Google.ai
google.ai
google.ai
It would be interesting to know if Google will be making .ai generic, or if they will make a special exception for themselves considering they do not allow others to change the geographical targeting of domains registered to a ccTLD.
[0]: https://support.google.com/webmasters/answer/62399?hl=en&ref...
If Google sites tend to float to the top for you, you're probably an avid Google user (don't forget search results are personalized) or searching for terms in which they rank highly. Extrapolating such an experience to Google taking advantage of their position to create a non-level playing field for search results is a biiiiiiig claim, given the implications. Back it up.
[0]: http://searchengineland.com/google-penalized-breaking-seo-ru...
Almost every single search query returns search results linking to content surrounded by adsense ads. I'm still waiting for them to de-prioritize those webpages, and link to results containing little to no ads. IIRC back when YT still ran exclusively on flash, they de-prioritized other websites for flash content but ignored the offensive over the top ads on YT's front page. Memories fade, but at this point I hazard a guess that even Google's own employees don't believe that their search doesn't favor them.
I am a former Google employee, and your guess is wrong, in my experience. Most I've interacted with on this topic correctly perceive Google search results as important to maintain as neutral, given the consequences for not doing so.
This thread started as discussing search engine ranking for .ai. A claim was then made that Google always prioritizes its own properties, and the person who made that claim discarded the context of search results. We're now talking about Google+ sidebars/shopping boxes, Android integrations, and so on -- that was not at all what I'm talking about. I'm specifically countering the notion that PageRank is manipulated to support Google properties. That's all.
I'm not here to defend Google. (I am actually quite negative on Google.) I'm arresting the claim that Google manipulates search results to promote its own properties, because that's just fake news. Sorry. I am aware the EU and others considered Google+ sidebars and Google Shopping boxes as manipulating results, but they are orthogonal and only manipulate results technically. There is absolutely no code in search that adds PageRank to a property simply because it is owned by Google, and the burden is on the person making that claim to back it up.
Companies go about this type of self-promotion in sneaky ways, because they know such tactics won't last long if the first line of code their employees see is `if 'google' in domain: pagerank * 5`. But it'd be naive to pretend like there aren't some people quietly attempting to promote Google's own properties within Google search results, with varying degrees of self-awareness of this fact, from "completely oblivious" (e.g., the search engineer's assumption that a Google-backed property is more likely to be the "correct" search result for a relevant query, and that therefore there is a problem if Google-backed properties are not highly ranked) to "explicitly tasked with finding subtle ways to favor Google's own offerings within the algorithm".
Large companies are necessarily masters of PR/indirection, and they will work to retain plausible deniability, especially after the MS antitrust case.
Edit: Just saw your edit, yeah, I'm broadening the topic. We'd have to shut down every single comment thread on every website if that never happened :)
Where is the code for PageRank algo? I would like to see it.
Search is a feee service offered by a private company. Users of search don't pay for it so I dont see how monopoly or antitrust legal issues arise. As long as they are honest about what they're doing, and don't manipate ad results (which is a service that does have paying customers) the non ad results could be generated however google sees fit. Since it's free google should be able to set whatever terms it wants - if you don't like it don't use it. No?
I've made this comment in a couple of other threads, which subsequently veered off into other territory, so forgive the repetition, but it's a really interesting topic to me. The open-source distributed tensorflow stuff is pretty nice, but it still requires a huge amount of hand coding and tuning the machinery, reminding me quite a lot of just rolling the damn thing in MPI yourself. I'm very excited to see where distributed tf will be in a year or two, but it's a chore today.
Depending on how much these TPUs and other Cloud ML Engine developments help, I'd gladly abandon the attempt to roll it myself with the distributed tf.
The hope is that using Google's secret sauce to auto-distribute the execution graphs and associated data ingestion makes things "just work". At the moment, the documentation and examples for that are a bit all over the place and require writing models to conform to the newish tf.contrib.learn.Experiment API, which is also a bit underdocumented and underexampled. Using it for very large datasets (say >tens of TB) seems to be pretty challenging at this moment (to me at least). For a lot of use cases, BigTable seems to be the ideal ingestion engine for Cloud ML tf jobs, but there's no C native API. You can use BigTable, but you can only dump complete tables into tensorflow rather than querying for relevant data (since the queries cost money, a 5000-core jobs with just a few queries per core would cost you a fortune, so the ability to query BigTable in the tensorflow reader is disabled.
