parm289

Perhaps a bit off topic, but seeing the breadth of general discussion about online CS masters programs in this thread, figured I'd ask the HN crowd:

How is a program like the Penn MCIT Online degree[0] viewed by engineering and product hiring teams in industry? I am looking at transitioning from venture capital investing to SWE (and potentially product, given my business background) and this seems like a good option to facilitate that change - education in CS fundamentals (vs. a bootcamp) but still a reasonably short/practical program (10 courses).

Curious if there are engineers/PMs here that have gone through MCIT or similar to pivot into software from another field? Or, if any hiring managers here have hired graduates from these online degree programs and have insight/advice?

[0] https://online.seas.upenn.edu/degrees/mcit-online/

Yeah, the disk space issue is what I'm concerned with. I'd prefer not to use local disk space to store a copy of my emails but I haven't been able to find a way to disable local storage in Mail.app.

This is great in terms of UI. Does anyone know if you can use Mail.app with Gmail and keep all emails on the server? I'm currently using Mimesteram which has a great UI and doesn't keep a local copy of all emails. But, I would rather use the native mail app if possible.

Got it, that makes sense.

What about a firm like Knight?

Why aren’t these firms insuring against this risk? Seems like any other risk that can be managed - pay premiums and so that a third party would fund any erroneous losses.

Was going to post this - have users submit the track data! That also might make it easier to expand to other stations where you aren't as familiar with the track layout.

Interesting, thanks. I had in mind a system that would basically regex match a string and add an entry to the ledger (similar, but it would probably be a bit more flexible than the built in system).

Separately, I saw you have also worked on a similar system for food logging. That got me thinking...what else could you track with a plain text file in this format? Fitness/workout info came to mind, as well as rewards points, sleep tracking, any others? Would be cool to have a single system to track all these things, which could be linked up to APIs like Fitbit, myfitnesspal, etc.

Have you ever thought about using a natural language interface with ledger? For example, the Mac/iOS calendar app Fantastical lets you create an event by typing or dictating a sentence like "lunch with mike at 2 on Tuesday at Wendy's", and the app will parse the input to create an event with the right parameters.

I ask after reading your vacation write-up. Seems like the input side could have been a lot easier if you had a speech or text NLP interface to ledger. "$10 entertainment expense paid from MasterCard today," for example. That could be the holy grail for people who are not inclined to keep a detailed ledger, or for situations where input is difficult. (Theoretically you could email/iMessage/SMS/slack that string and have it picked up by your ledger file.)

Thanks for the insight. Are there any particular social media sources you like to use? I find that Twitter and reddit are the most real-time, but reddit can often carry its own biases as well. Twitter seems like it could be the best social media news source if field reporters released news directly through their streams (ie not tweeting as a representative of FOX or NBC). Essentially the IRC chat you mentioned above, but public-facing.

Your comment was enlightening and compels me to ask: where can the public get access to live field reports such as the one you mention? If news companies transform these reports into faux-news, I'd like to bypass them directly and read/watch directly from the source. Are there any publicly-accessible resources that provide access to field reports or "real" news?

I posted this comment on another taxi medallion article, but figure it's hopefully more relevant here:

Just a little background on medallion finance:

A few summers ago, I worked in commercial credit and we did a financing for a "taxi mogul." He was replacing several cars in his fleet, and wanted to take out term loans for the full purchase price of the cars (approx. $30k each, IIRC).

The loans would be secured by cash flow, but the business also posted medallions as collateral. Each loan was attributed to the vehicle purchased with the proceeds and secured by that vehicle's medallion.

It's hard to value something like a taxi medallion. Medallions aren't liquid -- they are usually sold in very low volumes at auctions controlled by the TLC (in NYC). Additionally, the TLC limits the number of outstanding medallions. In practice, most of the medallions are concentrated in the hands of "taxi moguls" who started taxi businesses in the early 20th century when medallions were cheap (think $30k). The best approximation for value we had was the prices commanded by medallions at auction. When I was doing diligence on this deal a few years ago (before Uber), prices were accepted as $1.1mm per medallion. (In reality, that was at best the value of the "marginal" medallion sold; i.e. you could probably not put 10 medallions up for auction at $11mm.)

That means that a $30k term loan would have a loan-to-value of about 3% -- a dream for a bank, assuming the medallions can be seized and sold at market value upon default. That also meant that the loans would be approved almost regardless of the integrity of cash flow. Those characteristics allowed the more cunning taxi moguls to borrow a lot of money against their medallions, securing low rates due to the strength of their collateral posting, and lend the money out at higher rates to earn arbitrage.

