bwang29

I don't understand why most of the comment here are negative. On the bright side, being able to train a model in any sort of cloud (doesn't matter if it is Azure, Google, AWS, or IBM) and then directly create CoreML models, and iterate from model feedback, is a pretty special and unique position. It's a hybrid edge + cloud approach to running A.I. for end applications, and I do see it to benefit developers who want to roll their own models and create new applications particularly to computer vision, photography, and videos.

I think visually speaking, while the deblocked version is smoother overall, the blockiness added extra perception of noise/texture. The extra noise almost always increase the perception of sharpness and level of details, and you can see from the braids coming out of the hat that it does look sharper with the blocking version, although the wall is smoother. Would be great to include more sample images in the page.

I just thought there is a pattern on the "Are Older Than Scientists Thought" part to make a discovery exciting (it seems to be what is indeed exciting here), a quick search on "are older than scientists thoughts " you get

Universe Older Than Thought... The moon is older than scientists thought.. Land plants are older than scientists thought... Written zero 500 years older than scientists thought -... Your dog is 10,000 years older than scientists thought... Human Species May Be Much Older Than Previously Thought ...

Some [adjective] [commonly known subject matter/celebrities]'s [attribute] is/are [older/newer/bigger/smaller/...] than scientists/we thought.

I found this to also be great for sales pitch or any pitch in general, or to convince oneself for main life decisions (move to a new city, found a job, or buy a new expensive item).

I think the article is saying that there are way too many confounding variables in determining the probabilistic nature of longevity and so far due to the fragments of data, contradictions in data, challenges in tracking, uncaptured info, it is extremely difficulty to meaningful understand what people could do to slowdown aging without running controlled study on thousands if not millions of people.

However, say someone really wants to figure this out, they really want to know every component of every meal people eat, every activity or habit and every moment of their lives in terms of heart rate and metabolism and provide a device to easily track all these and make sure the data is for anonymized for research only, and they managed to convince millions of people to implant such device at their birth, and they figured this out in 500 years and found a formula to extend life indefinitely, I wonder if this would introduce the classic “the rich/wealthy lives forever” scenario, or the alternative scenario where the knowledge of longevity could hurt genetic evolution as a whole.

I think combat games teach a lot about improvisation, especially if you're playing against another human, you'd have to adapt in realtime the pattern the other person is using and trying to develop a strategy to counter the other person's attack/defense. In recent combat games, the startup/cooldown time are so precise that pro players literally need to remember, compute and combine maths in their game plays in real time into combos.

I definitely feel games like Sims and Simcity changed my perception of resource safety as a child. I played Sims/Simcity2000~3000/Civs for hundreds of hours as a kid, and I loved the "realistic" feels and setups of these games. When I started to play I always had later stage trouble when I'm overspending on building fancy roads and bridges in SimCity or buying expensive items up front. I slowly learned the way of setting up a less impressive base first, and swapping out parts and do re-designs and expansions as more resources became available.

I later found out all games have a specific pattern you follow to win, many of these patterns are available in online communities/guides/cheat sections where I'd spend tons of time to read, try and experiment. Those were probably the peak happiness and best time I've enjoyed as a little nerdy kid. I never thought too much a big deal of these experiences but nonetheless I believe definitely contributed to a base reality of how resource management work, and how I manage my personal projects, finance, and career choices as I grew up.

I'd be super curious to see research / longitudinal study coming up from the past 2-3 decades of gaming on kids, and see if conservative gaming styles leads to conservative financial decisions etc etc.

I'm not a father yet but I thought it would be a really cool idea to understand my kid through game play like this, but I also don't want the kids to be limited by my personal value though these interactions - sort of like the "observer and supporter" role the author played.

Sidenote: I'm still playing the latest Civ6 and still enjoy a lot from the dynamics in the economy managements with A.I..

One thing that might not be related but I found fascinating is that Jupiter is “木星” in Chinese, where “木” means wooden. The first time “木星” was mentioned in Chinese literature was in 东汉, year 25 - year 220 (around 1800 years ago). So Chinese people back then thought the planet was made of wood! And as a child, this planet has the name easiest to remember, because it looks exactly like a wooden ball.

