glimcat

Software development jobs tend not to care what you majored in if you can demonstrate competency and experience. Your average CS senior is a VERY junior programmer.

Also "math / engineering" is usually good enough to bypass the HR "must have degree" wall at companies where that pertains - at least to the point that you have the same chance anyone does to pitch them on interviewing you.

Things you can do that beat "switch majors" in terms of improving employment prospects:

    * Summer internships
    * Part-time gig
    * Write code, publish it on Github
    * Write a tutorial article
    * Create and market a software product

You also have to ask yourself, what are your goals for this? Are you looking for a comfortable W-2, or to spend some time working at startups, or to start a business of your own within a few years of graduating?

I'm seeing a range from snappy to several seconds (favoring snappy), so I'm going to guess "better caching."

Taking several seconds to load was common enough before that I would load the index & queue up all the articles I was interested in at once in new tabs, then grab my coffee. I may be able to just browse now, if this keeps up.

Things like Twitter often start as a "wouldn't it be neat if."

Then actually taking the trouble to prototype it, show it to people, and find out if it has lasting value to any group of people you care about. And then you have to improve it to better fit those peoples' needs, and marketing it instead of taking a "okay, now come and get it" attitude.

Each of those steps is a huge obstacle which most people won't even attempt.

We see the finished product and think "who knew we needed that." But it's exceptionally rare to jump straight from a golden idea to a finished product. That's a marketing narrative after the fact.

Even "finding an idea" is fairly rare, and its going to waste a lot of your time if you expect it to be the first step to building a business. Find an audience, learn about them, figure out what their goals are and what choke points are keeping them from attaining them.

Then help them.

While you may be the "visible culprit" - I'd hardly say you're at fault. The action you're attempting is not a "send email" action, even if the API provider has decided that it should send email notifications.

The implementation you'd probably want to see is for the API's email notifications to be batched if more than small-n trigger within a short period of time. Then the end user gets a notification that "1700 updates have occured" instead of getting 1700 emails.

The API should also be set up with exceptional event detection. Spikes which are several orders of magnitude above normal should get paused in the task queue and flagged for immediate manual review. Users do things you don't expect.

Is this separate from the line item about support (best effort / no SLA purchased)?

And if it's separate, in what sense does enterprise purchasing use "maintenance"? Software updates?

As long as we're playing the "Falsehoods Programmers Believe About Names" game again, here's the relevant patio11 article:

http://www.kalzumeus.com/2010/06/17/falsehoods-programmers-b...

If you try to validate names, or if you don't safely escape names along with your other user-input strings, you're gonna have a bad time.

"Software Businesses for MMO Addicts: The Dragons Wore Suits"

There are also numerous similarities between progression raiding and marketing. Replace phases with funnel stages, try different strategies until you figure out which one results in fewer wipes...

It's a developer tool. I bill more than that per hour. Lots of people bill more than that per hour. If it saves you time or stress, it's easily worth $70.

Article: "What The Rails Security Issue Means For Your Startup"

http://www.kalzumeus.com/2013/01/31/what-the-rails-security-...

"they can kill it if only they had a"

Nine times out of ten, if you find yourself thinking in this pattern, you're making excuses not to start.

Find a way to get started anyway. Do something. Make a mockup out of pieces of paper. Find and talk to potential customers.

There are things you can do now, and they're usually things you really should do before bringing on a developer, or taking investment, or whatever you're blocking yourself with.

If you get to "if only" and stop, you may be in love with the idea of being a successful startup founder, but you're finding excuses to avoid what it means to actually become a startup founder.

Or a jazz musician. Novelist. Whatever.

This approach is using the bounding box size and longest dimension to choose a camera distance and angle. It works better than static pose and it's computationally cheap.

However, it can oversimplify for objects which are significantly asymmetrical. A good place to go from here is statistical measures of object recognizability:

Betke & Makris, Information-Conserving Object Recognition.

http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.44.4...

For 3D models, you can use hierarchical space carving to simplify the mesh first. You don't need great resolution, you just need better resolution than a rectangular bounding box.

(Note: PCA can be used for recognizability, but is not ideal. Highish complexity & overly general.)

