HN user

willcannings

39 karma
Posts1
Comments20
View on HN

Most? Almost all my requests to the "Auto" model end up being routed to a "thinking" model, even those I think ChatGPT would be able to answer fine without extra reasoning time. Never say never, but right now the router doesn't seem to be optimising for cost (at least for me), it really does seem to be selecting a model based on the question itself.

You can distribute to "internal" testers (people you add to your developer account) without any review process. The app is only reviewed if you want to distribute to external testers.

People with depression do not generally have low serotonin levels, and it's unlikely low serotonin is the cause of chronic depression. If it was, serotonin agonists would be effective much faster than the 4-8 weeks they currently take. There's some evidence that increased serotonin levels modulate glutamate after about 4 weeks, and glutamate antagonists like Ketamine are effective on depression more rapidly than serotonin agonists (as quick as 40m in some studies).

Also the depression effect of alcohol is completely different to chronic depression, any mood effects you feel during a hangover are likely related to reduced dopamine.

Sleep is very important though, and, in Australia at least, melatonin is commonly prescribed for sleep issues that have a psychiatric cause.

You possibly didn't fool them - they may have suspected you were faking the symptoms, but when a patient shows up to a psychologist/psychiatrist on their own volition it suggests they're in distress. It could be considered malpractice to turn someone away who's seeking treatment.

A friend of mine spent two weeks in a hospital psych ward (in Sydney), and requested his records after he was discharged. It was surprising how much detail the doctors noted in the records, and how often they (correctly) picked up on avoidance and lying (yet they didn't give this impression in person). Based on this experience, it's possible the psychs your class visited gave you the diagnosis you wanted, but privately noted something very different you weren't aware of.

There has been a lot of progress in objective diagnostic tools recently - e.g there's a clinic in Sydney that diagnoses ADHD using EEGs, and selects medication based on response to the med (also measured using an EEG). And I read a paper last year that described a machine learning model trained on MRI images that could diagnose some of the organic psych diseases (bipolar, schizophrenia, depression etc.) with very high accuracy (70s/80s from memory), and distinguish between the diseases in borderline cases.

Psychiatry won't rely on symptoms alone forever, and don't think psychiatrists aren't already aware of the problems. But every psychiatrist who looked after my friend used whatever tools they had at the time to try and help. It's far from perfect, but they did their best and they really did help. And I'm really glad they did because he's finally a genuinely happy person.

It's interesting seeing differences between countries. Like the author said in the comments, 401k isn't an "issue" in Australia. I don't know how 401k works, but Superannuation in Aus is a compulsory 9%-10% of your salary that your employer sends to your super fund (that you can't access until you're 60). If you're a contractor you don't have to pay yourself super (so you could have more immediately available money), but you could also put away 9% just like your employer would. There's no insurance issues because we get government provided healthcare, and depending on the contract structure you go with there's actually a few decent protections an employer is required to provide you (if you structure it that you're a company contracting to another company then you're out of luck there).

So realistically the major difference is "security", which to be honest there isn't much of regardless you being an employee or contractor. So at least in Australia, unless you're getting an increase in your hourly rate, or a company says they'll only continue to send you work if you're an employee, there's no major advantage to switching to be an employee.

Re #1 and #3, that's not really true. They performed a manual landing, and the procedure in any emergency is to 'follow the book' regardless of the type of plane being flown. Their job was to evaluate the damage & take actions informed by their experience, ensure the plane was in a state to land, and then land it. That's what they did, a computer didn't do that for them.

Other implementations you haven't mentioned like MacRuby are much faster. In my totally unscientific and completely biased benchmarks (small tests related to the type of work I do mainly - Machine Learning) MacRuby is now faster than Python (2.6) and is closing the gap on V8 (which, in the tests I wrote, was even faster than Go, and less than twice as slow as C).

"By lacking sophistication, I mean a system of alternative capital (VC) that fully recognises Startups as a serious way to generate wealth"

That's definitely changing, e.g Innovation Bay, Startmate, the Sydney Angel Sidecar Fund, and others. VC/Angel investing is not at the same maturity as overseas yet, but it's definitely changing.

