I don't think there are actually major legal concerns. Copyright protects reproduction of a specific image. Looking at an image and producing something in a similar style is not copyright infringement, it's called being an artist. The law on this seems pretty clear.
The UK recently announced plans to make this completely explicit, to remove any remaining doubt:
"For text and data mining, we plan to introduce a new copyright and database exception which allows TDM for any purpose. Rights holders will still have safeguards to protect their content, including a requirement for lawful access."
I was wondering about trademark issues with a model that can draw new pictures of Mickey Mouse/Homer Simpson/Hatsune Miku if prompted with their names.
Google's arrogance here is astonishing. The police say no fault, and it's the subject of an NYT investigation, and they still won't restore the account. What hope do the rest of us have?
I've been an Android user for a long time, but I think this might finally push me to switch to Apple. I'm just disgusted by this.
The Eu starting to fine the living daylights out of Google for not allowing people get their data from their account per the new digital gatekeepers act that the big tech has 6 months to comply.
They did it to Microsoft when Microsoft refused to comply with the browser ballot box initiative, saying that it was "technically near impossible". They started fining them 500,000 Eur/day or something. Then Microsoft magically made the ballot box happen within 2 weeks.
Apparently having a "0 tolerance policy" on something means that even when an accusation is proven to be false, you'll still punish the accused. I am disgusted and Google should be ashamed.
You’re right, they’re not comparable: Apple was going to be scanning for CSAM on your device, without the photos even reaching their servers (by uploading a backup, texting a copy, etc). Google, so far as I can tell and as corroborated by the article, doesn’t do that. Apple and Google both already scan anything uploaded to your Photos account for CSAM and report to authorities.
No,Apple's proposal was for only scanning stuff that was being uploaded to their servers. And only matching hashes with known material, not general "nude child!" detection.
While there's problems with that too, it is a better design than what Google does.
As the article states, the scanning only occurred if you had iCloud Photos enabled. From what I recall, it worked like:
1. Use on device ML to scan for child porn before iCloud Photos upload.
2. If not found, issue a cryptographic ticket for iCloud upload and include in upload request.
…the whole point of Apple’s scheme as far as I know was to keep as much of the processing on device as possible while also keeping illegal content off of their cloud servers.
Yeah, Apple was annoying (not even that really, it’s just the idea rather than the actual effect that bothers me), but ultimately fairly benign compared to what Google does.
Some of the tools for interpretability can be useful (particularly for debugging), but I think the broader idea that we always need to be able to understand our own models is basically wrong.
For example, if you want AlphaGo to explain why it made a particular Go move, what kind of an explanation is possible? In many cases the only explanation may be that the move leads to a higher probability of a win. There simply may not be a more “compressed” or “high level” answer. Even human Go players often cannot explain why they choose particular moves, other than references to shape and feel, which is basically another way of saying their evaluation of the move leads to a higher win probability. There are a lot of domains where we may just have to accept that that _is_ the explaination.
To zoom out a bit, our greatest discoveries have historically been about finding the rare places where the universe is computationally compressible. Boiling a kettle is almost unimaginably complex to describe in terms of the interaction of elementary particles. But you can make very good predictions about that process using an equation that fits on a cocktail napkin. There may be other areas in which the universe is compressible only to a lesser extent. The parameters of AlphaGo are vanishingly small compared to the size of the Go game tree, but are very large compared to the equation we can use to predict the kettle. There may be many problems where the best descriptions lie in this intermediate domain, a domain which we have never really had access to before (except via biological brains).
So if learned models give us access to some truths without access to their (human intelligible) explanations, I think we need to just embrace that. If you allow yourself a new way of seeing, you can see new things.
> So if learned models give us access to some truths without access to their (human intelligible) explanations, I think we need to just embrace that.
The question is: if the model is not interpretable or understandable, how do you know that what it gives you is, in fact, truth?
You basically need some kind of external validation of the results. In the case of Go, the rules of the game and competion basically provide that in a very authoritative way. I don't think this is the case for all that many domains.
As I see it, behind the desire for interpretability there are two main concerns:
* could a model that gives really good answers in all the cases we have tested still give catastrophically wrong answers in some cases we have not foreseen?
* could the model be relying on some flaw or bias in the training data which we haven't realized?
In younger days I'd of questioned you to the nth degree to rationalise it all. But with age, I've learned that it's cool to just not like something - after all, we happily accept such personal choices when it comes to food. So why not with some arbitrary software more suited as an outlet for any negativity. Indeed, the World would be a far better place if they just focused their hate onto some bit of software.
So with that said, I totally respect your irrational burning hatred for AMP.
Is the email address something someone else might enter by accident? I constantly get emails intended for someone who has the same surname and first initial as me (and presumably a very similar email). She signs up for all sorts of things and accidentally enters my email (it's quite annoying). I even got her plane tickets once. And I can see loads of FB advertiser matches that are clearly her rather than me.
Advertisers want more control over who they target than just geographic regions. I'd bet that Facebook let's you select demographic targeting using all of the information they hold on your contacts, so by uploading every phone number you can target people based on geography, age, income, family size, etc.