Edit: I digged deeper and found that it was a coincidence that Jupiter was called the Wooden Planet. The Chinese assigned the name not because they were able to observe the wood like texture (due to the lack of telescope), but they were able to observe the existence of the planet with naked eye as early as 2300 years ago, along with the Mercury, Venus, Mars, and Saturn. These five planets were assigned to the name in the order of "金 gold (Venus)","木 wood (Jupiter)","水 water (Mercury)","火 fire (Mars)"."土 rock (Saturn)" superstitiously. What's really surprising is Venus does look like gold in color and Saturn does look like wood in color. What a coincidence!

Can I contact you on the training data idea?

Can I reach out to you somehow @gascan ? Want to learn more.

There is already a "sesame" score built in Sesame Credit by Alibaba (https://en.wikipedia.org/wiki/Sesame_Credit)

The Wired article is a little more exaggerating (https://www.wired.com/story/age-of-social-credit/) but here is the rough breakdown of your score according to https://www.zhihu.com/question/27844479

1. Identity Characteristics (15%) Public Security Real Name Authentication Identity Information Information Stability ... 2. Credit History (35%) Credit Card Repayment History Micronesia Repayment Record Utilities Fee Ticket ... 3. Compliance Capability (20%) Payment Account balance balance balance balance car information real estate information ...... 4. Personal connections (5%) relationship circle circle of friends credit level of social influence ... ... 5. behavior preferences (25%) account activity level consumer level payment level consumer preferences ...

This would be great for DSLRs if integrated especially those who are shooting wildlife at high speed. Contrary to my own previous belief, I thought I would need a smaller hard drive in terms of storage space when I have a larger or faster SD card, but the opposite is true. The faster and bigger the SD card is the more photos I ended up shooting with more duplicates and the more of a need for a home NAS or bigger hard drive.. I think the issue is I never considered SD card as permanent storage media and I would always create a copy of it somewhere else.

For those who wants to find the "Crucial Message" quickly, I think it is hidden in the 4th last paragraph :

"That’s why there is and will be an even greater pushback against the idea that a small group of companies and executives can reap vast rewards, dictate the architecture of the online, cloud, AI, and robotic worlds and maintain an insular, parochial, and narrow worldview marked by groupthink."

Or was it somewhere else?

To be fair, it is not a small group of companies, but rather clusters of companies working decades together to make it all happen, and I was hoping there was at least a tiny bit of appreciation in the article. But I quickly realized that while the author did the name dropping he had to point out Amazon is not technically in the Valley but part of the equation. I was hoping he could find companies like Western Digital is also making most of the hard drives that contribute to whatever agenda he thinks the silicon valley is doing.

A photographer could accidentally upload or AT SOME POINT labelled his/her photos in the free domain, but ALSO upload with a license to a paid photo stock site. This is usually the root cause of most issues. Another possible reason is the photographer's harddrive got hacked, camera stolen etc, and the photos in full resolution was uploaded maliciously to free CC0 websites. Both scenarios are hard to trace and hard to attribute when an infringement happens, because it involves too many party to do diligence and collect evidences. Most likely the party that uses the photo will have to pay.

Canon and Nikon still provides no on system encryption although photographers have been asked for it for years, so basically if someone stole your camera, they also owns the copyright of photos from those RAW files.

This is almost a sad day for the photo community. VCG has been suing Chinese startups and companies for licensing rights for a while [1], many of the licensed photographs appear in "free" or "CC0" websites. A year ago, we received a request of paying a 50k RMB penalty for a set of images downloaded from free CC0 websites as well as a set of images previously appeared in PetaPixel which then quoted from 500px in 2014 (this was before VCG's investment in 500px) - however VCG claims they have copyrights on them. The way to waive this penalty is to buy VCG bulk licenses or sign agreements with VCG to purchase their products (if you buy stuff from us, we won't charge you from the previous infringement issues!). When asked about why these photos show up in full resolution in free websites in the first place, they didn't provide a clear answer. One reason is that it is difficult to shutdown "pirated" and free image hosting sites created under individuals. For all these penalty or payments collected, VCG almost never repay the contributor or photographer. But I guess neither does their counterparts in the U.S. (this is a much larger issue that I won't expand here).