Thanks. I emailed you a few glitches and such that I encountered.

This looks really, really nice so far, particularly for its initial release. There are a few things like discovery and analytics which obviously aren't in at this stage...but other than that, it looks much nicer than most of the similar sites that I've seen.

http://hoverboard.io/grayj

The decisive condition here is often "can a problem with the software create a halting error in someone's daily workflow."

I have training in designing such circuits and I'd still use an opto-isolator.

This is not manufacturing. It's a one-off project. It's not a highly refined design. You don't have to worry about finessing a marginally lower build cost because of what it will mean in 10k unit quantities. Just use the opto-isolator.

No it doesn't.

The only way they display this information is if you (a) add it to your profile, and (b) make that information public.

If you use LinkedIn to tell people when your birthday is - and I have no idea why you would, given its intended use case - then that's your doing, not theirs.

Use this link to edit your profile:

https://www.linkedin.com/profile/edit?trk=tab_pro

In the "Additional Info" section, either set Birthday back to "Month..." and "Day..." or click on the icon next to it to reduce its visibility to only your network or connections. Then click the "Save" button.

P.S.: You can always lie about your birthdate.

The display clause prohibits advertising your party / movement / whatever in the Supreme Court building. It's a specific exemption intended to avoid influencing the impartiality of the group responsible for making determinations about the Constitution, which strikes me as not such a dumb idea.

It is unlawful to parade, stand, or move in processions or assemblages in the Supreme Court Building or grounds, or to display in the Building and grounds a flag, banner, or device designed or adapted to bring into public notice a party, organization, or movement.

There are some genuinely stupid acts of government out there like free speech zones and all the security theatre, or attacking photographers. This is not one of them.

Want something to get mad at the government about today? Congress is currently playing an $85 billion dollar game of chicken.

http://www.cbsnews.com/8301-250_162-57572068/sequester-its-j...

Beating $15k is certainly possible. If it's hard in your area, there's probably work that can be done remotely paying 10x. The catch is that you'd have to build up the reputation and the network of satisfied clients needed.

I know a few very experienced engineers doing work in robotics. There's a fairly high variance in salary. Of those, the one I'm fairly certain clears >$200k does so largely through things like consulting and appearing as an expert in patent disputes. His day job is maybe half that.

You can make lots of money and do what you love, but sometimes the thing that pays the bills and the thing you enjoy aren't the same. You'll get better results if you work out an optimal solution for yourself, versus being wholly dependent on an employer to give you a single monolithic package - but it will also take much more effort, and likely many failures.

Lots of the ones doing work in industry are in areas like calibration and maintenance, vs. design and research. Part of the issue here is working out the specifics of what "decent paying" and "in robotics" mean to you.

The problem with "learn x" from a motivational standpoint is that it is extremely open-ended. Say I want to learn more JavaScript. How do I know when I've met my goal? If the goal is indefinite, how do I measure progress? This is very demotivating.

Instead, try to find something completable which requires the skill you want to get better with. Build something, even if it's mostly frivolous. You need a destination, a path from here to there, and the ability to measure progress along the path.

So, the really cool part is this:

"At Watsi, 100% of your donations directly fund medical treatments. Watsi.org is separately funded. They pay all their operational costs from their own funding, and none from your donations. They even eat the credit card processing fees."

This works great with Watsi's crowdfunding-style model. It probably wouldn't help with things which need legs on the ground full-time, but it would be interesting to see more nonprofits looking at an "out of channel" donor model.

P.S. If you haven't clicked through to read PG's announcement and check out Watsi, go do so. Really. This is something YC is going to look back on and be proud of being part of 20 years from now.

https://watsi.org/faq

Reusing existing tweets is a common practice for Twitter bots.

Research-grade bots typically combine this with some logic like taking tweets from sources external to your current network graph, or trying to keep the bot's persona reasonably self-consistent. E.g. see the RealBoy project:

http://ca.olin.edu/2008/realboy/

Spam bots generally have a much lower bar. The majority seem to just use a follow-back paradigm, i.e. they put some spam message in the user profile, follow people at random or according to some search parameters, then unfollow a couple days later if you don't follow back - this last, since Twitter decides you might be a spammer if you let your following/followers ratio get too high.