At any rate, I've been banging around on it for a few weeks and am really hopeful. I will follow Cloud ML Engine's career with considerable interest.
Very excited about the NVidia chips though. Would be happy to run TensorFlow with them on my own hardware - though I'm more excited about the day when client software and hardware make that easy and cheap.
OTOH, I like to think of Google using my data as a form of a vote. The more data they have on me and tailor experiences using my usage data, the more useful it is and it will be designed to reflect that. So while in elections you may only get one vote for your choice of candidate, Google building on my interactions with an app will mean my voice is taken into consideration.
This is one of the reasons why I tend to share my crash/usage data with developers be it Google, Apple, Microsoft, etc.
— Colossus: The Forbin Project, 1970
I think we need to change our attitude here, because I don't think this is right. I think it is a self fulfilling prophecy. There is a certainly middle ground, and we are giving it up if we assume that companies already have everything. It is a false dichotomy to believe we either give all our information up or we go live in a cabin in the woods somewhere. Data minimization with an eye for the actual trade-offs in each situation is key.
I definitely feel google tech, phones included, should be much cheaper for the data their users give them. Their flagship devices are generally price-parity with Apple's - and Apple's not mining your data the way google is (to my knowledge).
And what's the privacy issue with indexing, and heck, still showing ads? As long as they're not sharing your data, where is your privacy being violated by indexing and being shown ads when they already have the same data? That makes no sense.
The issue is "fuck you, I am paying you to use your service without having you getting your greedy paws on my data." If they start indexing data from actual customers, then they're reading it. That's a breach of privacy and contract, plain and simple. And at that point, well, you might as well not pay for their service if it doesn't bring you anything more than a free tier plan.
If I'm paying you to keep my private journal safe yet accessible, you bet I'm going to be pissed if you start telling me at which page something is. Maybe you just remembered a content -> page mapping, but you still read my damn private journal.
My main beef is with what we do and don't call a privacy violation. If your issue was that use of your information or identity by someone else (especially to make money) entitled your to fair compensation (or otherwise it shouldn't be done simply because of unfairness), I would agree with you. If your claim was that a PERSON (or their machine) obtaining any new information about you is a privacy violation, I would get that too. But you're claiming a machine that can violate your privacy merely by indexing and displaying things back to YOURSELF that it already knew. That makes no sense to me. No one is gaining any extra information about you when a computer indexes information it already has, so while it might be unfair, it simply cannot be a privacy violation.
Just hard to imagine companies having their domains shutdown for the acts of an employee who happens to have multi-login enabled.
The ban didn't happen overnight after he enabled multi-login. Took a few months and then got an email from Google that all related accounts and any accounts belonging to me or to this dev will be banned without any notice.
This is the reponse to the admin appealing the ban (Note the words "associated Google Play developer account")
"Hello Mimi,
Thank you for reaching out to the Google Play Team.
After reviewing your appeal, we have confirmed our initial decision and will not be reinstating your developer account.
Your Google Play Developer account has been terminated due to multiple policy violations by associated Google Play developer account. You may also review the Content Policy and the Developer Distribution Agreement.
Note that Google Play Developer Console terminations are associated with developers, and may span multiple account registrations and related Google services. Do not attempt to register a new developer account. Any subsequent registrations will be closed and your developer registration fee will not be refunded.
We recommend that you utilize an alternative method for distributing your apps in the future. Please let us know if you have any other questions or concerns.
Regards, Arya The Google Play Team"
There might be exceptions for university students and the like.
Google still virus scans and allows searching on Apps. It's plenty indexed, even if they're not showing you ads.
Is this in the ToS?
Or put another way, they could currently be holding back 90% of their current research and only releasing 10%.
I'm not saying that is the case, it's just I think private research into some fields is much more opaque than folks (myself) realize. Relying on Google (or any other private company) to publish findings is likely not a path for long-term success.
That why we have company names, brand names, etc.
I occasionally see emails with lists of VP's and I know none of their names, and I work at Google.