So if company A used 500px's image in 2014 before VCG's investment, and company B quoted company A's article, and you re-tweet or re-blogged company B's article, VCG (who now has copyrights in China for 500px images) might go after you to collect penalty but never contact A nor B.

Advise to startups : to make sure you don't infringe any copyrights, do a reverse search on VCG's database (especially you have market in China) as well as Getty's/shutter stock's database and never trust unsplash-ish websites (they also have no model releases that will get you in trouble). These stuff could bite you back in years if you have a photo floating somewhere in your public Facebook, Twitter or Instagram account.

Now that VCG has acquired 100% of 500px, what would (or could) happen is they will crawl anyone who has referenced 500px photos in China and start to threaten to sue in exchange for opportunity to gain more subscribers for their media library. Last year a couple of social media accounts on Wechat was forced to shutdown by this [2]. VCG quotes different penalty for different images, and it is usually 1000 RMB per image.

[1] https://xueqiu.com/8818667120/74351790

[2] https://www.zhihu.com/question/48400883

I developed new appreciation for car manufacturing from this article. The way that one component depends on other common components reminded me of NPM packages.

I'm wondering what if they found one car infringing another car's design or engineering IP without fulfilling a license, or if they would find certain car makers "reinvent the wheel" simultaneously. A lot of car makers need to properly license certain design from other car makers. I wonder if this also helps facilitating and defending these licensable designs, or encouraging open sourcing design in cars.

I suspect the overall outcome of finding similar design and borrow existing design from one car maker into another is a good thing for bringing the cost down.

The article mentioned that "Tesla is striving and failing to build a pretty simple vehicle. The Model 3 is basically an electric Honda Accord. And Honda without noticeable effort builds and sells over 100,000 of those every single month in the US alone.".

I've always been too afraid to ask, with all the knowledge and best practices in car making, even consider Tesla isn't willing to hire other factories to make the car, what exactly is costing so much more time than it seems to its competition in terms of making car? Is it because Tesla bet too much time and resource on a much more autonomous production line which failed to deliver?

Title of article : "Apple Says All Macs, iPhones and iPads Exposed to Chip Security Flaws"

First paragraph: "Apple Inc. said all Mac computers and iOS devices, like iPhones and iPads, are affected by chip security flaws unearthed this week, but the company stressed there are no known exploits impacting users."

It would be silly if actual human file these claims so the claims are very likely to be filed by some algorithm, which somehow found matching pattern in the noise.

Maybe:

1) Some copyrighted music has a period of noise that closely matched the segment of noise of the video.

2) There is adversarial pattern in the noise in the video that confused the hunting algorithm.

Quote from article "We need to move faster. Validation at Intel is taking much longer than it does for our competition. We need to do whatever we can to reduce those times… we can’t live forever in the shadow of the early 90’s FDIV bug, we need to move on. Our competition is moving much faster than we are".

Competition pressure could make a company's new product worse than (in this case, less stable than) their previous products, e.x. Samsung phone explosion. I still remembered the story was Samsung wanting to release their phone ahead of iPhone and I would imagine the testing went through a similar stressful time as Intel.

Of course not all cases of taking such risks would lead to disasters - just imagine Intel rushes on releasing new chips ahead of competition and 99 out of 100 times it ended up performing well. But a unique character in Intel's case is these bugs, unlike a faulty battery design, are accumulative and additive to future product development, which means a few small wins in catching up with your competitor could also lead to massive failures in some next major battle.

Now imagine Intel's competitors are going through the exact same scenario. One possible outcome is both Intel and its competitors' products become less stable and more buggy over time, and until everyone's stuff seems to be broken they probably never have time to fix them.

Added another source to assist discussion for more context: "https://www.hollywoodreporter.com/thr-esq/spotify-hit-16-bil...

And Quote :

"The Settlement Agreement is procedurally and substantively unfair to Settlement Class Members because it prevents meaningful participation by rights holders and offers them an unfair dollar amount in light of Spotify’s ongoing, willful copyright infringement of their works," ...