There was a good amount of discussion about the medallion bubble -- all it would take is a significant increase in the number of medallions authorized by the TLC or a few failed auctions, and a medallion sold at a large haircut, for the value of all medallions to plummet. Granted, the drop in value might not trip loan covenants, but it would significantly erode the balance sheets of these businesses. At the time, we didn't expect that there would be an external force that would hurt medallion values.

Honestly, skyrocketing medallion prices made it clear that additional ride capacity was needed/demanded. The interesting fact is that the medallion market wasn't disrupted by the issuance of additional medallions, but rather a drop in the demand for yellow cab rides -- a scenario that taxi moguls likely hadn't planned for.

All in all, an interesting asset class that most people aren't aware of -- those 4-letter signs on taxis hold no meaning to riders, and almost nobody on the street would guess that they represent assets worth over one million dollars.

Just a little background on medallion finance:

A few summers ago, I worked in commercial credit and we did a financing for a "taxi mogul." He was replacing several cars in his fleet, and wanted to take out term loans for the full purchase price of the cars (approx. $30k each, IIRC).

The loans would be secured by cash flow, but the business also posted medallions as collateral. Each loan was attributed to the vehicle purchased with the proceeds and secured by that vehicle's medallion.

It's hard to value something like a taxi medallion. Medallions aren't liquid -- they are usually sold in very low volumes at auctions controlled by the TLC (in NYC). Additionally, the TLC limits the number of outstanding medallions. In practice, most of the medallions are concentrated in the hands of "taxi moguls" who started taxi businesses in the early 20th century when medallions were cheap (think $30k). The best approximation for value we had was the prices commanded by medallions at auction. When I was doing diligence on this deal a few years ago (before Uber), prices were accepted as $1.1mm per medallion. (In reality, that was at best the value of the "marginal" medallion sold; i.e. you could probably not put 10 medallions up for auction at $11mm.)

That means that a $30k term loan would have a loan-to-value of about 3% -- a dream for a bank, assuming the medallions can be seized and sold at market value upon default. That also meant that the loans would be approved almost regardless of the integrity of cash flow. Those characteristics allowed the more cunning taxi moguls to borrow a lot of money against their medallions, securing low rates due to the strength of their collateral posting, and lend the money out at higher rates to earn arbitrage.

There was a good amount of discussion about the medallion bubble -- all it would take is a significant increase in the number of medallions authorized by the TLC or a few failed auctions, and a medallion sold at a large haircut, for the value of all medallions to plummet. Granted, the drop in value might not trip loan covenants, but it would significantly erode the balance sheets of these businesses. At the time, we didn't expect that there would be an external force that would hurt medallion values.

Honestly, skyrocketing medallion prices made it clear that additional ride capacity was needed/demanded. The interesting fact is that the medallion market wasn't disrupted by the issuance of additional medallions, but rather a drop in the demand for yellow cab rides -- a scenario that taxi moguls likely hadn't planned for.

All in all, an interesting asset class that most people aren't aware of -- those 4-letter signs on taxis hold no meaning to riders, and almost nobody on the street would guess that they represent assets worth over one million dollars.

Why don't we launch the nuclear waste into deep space? Seems like that would avoid the burial problem described here, and since space is mostly, well, empty space, wouldn't inflict much harm to other bodies.

Great idea. To start, why not have the app find episodes that match the user's commute time? For other episodes, an app like Overcast could first apply SmartSpeed and then speed up the entire episode (ie 1.1x) to match the commute time. I'm finding myself looking for good (financial) news podcasts that are < 20 mins per episode.

Awesome, thanks.

For those of you on iOS looking for a polished transit app with support for multiple cities, I highly recommend Transit.[0] It allows you to quickly map trips as well as easily save data for offline access.

[0] http://thetransitapp.com

Any plans to open source the code?

I agree. It seems that Uber could be for transportation what Seamless is for food (at least in midtown Manhattan). Currently, banks etc. either hire company car services to take employees home late at night or reimburse late-night taxi expenses post-hoc. It would be great to simply charge these late-night rides to the company Uber just as we order dinner on the company Seamless.

Is your system open source? Would love to take a look at how it works.

Me too. I think there could be a sizable market for a tablet or app that does notetaking and handwriting really well. The iPad seems limited by its touchscreen tech/software, which seems to be optimized for fingers and not fine-tipped digital pens.

Which app and pen do you use for handwriting notes on your iPad? I've tried a few but I haven't liked any of the input interfaces. Writing in a small area of the screen is cramped, but when the whole screen is available to write (e.g. Paper, the drawing app) I find my arm constantly getting in the way.

You might be interested in the axlsx ruby gem: https://github.com/randym/axlsx

Haven't used it, but it allows the creation of pretty complex spreadsheets from ruby code. Not sure if it is able to read from spreadsheets, but it could be one part of the service layer.