Take a look at some of the other projects on Kaggle - a lot of them have no money or roughly similar amounts. One of the main crowds Kaggle attracts is University students and staff, and they do these projects because they're interesting and fun. It's similar to a shared task at an academic conference, which is the sort of thing most of the active ML/KD&DM community are interested in.

You already have a lot of advice :) But I'll add that if you focus on negative memories a lot you may need to consider if you have OCD or an anxiety disorder, and the anhedonia could be a sign of depression. The best treatment for these disorders is a short term medication with longer term psychological treatment (normally cognitive behavioural therapy).

But, if a psychologist or psychiatrist is out of the question, you can try some cognitive behavioural therapy online for free and in private: <http://moodgym.anu.edu.au/welcome>; The study behind this website actually showed it was more effective than visiting a therapist in person! There's a decent number of CBT exercises to complete, and it seems quite a few people get better simply by using the site without any psychological intervention.

But more importantly: it's sometimes possible to channel your negative emotions to good use. But sometimes it's not, and that's OK. Forget the advice that you should just 'man up' - if you feel the need to reach out here then it's a sign that what you have may be more than just a low attitude. It's OK to ask for help, it's OK to get it, and regardless of whether there's a stigma attached to mental ill health, it's OK to get professional help. There's nothing 'wrong' with you, but you can get help for everything you've mentioned if you feel you need it.

Yeah their credits are a bit sneaky :) Twilio only supports US numbers which is a problem. If you're looking at telephony companies, Tropo (https://www.tropo.com/) is only 2c per message and sorta supports international numbers. Their 'support' is basically "if it works, awesome, if it doesn't, well we never said it definitely would". It's worked well to Australia for me so far. International support is coming at some point, though I don't know if it will still be at 2c per message.

That doesn't scale to the rest of the world though. At least here in Australia you're free to port your number to any carrier at any time (I'm on my third carrier with the same number I've had since I was in high school), so you can't use the number on its own to know what network you're sending to. I'm sure this is the same in other countries as well.

Beware of moonshado unless you're sending marketing messages - their support pages state that all texts sent from them have this footer: "Msg&Data rate may apply txt STOP to opt-out". Not something I'd want if your users are paying to be sent messages (like with this service).

I'm looking at using CDYNE for texts. At volume they also charge 3c, but there's no distinction between networks with their price so you pay the same to text Nepal as you do San Francisco.

I think it's better to say that they do it exceptionally well with scale. Most algorithms are conceptually simple (once you understand them), but to make them efficient on the sort of scale Google does is sometimes hard.

That said, I don't think tagging for them would be very simple like you say. For a start they're dealing with multiple languages, probably many languages without any human annotated training corpora. Even for the languages with training data, web pages are difficult to tag & parse because they often contain very 'slack' grammar and domain specific/slang words. The standard English training corpus is the Penn Treebank (Wall Street Journal text), can you imagine trying to read and understand youtube comments if all you'd ever read was the WSJ? Even tagging search queries would be difficult because they're not even sentence fragments you could use viterbi with, they're often just words strung together without any grammar construct at all so you can't rely on the tag order you know from your corpus to help you tag a query.

So I'm very impressed that they're doing any tagging at all, on the scale they're doing it at, and with presumably decent enough results for it to be useful.

I normally read. Nothing that will get me excited or thinking too much, like some easy books or a newspaper. But - if you frequently can't sleep it's possibly a sign you're stressed and need to take a break. It can feel stressful even just thinking about taking a few days off! But the productivity you'll lose by not working will be more than made up by your increased productivity when you get back and (hopefully) feel more relaxed and focused.

> Really? That's news to me. How do they check?

Typically diagnosis is made after an interview (about an hour) and questionnaires. If there's some doubt about the diagnosis you may need to bring in old school reports. It's also possible to diagnose using brain scans now (e.g http://www.sydneydevelopmentalclinic.com.au/brain_scanning.h...), and medication can be trialled and selected on the basis of an individual response to the medication (shown with another brain scan, or using testing, e.g http://www.sydneydevelopmentalclinic.com.au/medication_testi...).