"In reaction, Spotify has been questioning whether Wixen has really been authorized by its clients (including Andrew Bird, Kenny Rogers, and Jim Morrison's heirs) to take its aggressive actions. Songwriters have administrative agreements with Wixen allowing the publisher to negotiate licensing deals, but Spotify has pointed out these agreements are silent about litigation."

Response doesn't make sense at all

"This title is from 2007, so we are unlikely to devote any valuable engineering resources to this issue, which is most likely caused by outdated API modules"

Doesn't make sense #1: Reason to not fix because it is from 2007, what about counter strike?

Doesn't make sense #2: Fixing the issue implies devoting "valuable" engineering resource, does it mean they don't have enough engineers or such issue is extremely difficult?

Doesn't make sense #3: "outdated API modules" was that likely from an educated guess and what does it mean even.

It sounds really foolish for a reply of this calibre to a community of supporters. I love AMD for what they're doing technically but I hope someone can at least promise to fix it soon before this hits the larger community.

I don't understand the purpose of publishing small technical pieces, put it there, and highly likely forget about it later and never come back to make anything really meaningful out of it.

This is at the best Google's "user/product study" to figure out if a product concept is worth developing further, but if you don't have a smart plot and meaningful conversation generation in the comic-gen-app, you won't be able to understand the actual reception of such a concept and risk killing a good idea pre-maturely.

But hell, why would theverge cover apps for the purpose of user study testing from Google.

It should be very doable and we do have a self hosted solution at https://photoeditor.polarr.co , which nobody uses as it is buried in terms of SEO (site like this https://www.pizap.com/pizap has much higher page rank). We'd love to keep the app on the web store as the SEO and amount of reviews we built up made us really stand out of the crowd.

This is really going to hurt Windows users who do not want to use .exe or Windows 10 apps, which are both very heavy and less secure when it comes to https or remote content loading compared to Chrome Apps (Windows UWP in this case).

Disclaimer: It's interesting to see that our app (Polarr) is in the screenshot (link: https://chrome.google.com/webstore/detail/polarr-photo-edito...)

Although ChromeOS is the smallest platform we currently support, we spent a lot of energy on it and it is so far, still the highest rated photo chrome app.

To put things in perspective. Between September to December, our Chrome App user breakdown are

Windows: 48.05% ChromeOS: 43.38% macOS: 7.11% Amount of users who used the app during this time frame: 326,459 (28.92% new).

Between Jun and September, our user breakdown are

Windows: 56.34% ChromeOS: 34.84% macOS: 7.25% Amount of users who used the app during this time frame: 280,013 (28.52% new).

The problem in here is more or less awareness of sharing or how to prevent accidental sharing of face data. I can see sharing to some photo editing app in anonymized fashion being useful for certain things and better than say, using a flat photo from Instagram or Facebook, but sharing faces on social media is a lot more intentional than a small and hidden agree to share with app button.

I think a better solution is when app is specifically requesting face data, there is a 2-3 seconds mandatory decision time with the default option to be off and prompt this decision in a different UI from the classic permission request dialogue. In this way the user knows the request is different And is given time to actually decide before accidentally tapping agree.

I'm thinking what this means to gas car makers.

Wave 1: "It's all about legacy and prestige, not speed/acceleration".

Wave 2: "The track handling isn't all that good, how can drivers take that corner at full power without losing traction".

Wave 3: "Alright, we will go electric too".

According to another source from BBC, the final bid for the work was $400m, with fees bringing the full price up to $450.3m. http://www.bbc.com/news/entertainment-arts-42000696

What's interesting from the BBC article is the image "The painting has been cleaned and restored from the image on the left to the one on the right" doesn't even look like the same person?

Another anecdote - In 1958 it was sold at auction in London for $60. By then the painting was generally reckoned to be the work of a follower of Leonardo and not the work of Leonardo himself.

There are a few apps right now that runs offline A.I. (instead of in the cloud) using the new CoreML in iOS 11 from Apple, including general object detection as well. But NSFW is a special unique use case.

Guess when you are a third party app running A.I. across entire user album you can't really do cloud anymore.

Disclaimer, developer of Polarr Album+ here (recently launched [1])

[1] https://techcrunch.com/2017/11/06/album-organizes-photos-wit...