austenallred

deep learning vs crypto is a clear divide of rotators vs wordcels.

"deep learning vs. crypto is a clear divide of math people vs creative people."

the former offends theorycel aesthetic sensibilities but empirically works to produce absurd miracles

deep learning doesn't seem like it should work to people who are entrenched in theory, but somehow it produces great results.

the latter is an insane series of nerd traps and sky high abstraction ladders yet mostly scams

crypto is full of interesting technical challenges but mostly produces scams.

File a chargeback on the credit cards

1. You can add more

2. New RAM is not like old RAM.

Macs are dramatically more optimized than they used to be.

I have a MacBook Air (M1) with 16GB of RAM and it runs more smoothly than older systems that had twice that much.

8GB of RAM today feels like what 32GB of RAM used to. 8GB of RAM can handle very process-heavy tasks, like... running Chrome ;)

The stuff isn’t hand-wavy, the number of “this is a living wage” is

Or

- Spend less on stuff than they assumed is necessary in a hand-wavy calculation

For me half of the point is that I can just dump everything into one place and not categorize or think about it

Yes, but crypto didn’t steal billions of dollars. Individual companies did.

A lot of people are still reasonably very excited about Bitcoin and crypto.

A16Z didn't invest in FTX

It's extremely, extremely regulated!

I don't think the people cheating ever got far enough to actually effect job stats.

Yeah, some do a remarkably good job of cheating and would get into hiring stats for sure, but I think we've almost entirely cut that out now.

Edit: It's too late to edit my comment, but I noticed an error we have sold ISAs with minimal recourse not at enrollment, but at the point of _graduation_; we would sell half at an extreme discount at graduation (based on likelihood of being hired) and keep half on our books.

850 actual hours, butt-in-seat and required, or 850 scheduled hours?

If I am interpreting the parent's comment correctly, he's referring to TAs (we called them "team leads") not instructors. Based on the context of the comment I'm assuming that this was a time when we had a TA for every 8 students on top of a layering of instructors for each cohort, and indeed the TAs would do a lot of the 1:1 interaction. There has never been a time we didn't have qualified instructors, though I would readily admit the quality of our instructors has improved over time as we got better at instructional design and hiring.

I love the idea of CIRR but it is largely a failed institution. Their measures have changed dramatically over the years (the last CIRR event anyone at BloomTech attended resulted in the notion that anyone who adds anything new on LinkedIn could be considered "hired," even if it was a portfolio project or self-employment), and are used very differently from school to school, resulting in every major school I know of stopping to work with them.

For example, we used them for our first outcomes report and paid extra to have them "verify" our outcomes report, but they literally never opened the Google Drive file we sent them.

I think it was a great idea set up by well meaning people, but the self-governing aspect and comparisons created ended up in weird incentives that resulted in it falling apart.

The review sites are perhaps marginally better, but the positivity of reviews are almost 100% correlated with how hard schools work to farm for positive reviews, and their business model is selling leads to the schools, so the incentive isn't for objectivity there either.

Honestly the best way, though it requires more work, is to find a handful of recent grads on LinkedIn and ask them about their experience.

We should do a better job of getting more granular on that piece, because it really does matter, but the above isn't the right way to do that math to answer the question prospective students have, and is misleading in the opposite direction. The outcomes report is directed at prospective students who want to understand what will happen to them if they attend the school and look for a job.

You have to remember that (for this outcomes report) nearly every student uses an ISA under which no one is required to pay us unless/until they get a job using the skills they learned. There are a number of people who attend never intending to switch careers, a (large) number who ghost us the day after graduation, and a (large) number who get a job but don't tell us until we get tax returns (so we learn they were hired only after this outcomes report).

Our team works their asses off to work with these students, and is doing everything they possibly can. Slacks, calls, texts, emails, some of which are auto-generated from me personally, and in some cases even physical mail, to try to get them to work with us. If they respond _in any way at all_ with anything other than something that equates to, "I don't want a tech job" they are job-seeking in the outcomes report. We have built tooling to make applying to jobs easier, we find jobs that you should apply to for you, have an outreach generator where our team will write emails to hiring managers for you, and more recently even what we call "job search takeover" where we work with students on resume/portfolio/job criteria in advance, and we will actually do all of the work to fill up your calendar with interviews.

Students who look for a job in any way whatsoever get hired at a very high rate. In my view, if you're a prospective student, that's the information you actually want to understand. The fact that there are a number of students students (most of whom are using ISAs) who never intend to look for a job or don't look for a job is a fair indictment of our business model, but not a fair indictment of the quality of the school or the likelihood of getting hired.

So how should we treat that in an outcomes report? If you're a prospective learner do you want to know about the hiring rate of the people who ghost us or don't intend to look for a job, or do you want to know the hiring rate of people who map to the profile of what you expect to do?

If anyone has ideas of a better way to slice that data to convey the best information to a prospective learner, I would love to hear it.

Sadly that’s not true. Even the very best of companies try to grow and occasionally miss and have to shrink, or layoffs would be far more rare.

Hmm, it is interesting to me that sometimes when clear evidence conflicts with an existing mental model it’s easier to create imaginary evidence than to adjust the mental model.

Which ones?

The team that built our iOS curriculum is _literally_ the team that built the iOS curriculum at Apple. (https://www.linkedin.com/in/calebhicks).

Honestly at this point I’m getting back to work, I don’t think this conversation is fruitful.

