Any talk of transferring anything cultural from current software development practice to somewhere else important... makes me nervous, by default.

Right now, we have huge problems in software development practice, with both design&implementation quality, and ethics.

You don't want your medical device or bridge developed in any way like the majority of software right now. You don't want your ER nurse to be smug about how smart they are, while they do shoddy work with fad tools and halfwitted cargo cult processes. You don't want your doctor or lawyer selling you out to 'analytics', for starters, like almost every dotcom startup does.

As startups try to bring the greatness of Webrogramming to other areas, are they going to suddenly say, "Oh, but this is actually an important area, unlike the information infrastructure of humanity, so this one we'll do with more responsibility than obviously we have been."

Now, of course, we’re using software to scale everything and data science to measure and optimize everything, but the core engine is a really great school and instructional design we created from the ground up, not code.

For example when a doctor asks you if you travelled recently they aren’t going to know which contagious diseases are local to where you visit. So if you tell them you went to Belize they’ll open https://wwwnc.cdc.gov/travel/destinations/traveler/none/beli... and see if anything there matches your symptoms.

For surgeons it’s mostly practice, practice and practice on how to cut you up and stitch you back together in the cleanest and fastest way possible.

That's what all of those medical texts in hospital offices are for. Instead of Google, pull text off the shelf with some notes or annotations.

Unfortunately, this may already be the case with people who graduate from medical assistant and physician assistant schools. From what I hear from doctors is that the quality of these graduates are sub-par. They are usually confident in their medical knowledge while repeatedly failing to show the conscientiousness needed to practice medicine without harm.

Frankly, this comment just sounds like you have an axe to grind with PAs. (Hence conflating MAs with PAs.)

PA is a Master's degree. They diagnose, treat, and prescribe. Unlike MDs, who are board-certified independent practitioners, PAs are not trained to work independently but as part of a team. In some cases that means they function as physician extenders, freeing up physicians to focus on more complicated cases. In other areas they focus on tasks delegated to them, for example central lines or other procedures, wound care. Some are used as First Assists in surgery, others are used for pre and postsurgical care, which helps surgeons do more surgeries. In any case, while the roles for physicians and PAs are much more similar, the condescension remains in your comment.

Punching down on MAs is inconceivably poor form. Punching down on PAs is still bad. Conflating the two gives away your game. You suggest harm... have any evidence to back this up?

Hint: you won't find any. There aren't a ton of studies, and quality could be better, but mid-levels (NPs and PAs) who stay within their roles have comparable outcomes to physician colleagues. How can this be, given the difference in training time? Easy. The roles are different. The subset of pts managed by PAs/NPs is different than MDs, because professionals consult and transfer care when appropriate. It's no different than a family practitioner or hospitalist consulting specialists. Nobody does everything, healthcare is a team effort.

Trash talking other professions, especially without evidence, is unprofessional.

So do you think this person is lying or not? Did these conversations in fact take place? That's the more basic question to ask first.

That said, the "only pay us back if you get a good job" commitment seems like lip service as long as the cutoff is $50K (for software dev at least). Lambda school's team is super vocal on Twitter and regularly shares stories of students who walk away making 3-4x what they made previously...if the program genuinely believes it can consistently graduate students who can earn $80-$150K in the market, then it should align its business model to that.

The median salary for a software engineer in the US is $100K. The current Lambda promise is "you only have to pay us back if you earn half the nationwide median salary in the 5 years following our program"...this does not strike me as a very bold commitment to the quality of the program.

To be very clear: I've seen successes of Lambda students and I believe that they have created a model that graduates students who are competitive in this job market. So, I'm not saying that Lambda is bad. On the contrary: I'm saying Lambda is excellent and they should be bolstered by that excellence to take a stance with a less conservative income share agreement.

From their site:

> Instead of paying tuition, students can agree to pay a percentage of their income after they're employed, and only if they're making more than $50k per year.

Certainly, in San Francisco, $50k isn’t a win. Luckily our median salary there is well into the six figures.

Lambda School’s income share agreement is also (to my knowledge) the only in the industry that is industry-specific. If you end up as a bartender making $60k we make $0.

Most 3 month schools are 15% for 3 yrs with a $40k min. Lambda School is 9 months, 17% for 2 yrs with a $50k min and a cap.

