeships in professional trades - things which no plausible "basic income" recipient could afford to get via private tuition, but which can be easily affordable as an organised programme.

I think that it takes a very special kind of deliberate wrongness to hear somebody say "we need healthcare and education in addition to financial support" and conclude "okay, so let's do none of these things".

I don't think that it helps anybody to offer endlessly reduced compromises in an effort to find common ground with people who are willing to do that. Let's not lose sight of the fact that the status quo - even in the US - is already a combined package of financial, medical, and education support. How about just continuing to improve all of them? We already know that they all work well.

It's a lot older than that, and is also the normal practice for governments in most civilised societies: even in the US, there is a progressive tax system and various forms of tax credits and cash benefits to give low-income people more money than they would have otherwise had. That argument has been had in the past by people like Friedman and, at least to some extent, won.

What's more recent is that we've got a few long-term studies like this one providing results showing that no-strings cash benefits do give strong beneficial results. This was long suspected, but that suspicion is now supported by evidence.

I'm sorry, I don't think you fully understood my point, as you appear to be presenting arguments against making financial support conditional on other things, which I was not endorsing here (that is a whole barrel of complicated which I don't want to open).

My point was rather than a financial-assistance-only approach is insufficient: we also need, at minimum, strong education and healthcare support, because there clearly exist groups of people who need them.

There's more depth here than "just give cash".

What we've seen in study after study is that if you take low-income people who are struggling and give them more money, they use it in ways that improve their lives, they become more productive, and this has profound long-term impact.

At the same time, I don't think anybody really disputes that giving moderate amounts of extra cash to a heroin addict is unlikely to improve their lives. They need medical treatment before anything else.

I've never seen a study which looked at extra cash injections for long-term unemployed. That would be an interesting one. I would be unsurprised to find that cash alone was insufficient to solve their problems; the most obvious thing they need is education.

My point here is that cash clearly helps in some - probably most - circumstances. But "just give cash" is insufficient; we still need to work on all the other things as well.

Please consider what kind of behaviour you would encourage. If wealth erodes every year then this strongly disincentives long term thinking and rewards business ventures which give quick cash and then burn out, like strip mining, and punishes planning for the future like pensions.

You want your tax system to reward long term thinking, not punish it.

(Tedious disclaimer: my opinion, not speaking for my employer, etc)

I'm an SRE at Google, where postmortems are habitual. The thing that jumped out at me here is that a production change was instantaneously pushed globally, instead of being canaried on a fraction of the serving capacity so that problems could be detected. That seems like your big problem here.

(Of course, without knowing how your data storage works, it's difficult to tell how hard it is to fix that.)

Sorry, this is incorrect. There are several very similar sounding terms here, which do not mean the same thing.

A "public company" or "publicly traded company" is a company traded on the stock exchange.

A "state company", or "publicly owned company", or "public sector company" is a company owned by a government and not traded at all.

The terms here are a regrettable accident of history, which often confuses people.

It works well for the way a lot of people use facebook, and very poorly for the way some people use facebook. There's a baked in assumption that you only "friend" people who you can identify from their photos.

(This one isn't a fake account, it's a "professional" account, so she's "friends" with a lot of people that she only emails for work purposes and has never met)

For a hilarious example of how this can fail, a guy I used to work with has a facebook account for his pet bird. It is friends with the pet birds of lots of other people. I was present on the occasion when he forgot the password to this account, and spent quite some time trying to guess the names of birds from their photos.

Facebook has a fallback authentication mechanism of showing them some government-issued ID (which could have worked for Violet Blue, and not for my colleague's bird).

> What am I missing?

To pick a few things I can talk about - sufficient replication to not lose data, data security that actually works (small companies can't - and shouldn't - afford disk shredders and internal access audits), targeted obliteration of customer data on request (including getting it out of the backups without rendering the backups worthless), and making sure that your exported data dump which contains massive amounts of personal information is only ever released to the correct user.

Doing any one of these things in isolation is not all that hard, because you can trade off against the others (for example, it is dead easy to do a data dump if you don't care about data security). Doing all of them at the same time is a large, expensive development project that would sink any startup.

At least in the EU there is a regulatory obligation to design your backup system such that data can (eventually) be deleted if the user so requests.

Fortunately there is no strict time limit on completing this process so a design that looks like "and after X months it will have aged out of our backups" appears to be considered reasonable. You are still responsible for making sure your backups are safely destroyed.

(Disclaimer: I work at Google, not speaking for the company, etc)

We can do both. Google stores, backs up, secures, and generally ensures your data is safe, and you can click a button to download it all if you think you can do any of that better.

This is probably too hard to ask every company to do the same, as it would exclude smaller companies who don't have the resources (this takes a lot of time and development!). So it makes sense for them to offer a different feature set, and accept that it's not going to be suitable for some people.

While you can do this, it means your device needs to be on (drawing power, running the radio, not on standby) for the whole stream, so the "normal", preferred model is for the content server to stream directly to the chromecast, and your phone is just the remote control for that process.

Exactly how this works is up to whoever writes the cast apps that you're using. APIs are over here: https://developers.google.com/cast/docs/developers

Announced < shipping. Soon...

Are they really different things? I inform, you advocate, he spews lies and deception.

Just sounds like different perspectives on the same action.