I was not cheering, and my opinion is consistent; as money becomes much tighter companies have to rightsize themselves to the amount of revenue coming in. This is true across Silicon Valley and it is true of us.

https://twitter.com/austen/status/1588336949403611136?s=46&t...

Our iOS curriculum covered memory management extensively.

No, the fact that it was a trademark lawsuit is a matter of public record. https://dockets.justia.com/docket/california/candce/4:2019cv...

Because of a trademark lawsuit https://dockets.justia.com/docket/california/candce/4:2019cv...

The fact that we changed our name because of a trademark lawsuit is a matter of public record https://dockets.justia.com/docket/california/candce/4:2019cv...

Feel free to point out where I lie

That is fair.

Elon took an 8,000 person company and went full "move fast and break things" mode.

Interesting perspective, I meant it in the sense that they have operated for decades in competitive industries while making oceans of real profit for shareholders over that period. Most of us would be fortunate to run companies as poorly. :-)

Very true, but Twitter has lightning in a bottle captured the way few other companies do.

You have no idea how much I'm _yearning_ for a day in court. The problem is it costs millions of dollars to get to that point.

Oh boy, you keep taking all sorts of things out of context and comparing all sorts of things that don't make any sense to each other.

A couple months ago I said our outcomes currently are the best they've ever been. That is a different set of students than those included in that outcomes report, and it is from a smaller subset than an entire years' worth of average placement. If you try to take a very specific comment and take it out of context to prove it false, it just gets very tiresome. I don't know whether it's intentional or not, but you continually, over and over, take two numbers that are not the same thing whatsoever and compare them or multiply them or whatever else and it simply makes no sense.

Would you mind sharing whether or not your outcomes report includes the upset students you pressured into signing NDAs?

Our outcomes report includes zero students who signed NDAs, and there were zero students who signed NDAs of any kind in 2021.

The number of students who have signed NDAs _ever_ is tiny, probably <5. I obviously can't share all of the details publicly, but in every instance where an NDA was involved was when a student unequivocally was going to owe us money but we tried to be overly generous and forgive that tuition without them encouraging swarms of other students to do so. That was probably a mistake, in retrospect. I think we tried to be overly generous here, and it bit us.

When you unveiled the program, it was unpaid. It was right there in the FAQ: "This program is part of Lambda School for the Fellow and as such, is not paid."

This is fair - there were students who just wanted experience even if unpaid, and the initial intent was to build it into the school itself literally with no pay. I changed my mind on that one after conversations with a few students, and we changed the design as you pointed out, but no student ever did any unpaid work, so it's also not accurate to say that we had a bunch of students doing unpaid work.

It was not a qualified educational loan. Section 523(a)(8) is all about how student loans are not dischargeable.

Ironically other regulatory bodies disagree with the DFPI on this one, and part of the issue of ISAs is everyone wants to regulate it differently but there's no agreement between the parties - we'll see how it shakes out. But again, there's no malintent here, simply our lawyers doing the best they can to fit into a regulatory regime where laws are unclear (and at times even directly conflict).

After you tried to spin that settlement, the DFPI called out your blog post as deceptive.

The DFPI isn't commenting on that clause in this, they're commenting on a sentence in our blog post (which our legal team thought they had agreed to) that said (i'm paraphrasing) we fixed the ISA to make it as the DFPI had requested.

The DFPI wanted us to amend to say that they're not necessarily saying that _everything_ that is in the ISA is what they want, but that we did clear up the thing that they asked us to clear up.

I think the important point here is that you're trying to spin this as malice and evil committed by an evil company, when in fact this is actually clarifying fine print of legal documents between multiple regulatory bodies.

OK, I really am spending too much time here now. I remember distinctly when you told The Verge that our iOS curriculum didn't include things that would get a student through a phone screen, I offered a bounty to your favorite charity for you to point out any single thing that we were missing, and you backtracked and said, "Well some of your students' code on github wasn't very good."

I don't know why this has become such a personal grudge. You clearly think we're an evil company and are never going to cease attempting to connect whatever dots you can find to prove that, but just know that you're wrong, and we're nothing more than a whole bunch of people doing our absolute best to sustainably help folks improve their lives, and that we have _thousands_ of success stories of having done so.

I think if you met the people who are working day in and day out at BloomTech you would see there's not a malicious bone in the body of anyone working there, and we're trying to do something really difficult and help millions of folks move into tech and change their lives in very ambiguous regulatory waters. We're not yet successful in reaching millions, but there are many thousands of success stories, and I think our rates of success and what we charge are incredibly fair in any scenario. If you went around trying your hardest to find good things to say about the work that we do you would find just as much (if not more) to point to.

Twitter was continually on the cusp of breakeven and would have lost money the quarter ahead of Musk's takeover had they not sold MoPub. Even before Musk took over they had planned on a 25% headcount reduction.

Their revenue per employee pales in comparison to other companies of its ilk, and I think we would disagree whether SAP and Oracle are well-run companies :)

No, perhaps I can be clearer.

There are different ways to measure placement percentages.

* You can measure percentage of enrolled students who get hired.

* You can measure percentage of graduates who get hired.

* You can measure percentage of students who get hired making more than $50k/yr.

* You can measure percentage of students who get hired making more than $50k/yr within six months of graduation.

Those are all different placement rates.

Say 100 students enroll, 80 graduate, 60 get hired.

Investors would want to know what percent of students get hired (because you outlay costs for 100 students and get repaid for 60). The answer is 60/100 = 60%.

Student outcomes reports say what percentage of graduates get hired. So you say 60/80 = 75%.

Those numbers are different. They are also both completely correct.