So our agreement is actually very aggressive as is, and will become more aggressive over time.

https://open.buffer.com/transparent-salaries/

Btw, love what you're doing. It might have helped me back when I was in a rural area. We didn't have Internet, though. So, would've been whatever I could do at libraries during narrow windows of computer use. These days it's McDonalds' Wifi for lots of folks. Do factor that into your offerings somehow if you don't already.

Good luck LS! Looks like awesome stuff you all are doing.

Don't you mean role/responsibilities specific? What happens if one of your students takes a web app developer role at, let's say, an auto parts manufacturer?

I don't think this is a fair comparison because it includes people with 10 years experience, and people with a 4-year computer science education. It would be better to look at the median income of developers with no more than 5 years experience and no more than 1 year engineering education.

Also the problem with increasing the minimum salary in the ISA is that it gives Lambda an incentive to reject applicants who are less likely to achieve that minimum. Personally I think Lambda should increase their $30k cap in order to align incentives for maximising the graduates income.

There are places for skepticism but the actual critiques made are shockingly wide of the mark in my opinion as someone who found the value proposition compelling.

I think if I asked myself at the beginning, what's the most you can hope for in going to Lambda School, my expectations were met by the education quality and the follow-through from the outcomes team. It's not perfect but the incentive structure forces constant improvement in the factors that matter most for getting us hired.

People point out it's not a CS degree and they're right. But the staff manages to teach concrete skills to get students hired, and the CS, data structures, algorithms stuff in the curriculum is taught well enough and with enough enthusiasm that it gives students the basis to be conversant with CS concepts and continue growing forward.

It might not work for everyone; some people are going to struggle to become programmers. But Lambda is trying their hardest to make it work for as many people as they can, and their heart is in the right place. And it is structured so if it doesn't work for a student their downside is hedged. Due to the $50k/industry-specific cutoff for repayment, the downside of going to Lambda but not making a career transition will be least harmful to the least advantaged students.

- Teaching assistants and section leads play a huge role in the quality of student experience. I think sometimes a supply/demand for filling these positions can result in not the best people filling them, which then can undermine the experience if your teaching assistant isn't the best.

- The system relies on teaching assistants for reviewing student code, and determining if students pass their tests. I worry what it means for ensuring students aren't progressing through the curriculum without having truly learned prior material. This goes back to having properly motivated and talented TAs too. Maybe some sort of randomized spot-checking, QA from full-time staff on this could help.

- I think sometimes the startup nature, remote-aspect, and fast iteration can mean various transitions through the curriculum aren't explained properly, but rather communicated piecemeal and students assemble an understanding of what's happening by discussing among themselves. There's a lot going on and structures are evolving constantly and sometimes instructors themselves can have trouble keeping track. I think being live in a room forces teachers to properly add ceremony and bookending structures to classes, but that can get dropped in the remote sphere.

Another thing, which is more a worry than a critique, is keeping the character/values and quality level there as it scales.

If only. We all know the majority of what we learned in school was not useful.

We should be striving to cut out the wasteful education that only serves as a signaling arms race https://www.theatlantic.com/magazine/archive/2018/01/whats-c...

We tend to underestimate that because we (engineers / people on HN) mostly happen to have a good education and interest in general knowledge.

Lambda School has a 9 month program that takes a 17% cut of your salary for the first 24 months of employment with a cap at $30,000. Which many students in software will hit, especially in high COL areas.

It's worth noting that this is more expensive than many/most universities, which charge less than $30,000 per year (9 months of instruction) even at the sticker price (which no one actually pays).

In fact, I wonder whether bootcamps are even cheaper than the full cost of university. The average student loan debt upon graduation for all US university students is $29,400 (again, for four years of college education with access to both applied and fundamental courses across many different majors vs. a 9 month boot camp focused on a particular skill set).

To be concrete: by taking out a loan for what lambda school costs, you could double major in both nursing and CS at a state U for about the same amount of post-graduation debt as you would have at bootcamps. Or CS and Econ. Or CS and Accounting. Or CS and Mechanical Engineering. In states that still invest in higher ed, you'll end up paying less (even with interest) than you'd pay lambda school.

There's a natural experiment on this question at Purdue, where there's an apples-for-apples comparison and loans are almost always cheaper than the income sharing agreement.

To me, these income sharing agreements and bootcamps seem like not great deals when you look at instruction time per dollar. And also when you look at the longitudinal durability of the skill-set that's being taught. They have a place in the market for sure, but they aren't the panacea to expensive education. They're not even cost-competitive.

They're just a different point in the design space of educational programs that extract as much of the added value of their product as possible.