If the leaders are thinking they'll thank the team at some point.
https://web.archive.org/web/20160310211414/http://www.hooli....
vs.
Or is "minority" only black and Hispanic now? These goalposts seem to shift so often I can never tell.
The single example of the CEO doesn't tell us anything about the demographic makeup of their engineering force, unless you think the example of the former is a good basis for inferences to the latter.
(But this is admittedly a total derail from the point of this thread.)
https://google.ai/ and https://ai.google/ both point to the same page.
One example: limited.international and international.limited are both available
Both of those appear to be blocked by the registry.
Edit: Nevermind, not blocked but flagged as "premium" and sold at ~double price.
We don't have the reversed version up anymore; it's now just a redirect.
blog.google works though
They recently published very novel approaches: https://openreview.net/pdf?id=rkpACe1lx https://openreview.net/pdf?id=r1Ue8Hcxg
Does anyone else find it odd that we're so far through the looking glass this past year that Richard Hendricks' latest venture seems not only plausible but a bit mundane by comparison?
Both parties get what they want, and nobody loses out. Win win.
In the real world, 99% of us would be bored out of our minds with rapidly-declining health, rapidly-declining self-esteem brought on by our unexpectedly poor handling of time, and a continuing unfulfilled search for stimulation that will mostly get filled by more mindless "entertainment" and consumerism/shopping.
Our social lives would decay as there'd be no in-built workplace community, and this is exacerbated by today's convention of small or totally childless families. We'd lose respect for one another. Productivity would plummet as few people would have any incentive to work cooperatively anymore (and if you can't imperil someone's paycheck, they will usually not contribute; for all of the open-source developers, how many are only in it because their paycheck and/or job prospects are somehow tied to it, and even including that number, how many more closed-source developers are there?). Volunteers don't take firm direction well, and that kind of direction is necessary to keep a successful, efficient enterprise running.
There are plenty of windows into this imaginary world without obligatory labor today, and they do not make it look like a pleasant place.
(Source: I am the eng lead of Google Registry, which run's Google's TLDs.)
Gandi seems to do this as well. I recently tried to buy one through them and they quoted $600, and required a corporate subscription. 101domains had the same one for a little over $100, which implied to me that there's a paper process with people somewhere and each registrar is pricing accordingly (since all .ai are $100/2yr flat).
Good timing on your comment, too, since buying-then-transferring seemed like a good strategy until I read it. Didn't realize .ai was so fundamentally manual. Although[0]:
> We expect that about mid 2017 we will support EPP and other registrars.
Also, we (Charleston Road Registry) are definitely not a small company, but we run ~45 TLDs with a team of engineers much smaller than 45, so we have a ratio of TLDs to engineers significantly higher than 1:1.
//Edited the fee.
That is an interesting avenue to go however, so thanks!
Google seems to be more and more appropriate for The Matrix quotes: https://news.ycombinator.com/item?id=9780632
I think the reason they're pushing so hard, actually, is that they're in a dogfight with IBM (i.e. Watson) over who enterprises will call if some VP gets the idea that they want to "solve problem X with AI."
So many searches I do are so "SEO'ed", I feel like it's 1999 again.
It also tries to "help" too much with fuzzy matching which starts to make it useless if you are looking for less common thing. For example, if you search for "nmake tabs vs spaces" it returns a bunch of results for GNU make and flame wars about tabs vs spaces instead of nmake specific info regarding usage of tabs or spaces in nmake makefiles.
I find myself preemptively using quotes more and more with Google. I kind of which their logic was segmented by user-type.
http://www.zdnet.com/article/googles-marissa-mayer-speed-win...
Wow. The price is quite good. Is there some website you bought this from? How many cores?
>>But a 24-core new workstation would be much more expensive
Tell me about it. I've been checking the prices of a dual 24 core workstation and it is around $7K.
http://www.ebay.com/itm/HP-Z800-Workstation-2x-Xeon-x5670-2-...
The one downside of this box is that the power supply is totally proprietary to HP. Well, the motherboard too. So if one of those craps out, you're done, unless you can locate a spare cheaply.
It's also reeeeaaally heavy and draws a lot of power.