As a European start-up founder, I agree with this. The unintended consequences of the GDPR will be to (1) cement the power of Google and Facebook, since they have logged in consent (2) kill third party retargeting companies, ironically French company Criteo being one of the promient examples (3) generally make life harder for Euppean companies, and sharpen the incentive to ignore the EU market where possible.
As a user I do feel some privacy regulation is a good thing. There are certainly good things about GDPR (e.g. a lot of third party ad tracking deserves to die). However, I am far from convinced that the GDPR is the saviour some people think it is.
> (1) cement the power of Google and Facebook, since they have logged in consent
I might be missing something, but won't they actually have to get your consent again, on everything, but this time in details and with option for you to refuse without service degradation?
> (2) kill third party retargeting companies
Good riddance. That would be the second most annoying ad technique today, after ads interrupting your video streaming.
So, both Google and FB track people around the web, as do many web-retargeters. The web-retargeters have no reasonable way to get consent, but big G and F do, so it's likely that they will fill the gap in the advertising markets created by the lack of web-retargeters, thus increasing their power.
The DGraph team have done some very interesting work. The recent addition of transactions is an extremely impressive technical accomplishment, and a very good long term move. Excellent to see a production-ready 1.0 release.
Like many universities, the value of YC to founders is mostly signalling and network. What you actually learn at YC is a minor component of the value (since you could learn most of it just by reading or watching videos)
As YC batches get larger, the signalling value of being a YC company goes down. If everyone is special then nobody is.
As YC batches get larger, I suspect the value of the YC network also goes down. This is maybe a little bit counter-intuitive, but I think a large influx of less highly-selected startups will dilute the network and prevent it from functioning effectively.
Unfortunately YC is not completely aligned with founders here. There is a strong temptation for YC to increase batch size. YC gets 7% of a company for far below market price. So increasing batch sizes makes short-term economic sense. However, if YC increases batch sizes too much and harms its brand, it will eventually break its own model. So increasing batch sizes could be seen as form of spending down the brand equity.
This all hangs on a few premises that may or may not be true:
1) You can't have larger batches while keeping quality constant.
2) YC doesn't directly influence the success of participants that much.
3) If the value of YC drops, good founders will eventually stop applying
I suspect an argument like this has been going on inside YC for quite a while. And the blog post was written by the original founders to try to settle the dispute in favour of the pro-scale faction.
1) I think the value of most universities (top universities) is the quality of the students they are able to attract. It turns out that becoming friends with smart and motivated people can be very very helpful. It also turns out that most skills in life are learned on the job - not in school. YC is very similar in this regard.
2) Being a participant in YC in '07 and 12 I can tell you that the value of YC increases as it gets bigger because there are more smart, motivated, and successful people to learn from and be motivated by. Also especially for b2b companies - there are a lot of potential customers - see Gusto.
3) YC has no incentive to increase in size if it can't maintain a high-quality program because top founders only want to participate in a program with other top founders. If we do something that convinces top founders to stay away - then we lose.
4) There is this strange assumption that YC had higher quality founders earlier in its life. I think that assumption is biased by the fact that it takes 10 years to build a successful company and so the breakout successes always look like they are from the distant past.
5) "a large influx of less highly-selected startups" - where is your data to back this up? There are significantly more applications to YC now than 10 years ago. The acceptance rate is significantly lower than it was before.
But ycombinator itself in lots of ways turned out to be a credential network like an old school law firm. Even including early associates becoming promoted to partners, etc.
But the ycombinator application process itself at least still doesn't rely on much if any credentialism.
"When we were kids I used to annoy my sister by ordering her to do things I knew she was about to do anyway. As credentials are superseded by performance, a similar role is the best former gatekeepers can hope for. "
Maybe that is close to the role Y Combinator has ended up with as it expands. Getting a slice of startups that would be persuing the same thing with or without Y Combinator, in exchange for acting as a credential and signal to investors.
But there is a big peer-network component to it too, which is more similar to universities than to law firms. There are sort of semesters and the startups identify by "class year." Though that probably happens a bit a the bigger law firms too. Entering associates get a network out of it even if they don't go on to make partner and end up in government positions or positions at other firms.
>3) YC has no incentive to increase in size if it can't maintain a high-quality program because top founders only want to participate in a program with other top founders. If we do something that convinces top founders to stay away - then we lose.
It says right in the post - YC's goal is mass production of startups via incubation.
I think Michael is saying "Just like it would be dumb to increase the size of a factory too quickly, it would be dumb to increase the size of our batches too quickly."
We're always looking for more ways to help more startups, sometimes within the framework of our original incubation program and sometimes without it.
If YC lives up to its potential, I suspect the startups that actually end up going through our accelerator will be just a small fraction of all the ones we've meaningfully helped out in one way or another.
> (since you could learn most of it just by reading or watching videos)
I think this misunderstands how most people learn -- people generally learn much more effectively by "active learning," aka actually trying to do the thing they are learning, and getting dynamic guidance alongside that active process, not canned lecture / video content.