I always have the impression that this perception that Lambda's model is not good always come from people in more comfortable positions that are not even close to the reality of who actually applies for Lambda.

I am from Brazil and did the career change two years ago, so there wasn't actually the option to do Lambda School for me. But it would be a no-brainer for me at the time and I am pretty sure I would be much better in my career now if I had done it (in knowledge and earnings).

4 years of college? Not an option for me. USD15k upfront for a regular bootcamp? Not an option for me. Take a loan with the risk of having to pay with a big chance of not getting a good job soon enough? Too risky for me (and in Brazil the high interest rates make it a ridiculous thought, but even with US rates it is too risky).

My path was self-taught through freeCodeCamp.org (awesome project!!!) and reading tutorials and documentation. I was able to get a frontend jr position, but I can tell that I was lacking a lot of knowledge/skills when I started (git, tests, agile process, design patterns, clean code, to name a few).

Now I am moving to LA and I am failing every remote technical interview because I lack the specific skills needed for being hired in the US.

Lambda School 9 no-upfront full-time months with close mentorship and teamwork, plus the alignment to help get me a job seems like a "too good to be true" option for me; not a "tricking naive laypeople who don't know better into paying their high price" that all "skeptical" commenters I have ever read in HN seems to think

It is not simply about sticker price.

It is about _risk_. You as the individual burden all the risk if you take a loan (or not) and pay upfront.

Here, even if your analysis is correct that you pay more, you must adjust it for risk. In finance this is your risk-adjusted return, but the same concept applies here.

Well, you know what they say about if something sounds too good to be true.

A student will have spent about 2,000 hours in Lambda School by graduation - the equivalent of ~4 university semesters, because we go all day every day and don’t take breaks.

And comparing the amount of debt vs the total possible cost doesn’t make for an accurate comparison either.

Also note that by the time a Lambda School student is at year four, if we want to factor in the opportunity cost of time, a Lambda School student has paid Lambda School back and has three years of earnings/work experience.

If you’re using time as the comparison: Would you rather have a degree and $24k in debt or three years of Lambda School, have paid the school off, and have earned an additional $250k?

If you enroll in 15 credit hours at a university, that means 15 hours in a classroom plus 15-30 hours of assignments and labs. (1 credit hour = 1 hour class + 1-2 hours outside class, on average, per week).

That is full-time.

CS may be somewhat different.

Many schools are also only in session between half and three-quarters of the year.

My universty/department's policy is to provide 2 hours of work per credit hour, and the students certainly claim to be spending even more than that! (The truth is likely somewhere in the middle.)

Good point about not being in session year round. It is fairly common to either take summer courses or to do an internship though, which I strongly encourage my students to do! Personally, I took summer courses every year of undergrad and did 5 internships in grad school.

As a university student, I respectfully disagree.

Also if you get through a few semesters at an accredited university and decide you hate it those credits will generally transfer somewhere else. The same is not true of unaccredited programs.

Unless, of course, you date someone of the "wrong" race or gender, convert to the "wrong" religion, join the "wrong" political group, or do anything else that your parents don't like. Then you'll have to pay sticker price, because your parents didn't submit all the forms the college required them to.

It's a priceless feature that every student would value differently... maybe that's worth a few thousand more?

This is a pretty common situation for people looking to pick up a trade skill to get out of the unskilled labor market, or for people looking to transition into a more desirable field of work.

if that's your metric, universities lose out wildly to all of the free education available online.

> It's worth noting that this is more expensive than many/most universities, which charge less than $30,000 per year

also a meaningless comparison; a freshman year of school alone doesn't get you anything.

Exactly. We also have no idea how these kind of income sharing agreements will work out long term at scale for the company. No idea what the cost of long term recovery will be. Personally I doubt these kind of income sharing agreements will be enforceable long term.

I start in May wish me luck

I have such a hard time accepting that people can't just teach themselves things using free online resources. It's been shown in many studies that self directed learning doesn't work, but it worked for me. And I can't stand the idea of smart people paying thousands to learn things they could learn for free.

It's a hedge, which is different than a true bet against yourself (i.e. a short). Semantics aside, surely hedging against yourself is not always a terrible financial decision?

For example, insurance is a hedge. Taking out life/health/car insurance is not always a terrible financial decision.

Taking out a large student loan is tantamount to taking a "long position" on yourself. You are betting that you'll make lots of money in the future, enough to easily pay off the loan.