What does Xeon have to do with training? You need a fast GPU (or apparently TPU). The CPU is relatively unimportant. Furthermore, you can roll your own algorithms (i.e. architecture) in TensorFlow. You don't have to use "theirs".
Now, you may think I'm crazy for designing something from scratch, I may be indeed, but how else can we discover/invent something totally new. I think I'm actually onto something given that the early results look quite promising.
I actually created a video demonstration of my AI Engine which you can find a link to it in my profile but I have done a crappy job of explaining the strengths of it [1]. Too much work, so little time.
Where I get stuck is figuring out new challenges to throw at the system. I find it funny that there is lots of discussion here about performance and tuning, and not so much about practice applications and valuable problems to solve.
Thank-you for reminding me why I don't use Gboard and instead the BlackBerry Priv's fine keyboard instead. The obsession of prediction in our culture is absurd.
Doing something myself is one thing, having it done for me is convenient, having it done for me before I asked is even better (I do not mean in all situations, only in some, and only for some things, etc...)
Been looking for something more secure than vanilla Android.
Concerning security, I think its basically an encrypted Android. The custom permissions feature is nice too but I think that's on generic Android too. It would be nice if there was an option to have no Google Play apps, but I have all permissions off on them anyways (except Gmail and Drive.)
Most importantly, AI and machine learnings are not synonym at all. People should regard AI as the overall goal, wanting computer to do something really smart on its own, very little to no instruction.
Going back 5-7 years ago when Siri first came out, it was quite a noise. But I honestly never found a compelling reason to use Siri until I started driving and I needed to call somebody. The problem is that I have some accent and I have a lazy tongue so I blur on words, so Siri does not always understand what I want to say. I am surprised the voice-to-text feature in Message is quite accurate (it can auto-correct by learning the next phrase and understands pauses so it waits for you to speak again), but Siri doesn't. So while I appreciate virtual assistant, their capabilities are very limited to a set of commands.
I do feel the AI community has made some good progress, from beating Mario game, beating top Go players, to self-driving car, the technologies supporting these initiatives are getting more sophisticated than ever (and I feel the tooling too is getting too competitive, too many choices). I am working on some simple home automation involving NLP (for speaking to the program), image recognition (who's in the house), and a couple self-execution routines such as make sure all nights are off if no one is in the house or reminds me doctor appointment every Wednesday. That's not AI, it doesn't do anything else beyond what I programmed it to do, it doesn't try to survive or better itself.
I know, they're coming soon, but I've been waiting for years and at some point the enthusiasm wanes when it's been right around the corner for several years.
Hey, maybe they're trying to solve a problem that's harder than anyone expected.
Anyways, I'll be impressed when I can have one.
Comparing pre-alpha with fielded is no comparison. Nuclear weapons haven't been going on and on about how great they'll eventually be.
I'd love a place where I can grok and find only useful info. Comments like this don't help.
> The worst thing to post or upvote is something that's intensely but shallowly interesting: gossip about famous people, funny or cute pictures or videos, partisan political articles, etc. If you let that sort of thing onto a news site, it will push aside the deeply interesting stuff, which tends to be quieter.
Intellectual curiosity often takes up seemingly trivial details and plays with them for no particular reason. Mostly nothing important comes of it, but sometimes something really does. In any case it would be a big mistake to try to push that kind of thing out of here—it's well within the scope of what HN exists for.
We detached this subthread from https://news.ycombinator.com/item?id=14361923 and marked it off-topic.
I think it was unfortunate that this generic landing page was chosen as the golden submission and other more specific and more technical submissions (such as https://news.ycombinator.com/item?id=14360653) were duped off the front page. If a more specific article had been chosen instead, we might have had a more focused technical discussion rather than domain name commentary and the same old non-specific privacy concerns discussed ad nauseam every day.
You know what, you're exactly right. Generic pages lead to generic discussion and that is uninteresting. We're well aware of that effect, so usually make a point of penalizing generic portal-style pages or changing the HN URLs to more substantive ones. For some reason we missed this case.
It's interesting as yet another demonstration of how reliable that effect is, I guess. Initial conditions have a huge impact on HN threads. First comment is another.