I've been doing some work with Twitter followers recently, to try and estimate real followers versus spam accounts. So far, I'm seeing something like 15% real, using relatively simplistic spam detection (likely high error).

I'm mainly looking at commercial accounts with five-digit follower counts. I also haven't run enough yet to know if the result is more up to my detection routine, or if I'm choosing accounts that bought followers at some point, or what. So don't take it as a general rule (yet).

Surveying preference isn't a valid way of determining product viability, or even of comparing how well two products well sell.

This is not to say that you shouldn't build it. Building stuff for the sake of building stuff is a good thing.

This is also not to say that you can't make a business around product validation. There are multiple successful businesses already in the area of product validation.

I just wouldn't recommend the "Hot or Not" approach, because it's bad and generates bad data. Think who else is doing this and get an idea for what approaches there are to the problem - for example, Kickstarter.

If you go here, you should be able to "reset" the admin account via the domain ownership verification process. This involves either uploading a file to the web server or altering its DNS records.

    https://www.google.com/a/cpanel/yourdomain.com/VerifyAdminAccountPasswordReset

Replace the "yourdomain.com" part with the appropriate string.

You know, I'm all for that - up until we invent Youtube and it turns up in search results for the person's name more or less forever.

I don't believe that the citation list is copyrightable ("copyright does not protect facts" [1]). Ergo, anyone with access to a papers database could contribute, or you could port it over from other citation graphs.

Many papers databases will give you a formatted list of citations, which would be easier than trying to extract it from a set of arbitrarily formatted PDFs.

[1] http://www.copyright.gov/help/faq/faq-protect.html

Just from an efficiency standpoint, you don't need other levers for new grads. A reasonable engineering salary will already be highly motivational for someone who is coming off a student's budget, probably with loans just kicking in.

Some percentage of them are going to flake. That happens when you take people who have spent their lives in an environment with one eval loop and place them in a new environment with differing expectations.

Of those who would flake, some of them can be made into great employees. But a bigger carrot is almost never effective at accomplishing this. The real need is generally along the lines of "effective mentorship" - which is far harder to implement than a revision to your employee benefits plan.

I know a guy who's working on this. In addition to the chores, they're working on some stuff for motivating kids to study.

Allowance as a service + motivational gamification.

http://empower.me/how-it-works/

Many instances of this type are properly unit tests, e.g. assert brute_method(foo) == optimized_method(foo) for foo in cases.

You can increase this further by abstracting away the "code monkey" aspect.

1. Specialize & productize your output, then offer fixed pricing. Customer gets a predictable result for a predictable cost, you produce something you can estimate to good accuracy while earning a higher effective rate.

2. Retarget your deliverables on business value instead of a technical spec sheet, then charge accordingly.

Note that both of these involve business development strategies, not improved technical skills.

The general premise is that there is an arbitrary, recoverable signal which has been convolved with the data signal. The generating function for this signal is a matter of record if you have access to the utility logs, but difficult to obtain otherwise without being physically at the time and place where the effect occurred.

I strongly suspect that it can be compromised under both of the following conditions:

1. You take recording A, then you take recording B at a time and place which you want to assert that recording A took place at. You recover the thumbprint from recording B, suppress the thumbprint in recording A, then apply the thumbprint from B to A.

This is not a trivial process, but you only really need a plausibly consistent result. A reasonably basic understanding of signal processing theory, a copy of MATLAB, and many pots of coffee should do the job. Then, you could automate most of it for the next guy.

2. You take many recordings at a series of locations of interest, while taking data about the power grid from nearby locations and from distribution nodes. You then attempt to predict the signal at a location from the characteristics of the surrounding area.

This is almost certainly possible, as generalization from distribution logs to the local effect is what makes their fingerprinting technique possible in the first place. It is not a trivial undertaking, and it's questionable how well it would be generalizable. But at the same time, it's largely a question of if you want the data badly enough to do the legwork, and whether you have a reasonably functional understanding of machine learning.

It doesn't help that user research is constantly getting conflated with how to do drop shadows and gradients.