If that is a good bet for you, go for it. But surely some people want a less risky option, either because they have less risk tolerance or they have lower expectations for their future gains, or whatever.

As another way of thinking about it, profit sharing is selling equity in yourself. If we accept your premise that profit sharing is "always a terrible financial decision", then surely no rational corporation would ever sell equity in itself except as a last resort; they would be financed by debt whenever possible. But of course we can observe this is not how things work.

Given that you could earn the entire cost of Lambda School (at the cap - the most expensive) in about three months of engineering time, it’s a rational decision even if you just speed up the process.

I'm not sure. But what's definitely true is that it's WAY harder as an adult.

Most self-taught programmers I know (including myself) learned to program as children/teenagers. Learning at that age is easier -- if not biologically, then at the very least because you have tons of free time.

The few I know who self-taught programming but didn't learn as youth already had really strong backgrounds in mathematics or physics, and leaned heavily on the skills that studying those fields teaches you.

That isn't really a valid argument here though because Lambda School is a full-time program to my understanding, so it's expected you somehow have 8 hours a day to put into it.

http://www.accsc.org/Accreditation/Standards-of-Accreditatio...

For one, I think incentive alignment is extremely powerful. Austen Allred talks about it a bit here.[1] (Great interview). Although reducing risk for the students is a benefit, it seems that the effect it has on the quality of lambda's education is much more important. In fact I think that a good source of startup ideas is to simply find existing systems with misaligned incentives and then figure out how to align them--I'm planning to try something like that in the music industry myself.

Also, I don't think profit sharing is a terrible financial decision. You're not just paying to reduce risk: you're paying to get a better education. You could take a normal pay-up-front bootcamp and slap an income share agreement on it, but that won't recreate lambda school. (Austen makes this point in the interview as well).

As for self-directed learning: it also worked for me, but I do think a lot of people need extra support before they can do it effectively. I was fortunate in that I was introduced to programming as a kid and thus had lots of time to just mess around. It's harder once you've already entered the grind.

[1] https://blog.ycombinator.com/a-cs-education-thats-free-until...

Another thing is the teachers/mentors. I'm not sure about Lamba School's quality but in general having a teacher or mentor who can guide you is very beneficial. The thing with programming is that there's so much you have to learn at first.

I am also completely self taught. I learned Ruby to apply into one academy but got rejected. But I continued to work through their recommended materials anyway and now I work remotely fulltime as a software developer.

I learned a lot about Rails from Michael Hartl's Ruby on Rails Tutorial book.

I learned the most through doing, though. I would work on building my own website and looked up anything I didn't know.

Higher Education is not 'Job Training'.

Also, nursing is not software.

I would love for nursing school to take on some more advanced methods, but I'm doubtful lambda is the way to go.

My hunch is that aside from some very critical basis things ... that much of nursing is learned on the job.

So this is actually in line with where we want to be in the future.

On the residency point, it adds:

"As part of the multi-pronged approach to alleviating the doctor shortage we also need additional federal support to produce about 3,750 more doctors a year by lifting the cap on federally funded residency training positions. Teaching hospitals are operating 10,000 residency positions without Medicare support, but cuts to Medicare and other clinical reimbursements jeopardize the ability of teaching hospitals to cross-subsidize with clinical revenue these positions.

The AAMC strongly supports bipartisan GME legislation introduced in both the House of Representatives and the Senate, the Resident Physician Shortage Reduction Act of 2019 (H.R. 1763, S. 348), which takes an important step towards alleviating the physician shortage by gradually providing 15,000 Medicare-supported GME residency positions over a five-year period. However, legislation alone will not relieve the doctor shortage.

In addition, the AAMC supports non-GME incentives and programs, including Conrad 30, the National Health Service Corps (NHSC) and Public Service Loan Forgiveness (PSLF), and Title VII/VIII, which are used to recruit a diverse workforce and encourage physicians to practice in shortage specialties and underserved communities."

Dr. Wilensky says that most likely, hospitals lose money on residents in the first year, when the doctors just out of med school waste a lot of time and money on unnecessary tests and treatments. But residents’ value goes up rapidly because the learning curve is steep, while their salary increases are not.

In fact, one data point that suggests hospitals are currently making a lot of money on the more seasoned residents is the quantum leap in compensation on the day doctors convert their status from trainee to attending physician, under a new job contract. How can a doctor be worth paying only $60,000 on Friday and then at least twice that on Monday? Does the doctor’s marginal revenue product — that is, how profitable the doctor is for the employer — actually surge that much?

