gmurphy

Yep, that was me.

The reasoning wasn't that we didn't like it and wanted to kill hidden features (as above, I loved it), but more that all things, even small or hidden ones, come with small and hidden cost, and cost is the end of large projects. It sounds a bit daft to many, but just having the code in the tree is a burden - tests run, translations need to be updated, and minor dependencies all over the place increase development costs persistently. For example, consider the staff who added the "audio indicator in tabs" functionality and how the presence of a separate tab system in the tree might cause them extra work, or how their work may inadvertently cause critical bugs in sidetabs that don't have an owner to fix it or care for it.

To keep Chrome nimble we attempt to cut and trim as much as we can, knowing that that annoys some. We don't always get this right, but it's better to aspire to do fewer things well than to be generally OK with crufty bits here and there.

I realize this explanation probably won't decrease the sadness, and you've probably heard something like it many times in the bug tracker, but if not, I hope it explains things a little.

This is a great use of Chromium!

Storytime! When we started designing Chrome back in 2006, and despite having spent the past year working on Google's Firefox team, my favorite browser was iRider, which was an IE shell with tree-style-tabs. It was too power-user for what we were going for, but some of the concepts (pinning) live on in Chrome today.

LEGO itself had such a service, but shut it down in 2012: http://ldd.lego.com/en-us/subpages/designbyme

A Googler posted on their blog that they were looking, so I emailed him and he referred me. At the time I had dropped out of an EE+CS degree, had four years of suit-wearing professional experience writing enterprise webapps in PHP and ColdFusion, was halfway through a part-time Masters Degree in Design, and had a bunch of small JavaScript/C++/Art side projects. They hired me as a SWE, am now in UX.

I won't pretend that I wasn't very lucky in timing, interviewers, etc, but you won't get anywhere if you don't roll the dice.

I thought the same thing once, but applied anyway on a whim. Have now been at Google for 11 years.

It can't hurt to try.

Chrome doesn't add settings for things when extensions can do the job because doing so adds to our support burden and slows further development - it's better to appeal to fewer people and have a simpler and easier/faster-to-maintain product than to appeal to more with higher maintenance complexity.

Try using Steam Streaming - it's some work to get it to mirror a desktop rather than a game, but because they assume high-bandwidth connections, it does 1080p at 60fps for me, with super low input latency.

My father used to teach microcontroller programming (after doing it for years at TI); he LOVES this game, wishes he had it when he was teaching students just a few years ago, and has recommended it to the professors who succeeded him.

We considered indicating this in Chrome, but the problem was reliability - because JS can execute when you click, we actually can't tell you what will happen next, and we don't like showing unreliable UI (extensions, however, are great for this sort of thing).

You should try Steam streaming - you can run your full and regular desktop in it (by using fake passthrough apps), and is how I often use Lightroom and Solidworks from my Macbook in bed.

Coming from VNC and Remote Desktop, it's remarkable how low the latency is. In many cases, because I can get a signifcantly higher framerate by remote rendering, the latency is better than if I was doing stuff locally.

For games, it's worked pretty well with everything you'd be happy playing with a gamepad (e.g. GTAV, but not CoD).

I don't believe the small mechanical difference in click activation actually makes a huge difference to uptake numbers, since the big leap is making the choice to have plugins not run by default, with the activation path making no noticeable difference since by then you've already filtered out most people, who just don't care about plugin activation. This feature is driven by the Chrome security/privacy teams, who would have reverted it by now if they saw a regression in usage.

Chrome moved to context click + menu because left click is clickjackable, so doesn't help very much with the security issues.

As a large amount of what I do at home is games and Lightroom, what I'd really like (as an incredibly niche product) is a pure terminal laptop that is just a fast wireless screen/keyboard/mouse for the beefy desktop elsewhere in the house.

Today you can kludge it together with a regular laptop and Steam Streaming (latency is impressive, but not perfect), or by using a low-latency/long-range wireless HDMI system like the Nyrius Aries Pro/Paralinx with a lap-sized monitor and wireless keyboard/trackpad.

The second method works (and also works with consoles), but is hampered by the same-room-only short ranges of wireless keyboard/mice, and is a mess of wires and USB batteries.

Right now there aren't any systems that do it all well, so you want to just pick the best tools for each job, then glue it together with a hub.

A surprising number of things have relatively IP interfaces, so in my house I scripted it all together by writing http://route.io/ - this lets me do things like control Sonos from my lightswitches, unlock my house when I walk up to it, or turn off everything with a single command.

Here's quick demonstration that you can easily detect 60fps and that the referenced picture is valid - grab a window (or even your mouse cursor) and wiggle it around your screen. Unless your computer is ancient, it was probably updated at 60hz. Yet it's not very smooth - you can clearly see multiple steps with gaps in between.