> The test for substance is a lot like it is for links. Does your comment teach us anything? There are two ways to do that: by pointing out some consideration that hadn't previously been mentioned, and by giving more information about the topic, perhaps from personal experience. Whereas comments like "LOL!" or worse still, "That's retarded!" teach us nothing.
As you may have noticed, there are a couple of comments regarding the use of the .ai TLD. A legitimate conversation can be held on whether or not .ai should be generalized for "Artificial Intelligence", possibly to the detriment of people in Anguilla.
The only site on the internet that will meet your requirements of "a place where I can grok and find only useful info" is one you write yourself. What you consider to be "useful info" is going to differ from what other people think.
And, of course, you should consider that your comment is, if anything, a more egregious example of a useless comment, detracting from the conversation.
You can hide comments with the [-] symbol. Use it.
As good a time as any for a reminder of how this philosophy actually plays out at Google: https://www.theatlantic.com/technology/archive/2017/04/the-t...
https://www.forbes.com/sites/kashmirhill/2014/06/28/facebook...
There's no comparing the incentives, resources, and potential for harm between the state and citizens.
The point being, the harm the government is doing right now is obscured from the people, but with AI trawling all public data, these issues can be revealed much sooner.
Giving each side force multipliers doesn't even the playing field, it makes the absolute gap even larger.
Secondly, your point would be valid if the force multipliers applied equally. This likely isn't true.
So which side gains more advantage? If the government could sufficiently obscure their actions such that the data weren't available to analyze, that would give it the advantage. However, the government is constrained by certain transparency requirements that don't apply to citizens, so by default, citizens know more about their representatives than their representatives know about their citizens.
The executive branch losing a court case isn't answering to the people. It's answering to the judicial branch. Being granted explicit permission as a citizen to do something you legally should be allowed to do isn't gaining anything at all. An innocent person being allowed to remain innocent again isn't giving anything to citizens. Those are basic rights..
> If the government could sufficiently obscure their actions such that the data weren't available to analyze, that would give it the advantage
The govt already does this across the board. I'm not certain how you could argue otherwise. Taking a issue in the news recently, we still don't have national rates on use of force by the police across the country. As in literally no-one knows what the police is doing because centralized records aren't kept, and every area does it differently. Police departments have fought body cameras. Fought enabling citizens to even see how they are executing their jobs.
It was a sliver of data available under Obama (although to his credit, his admin tried to improve it), and yet: https://www.engadget.com/2017/04/14/trump-admin-killing-open...
Statisticians already do this with elections around the world, for instance.
And the chocolate ration has been increased to 20g this week!
(apologies to George Orwell)
Google decides if you are part of the everyone included in their AI for everyone
This is empirically not true, especially if you look at the policies of and donations by Google, Twitter, and Facebook.
Google donates to both parties; Twitter obviously has a very laissez-faire to the Right's voices, including the POTUS, and its COO recently suggested that if Trump canceled briefings, he could take questions directly on Twitter. And the evidence and rumors continue to swirl about how multiple Republican Party candidates campaigns used Facebook's own incredibly specific ad targeting, which they did not revise until after the election. (cf. Cambridge Analytica, contracted by Cruz and then Trump to "manage" the election outcome.)
[1] http://thehill.com/blogs/pundits-blog/media/303295-how-faceb...
[2] http://gizmodo.com/former-facebook-workers-we-routinely-supp...
I'm curious, what in your opinion are some "right leaning" fact checkers or "agnotistic" fact checkers?
We're all human and we have our biases, whether we like it or not. It doesn't take much effort to tweak an assertion to go from "Mostly false" to "mostly true" - it just depends on which one the writer favors. My BS detector goes haywire whenever someone claims to offer non-subjective anything. tl;dr investigate facts and claims yourself and never trust those who have power over you, even if you vote for them!
Also, what are some "right-leaning" fact checkers then?
Twitter isn't laissez-faire with right leaning voices. They actively suppress them. They remove hashtags from trending daily. They mark posts as containing explicit content. It's getting ridiculous how much all of these companies are trying to silence the right.
Facebook will take almost anyone's money on their ads. But they were also caught this last election cycle actively suppressing conservatives. It got so bad Zuckerberg had to come out and do damage control by inviting conservative commentators to a sit-down or piss off half his US userbase.