We don’t know exactly how profitable individual residents are for hospitals because the hospitals can’t or won’t do the complicated accounting to figure it out, at least not publicly. But we can guess, at least, where hospitals believe they make the most money, based on how they’ve allocated their residency slots (both those Medicare subsidizes, and the 17,000 additional jobs that health care organizations have managed to create through other funding in the last decade).

Hospitals are not eager to find out the truth—I wonder why? Certainly doubling or tripling one's salary upon graduation indicates that later-year residents are highly profitable.

Most simply, the match should end and hospitals should be allowed to bid competitively on residents, just like they do for every other job. Then the market will clear and we'll see how much they're actually worth. First-year resident salaries may decline but I bet third-year resident salaries increase.

It's not clear the extent to which less educated medical providers are a problem: https://www.upi.com/Study-finds-no-difference-in-care-betwee... or https://annals.org/aim/article-abstract/2716077/intermediate...

Looking at the American med students I know, getting through med school in four years seems to require giving up all semblance of a private life and driving yourself to the edge of exhaustion/over the edge into a mental break-down. I don't think there's a way to do it in three years or even less.

That said, there might be a market for specialists with less training than that. There's no real reason to make e.g. an aspiring orthopedic surgeon sit through the entirety of med school, including all the dermatology/pediatrics/gynecology etc stuff.

But as another commenter mentioned, you'll have a hard time getting the regulatory bodies to accept your "eight months to orthopedic surgery" students as fully qualified doctors

I think the difference is that many students enter German (and other non-US) med schools immediately after high school, while US med schools only accept students who already have a 4 year undergrad degree.

So in reality US medical training takes 8 years (4 years of pre-med undergrad, 4 years med school) which makes the 6 years look relatively efficient.

Also those non-US doctors graduate with substantially less debt.

I believe it aligns the interests of: 1)student 2)school 3)employer

Far better than the existing conventional university model in teaching tangible trade skills.

Lambda School can and will adapt to the ever changing needs of employers far faster than universities because Lambda School’s very existence depends on it.

However, does the same apply to regulated and licensed “guilds” like nursing, medicine, dentistry, etc?

Will “guild” like protectionism of their profession impede efforts in this direction?

Lambda School already seems to be making enemies understandably in the traditional university computer engineering and upfront paid bootcamp sectors.

Will it make enemies with professional regulatory bodies who may be biased to prefer higher education hurdles and fewer licensed practitioners to protect their income?

There was a recent post on how a simple thing like checklists can really cut infection and death rate by half or something. And there were startups that attempted to do this at the doctor level but failed to be able to drive change.

Would you be the GitHub of nursing - drive change from bottom up. Build checklist creation and adherence at the nurses level ?

I'm not so sure. The solution might not be to train more nurses, but to retain the exiting nurses. Nurses have a really high turnover rate in the U.S. The problem is nursing is demanding work and it often requires long hours.

> Nurses who work more than 12 hours in a single shift and more than 40 hours a week are likely to leave the nursing workforce within a year. Overtime for nurses should not be commonplace. Instead, hospitals and other healthcare organizations should reduce the length of shifts and the number of hours in a workweek. Hospital administrators should not push nurses into working extra hours.

https://rnbsnonline.unm.edu/articles/high-cost-of-nurse-turn...

> A study by Compdata Surveys of 11,000 healthcare employers with more than 11 million employees found the average turnover in healthcare jobs in 2017 was 20.6%, up from 15.6% in 2010, putting healthcare’s turnover second only to hospitality’s.

https://www.beckershospitalreview.com/finance/will-2018-be-t...

> Turnover rates also varied significantly across countries with the highest rate reported in New Zealand (44·3%) followed by the US (26·8%), Canada (19·9%) and Australia (15·1%).

https://onlinelibrary.wiley.com/doi/abs/10.1111/jan.12483

This might be a lesson in capitalism, but maybe not the lesson that The Economists thinks.

So, 2/3 of what they're paid (presumably after Austen Allred's and the VCs' take), is marketing, leaving the leftovers to pay for actual instructors. If I end up in the hospital, I want a nurse with an actual education, not "nursing boot camp."

We have ~80 full-time employees and about 200 part-time. 40 of the full-time and all 200 part-time are dedicated to teaching/career coaching.

growth (what you call marketing)

student success (student support)

* engineering

* analytics

* finance

* operations/HR

The marketing team to instructor ratio is about 1:50