Now do the same on a 144hz monitor - it is significantly different, but still not as smooth as it could be.

Yep, z-wave locks have keypads - it's what I have (Yale Real Living), and is a great solution. The problem I was referring to was the new class of Bluetooth-based locks that rely on your phone, but don't have keypads (Lockitron, Kwikset).

The lock does have some lag in it, but I just try to keep it awake by pinging it, and rely on early detection of the beacon to get the request in early. Definitely not a great solution though, and I'd love to find something faster.

RaZberry. You can see the (hacked-together) code in: https://github.com/glenmurphy/route/blob/master/examples/cam...

Smart locks are awesome, I have a z-wave lock in my house, hooked up to a home automation system I wrote so that my front door unlocks whenever I walk up to it. I love that I've reduced the "wallet/phone/keys" patdown to "wallet/phone".

Unfortunately, many of the smart locks that use your phone as an unlock mechanism don't have a good solution for what to do when your phone battery runs out. Usually the answer is "use a key", but the new problem is that once enough of your locks are smart, you stop carrying keys.

In the home, keypads work well (and they can easily be set up for guests, etc), but that seems prone to memory error in hotel situations.

I score 33/36 on this test and very highly on others like it, yet I consider myself pretty rubbish at reading people's emotions. Still pictures are very different to real life - here, I can rely on pure technical understanding trained by years of being obsessed with art (looking for eye creases, angles of eyebrows, and imitating the expression in my head), without the distraction of people saying things in direct conflict with the range of things briefly written on their face.

I suppose it's a works-in-a-test-setting trainable sort of thing. Thanks, comics!

It's not so much the getting out, as the not having to do anything - you get 90 seconds of HN time!

  > I can run it on my laptop, my other laptop, as well
  > as my desktop machine that just gathers dust these days.

I'm not sure how you're doing this - a killer problem for me with CC is that there's a two-machine limit for product activations. I used to own two Photoshop licences so I could run it on three machines, but you can't get multiple Creative Cloud subs under one Adobe account.

CSS lets you color links based on whether the user has visited them or not.

Most of the Asteroids in the game are links - they use those same CSS rules to color themselves based on whether you've visited them or not.

Script on the page can't tell what color the asteroids are, but it can tell what asteroids you've clicked on. By making the visited-link color the same as the background color of the webpage, we get the user to give away the color of the links, because they'd only be clicking on the links that they can see - the ones with a color different to the background.

You could use Google Alerts (or equivalent), which gets you the name and frequency, then embed the reason in the search string in a way that won't match anything.

e.g: Tammy Glotfelty -"reason: keira science experiment"

The visibility of the shield was designed to be proportional to the number of people who care/should care about it.

Gwern is male; not only is Gwern a boy's name, making it a reasonable assumption, but he publicly declares himself as male.

For example:

https://plus.google.com/103530621949492999968/about

http://www.last.fm/user/gwern

I've struggled with sleep issues since I was a small child; even after doing all the reasonable exercise/caffeine/no-bright-lights things people recommended, my brain sometimes just wouldn't shut down. After some introspection, I came to realize that my body seemed be on a 25 hour day, and it just wanted to stay awake an hour later each time.

Then I read gwern's article on melatonin[1], and it changed everything. I don't get the shorter sleep he describes, but I take it once or twice a week to give me what feels like complete control over my sleep cycle. In addition the alertness/grumpiness benefits, it is truly great just being able to go to sleep and wake up at the same time as my wife (who seems to be on a 23-hour day).

[1] http://www.gwern.net/Melatonin

Didn't appear to work for me, as I'd already (apparently) claimed a $1.25 free credit. D'oh.

I did not have a degree when Google hired me as an engineer; the engineer who referred me was also a drop-out; my manager, now a VP, also dropped out of college; and I have a co-worker who dropped out of high school.

It's not the normal path, and you need to fill that blank in your resume with interesting personal projects and work, but Google is perfectly willing to hire designers and engineers without degrees. We do it all the time.

My problem is that I already have (many) of those relationships, but because of these difficulty issues, I'm stuck in those relationships.

I'd really love it if someone could make it as as easy to buy design help as it is to buy stock photography.

There is an astronomical difference between the effort required to buy tens of thousands of dollars worth of imagery, and the effort required to buy tens of thousands of dollars worth of someone's time. The former is a matter of a cart and credit card, the latter requires lengthy emails, contracts, legal approvals, timesheets, and all other sorts of mumbo jumbo and for most projects, is enough work that I often just don't bother.

Though those roadblocks are there to protect both parties, I suspect that there's an artificially large delta between cost and risk.