You mention fees and AML/anti-terrorism regulation...it seems that the major impediment to cross-border wire transfers and ACH is regulation, not a lack of infrastructural capability. Assuming most countries have the same goals with regard to regulation, it seems feasible to develop some overarching regulatory system which would permit international transfer. I wonder if one can transfer money easily within the EU -- that could serve as a next step or a guide to implementation.

Just brainstorming here. I really do suspect that most of the benefits of cryptocurrency are due to the way it interacts with technology, which can be imitated through just a bit of innovation in the banking industry. (Imagine banks allowing customers to script transactions, like Bitcoin, or providing direct APIs to bill-pay and ACH transactions.)

For every discussion about Bitcoin, I wonder if improving the ACH system and/or writing services on top of it would be a logical next step to make the existing banking system more user-friendly.

For example, Venmo[0] is simply an app layer over the ACH system. Sending someone money is as easy as sending a text on iOS, or an email on your computer. Payouts to your bank account can be same-day but usually take 1 business day.

Other than the pseudonymity (which is admittedly one of Bitcoin's stated value propositions), doesn't Venmo capture most of Bitcoin's advantages? What's preventing someone from using the Venmo API to, for example, write a reddit micro-tipping bot? Is it that an app having ACH access to people's bank accounts is scary?

Those aren't rhetorical questions -- I'm genuinely interested in knowing what advantages cryptocurrencies have for the layperson compared to an enhanced ACH system where security is already taken care of.

(I do use Venmo, but am not affiliated with the company otherwise.)

[0] http://venmo.com

You raise a good point about mood tracking. I've found that iTunes's genius shuffle does a decent job of shuffling a single genre of music, so you don't generally get harsh interruptions.

The holy grail is most definitely a library (or app) where you click "shuffle all" and enjoy every song it plays. Songza, genius shuffle, 8tracks etc. all try to tackle this, but I still think we're far from that point.

Spot on. This would be great as a way to learn new frameworks/libraries as well.

At the high school level, I think your analysis is generally correct. Most students who aren't already interested in computer science will have the same attitude toward a programming course as they have toward foreign language courses.

However, in college, the paradigm is different. At my university, programming is taught from two distinct perspectives: theoretical (as in the computer science major) and practical. I'm in the business school, and we have courses that teach VBA, SQL, Python, and Matlab. These courses focus on using these languages as tools to solve problems and gain competitive advantages through optimization, efficiency, and data analysis (think of Steve Jobs' bicycle analogy). For example, one course teaches Python and Matlab in the context of text analytics -- students use Python's text handling capabilities to parse text and collect data, then use Matlab to visualize the data and detect trends. VBA and SQL are taught together as tools which increase productivity in business, where scripting Excel can save countless man-hours and being familiar with databases can make interfacing with data analysts and IT people easier. These courses are taught with a distinct "hacker" or "scripter" mindset.

Students inevitably have difficulty with syntax, logic, etc but get past that with practice over the course of the semester. Programming is perceived by students as a problem-solving tool, like a TI-84 calculator helps them solve math problems and graph functions. They will not be creating enterprise applications or studying theory, but learning to script measurably enhances productivity -- and that's where the value is.

(Anecdotally, I'm not a programmer but am familiar with Ruby as well as the languages listed above. While doing an academic research project this semester, knowing Ruby made it trivial to quickly organize and manage a trove of 10,259 text documents. Without programming knowledge, that task could have taken quite a few hours.)

Just curious -- what analytics would you be interested in seeing?

I actually started hacking on this idea a few weeks ago. A personal API could have tons of uses. Many others have commented on the benefits of aggregation, but this could allow users to control and selectively sell/provide information to companies, apps, and advertisers. Applications could be built which utilize the personal APIs of friends. In Naveen's example, my app could compare my fitness level with my friends, recommend new restaurants based on ones my friends actually went to, track progress on goals, etc.

What I'm trying to say is that with a personal API platform, apps simply become a front end to standardized personal data. Switching costs would be reduced significantly, which would encourage a lot more innovation.

Here's a spec I've written out for my work on my own personal API. It's not done yet, and the code isn't open sourced yet, but hopefully it will inspire others to hack on the idea.

https://dl.dropboxusercontent.com/u/570656/spec_public.md

I agree that the technology is impressive -- after all, serving a billion users boggles the mind.

I was getting at the faults of the platform, though. It seems that people are satisfied with Facebook's groups, chat, and events features, but not the news feed/status updates portion of the site. I was wondering aloud if a social network like Path, which seeks to encourage more honest sharing as opposed to a greater volume of sharing, offered some insight into why people are becoming dissatisfied with Facebook.