> It's not an insurance policy at all. It's a crazy tournament. It's a zero-sum tournament. If you were the president of Harvard or Stanford and you wanted to get a lynch mob of students, alumni, and faculty to come after you, what you should say is something like this: We live in this much larger, more global world. We offer this great education to everybody. So we're going to double or triple our enrollment over the next 15 to 20 years. And people would all be furious, because the value of the degree comes from massive exclusion. And what you're really running is something like a Studio 54 night club that's got an incredibly long line outside and a very small number of people let inside. It's branded as positive sum, everybody can learn, but the reality is that it is deeply zero sum. -- Peter Thiel (former YC partner)
120K for 7% at SF area is indeed not attractive, not to mention you must move there and find your own working space etc. It might be interesting for the very early stage startups and money-hungry (no-kids-yet) young founders. Yes the connections and advice are priceless but still money talks the loudest for most start-ups.
Both my co-founder and I have kids. I would have given up 15% to do YC if I had to. The money is beside the point, but it’s enough, even in the Bay Area.
It was, hands down, the best and most important thing our company has done.
And we were profitable before YC began, so we didn’t need it.
So why is the YC/120K the most important thing your company has done even though you don't need it financially, YC somehow helps you to expand quickly?
The YC provided money is the least important benefit to doing YC. The parent very specifically notes that money wasn't the most important thing, you appear to have glossed over that.
The value of getting accepted and how that acts as a filtering mechanism to prime venture firms. The network: YC alumni, investors, executives in the industry, mentors, etc.
In other words, getting into YC will usually: enable you to raise venture capital from the best VC firms far more easily; it'll open doors that you previously likely could not open (especially for newer entrepreneurs); it'll help you with recruiting talented and experienced employees (critical to success); it's a halo, by getting accepted into it, you get to borrow reputation from what YC has built up.
We aren't this either. 2 cofounders 6 people. 1 with a kid, startup around for roughly 2.5 years before doing YC.
In a year since YC we've more than tripled the team to more than 30 people now, of which again: most have kids.
Ultimately for us, the network among other things has been the best part. I would definitely do it again.
That being said, my anecdotal opinion: A weak point for YC is enterprise. If you want additional customers, there are a few big YC companies, but not many. Even then, when we did it, we found some help and still got great advice.
If you sell to startups though, it's actually a great way to get customers.
Something that hasn't really been around long enough yet, but also potential is the new vertical accelerator.
Our batch was a bit too early for it, but they are trying to solve real problems these companies h ave.
I was 35 and had two kids when I went through YC in 2013. My cofounder was roughly the same age and also had two kids. We weren’t the oldest and we didn’t have the most kids. The idea that YC is just for 20-somethings and consumer/tech apps is a myth that may have been true early on, but is no longer based in reality.
It does help to have a low personal burn rate. If you have high expenses, it’s difficult to quit a decently well paying job to start a startup. But that doesn’t have anything to do with YC. If anything, it’s marginally easier with YC backing because at least you can slow down the personal burn.
That effectively sets the average valuation at $1.7 million for every startup accepted into YC. While I'm sure there are exceptions, startups with really strong initial traction, that seems like a fairly reasonable (if not elevated) valuation.
> Like many universities, the value of YC to founders is mostly signalling and network. What you actually learn at YC is a minor component of the value (since you could learn most of it just by reading or watching videos)
While this sounds true to an extent, YC can provide more tailored feedback than a university generally does. Most of what one does in university is generic and can be learned through tutorials, but YC can provide custom feedback to a specific startup's situation. (And good educational programs will also focus on providing feedback on individual projects.)
I also think another factor is over the past 10 years, there has more investor money thrown at early stage startups. That just means that it harder for companies to break out. Due to that fact I am not sure accelerators even YC will provide much more advantages.
I've always wondered about the viability of an email client that behaves like a chat client. Instant send with the enter key, messages in bubbles, etc. You get all the fun of IM without additional infrastructure or services, and in a standard way too. No walled gardens of different apps.
Could be a pretty cool UI solution to have a chat client that uses PGP encrypted e-mail in the background, but Gmail SMTP has a send limit of 100 e-mails per day. You would run into that rather quickly with a chat UI.
I guess the chat client could do all the e-mail sending from some of their maintainer e-mail servers and not from under the chat participant e-mail. Then we would start losing comaptibility with other e-mail clients though. Although not completely, as we could set Reply-To headers and when another e-mail client replies to user1234@chat-client.com, we could just forward it to the correct e-mail.
Kind of. Imagine more like Google Groups or HN but with notifications when someone replied.
I guess if you use mailing lists for topics you can subscribe to, and cc'ing people for @mentions inviting them to reply, then yeah! Email can do it all. So why is email considered such a time sink? Because people don't filter conversations coming from the outside. But if they just set up a filter, then literally email is a Slack killer, no?
The UK recently announced plans to make this completely explicit, to remove any remaining doubt: "For text and data mining, we plan to introduce a new copyright and database exception which allows TDM for any purpose. Rights holders will still have safeguards to protect their content, including a requirement for lawful access."
https://www.gov.uk/government/consultations/artificial-intel...