rabbits77

In a van down by the river.

If you think tuning GP parameters are challenging wait until you try tuning hyper parameters for a DL model!

It is just a press release, to be fair to DeepMind, and I guess they can promote themselves however they wish.

My original comment was more from the context of seeing neural network models in practice perform barely any better, if at all, then classic ML models. Just as those comparisons were revealing similarly I was suspecting this use case may be the same to another classic technique.

GP is certainly not the shining star of AI right now but it is actively researched and perusing Google scholar on the subject will show you plenty of interesting, but less heralded, results.

There are probably several meaningful metrics for this problem that can be examined. If nothing else it is a simple matter of grading the solutions of each, like a university assignment. Also, typically classical techniques are less resource intensive then any neural network methods; the energy savings alone when considered at production scales would be significant.

What I always find missing from these Deep Learning showcase examples are an honest comparison to existing work. It isn’t like computers haven’t been able to generate code before.

Maybe the novelty here is working from the English language specification, but I am dubious just how useful that really is. Specifications are themselves hard to write well too.

And what if the “specification” was some Lisp code testing a certain goal, is this any better then existing Genetic Programming?

Maybe it is better but in my mind it is kind of suspicious that no comparison is made.

I love Deep Learning but nobody does the field any favors by over promising and exaggerating results.

I was wondering how he ended up and the biography on Wikipedia is unclear on how his life may have improved after publishing his results. He published in 1978 and passed away fifteen years later. Other than a note that he joined The Party in 1991 there is no information on his last years.

I am not sure what to make of the information that is given. Does it imply he was treated poorly, even tortured, until 1991 when he relented?

I agree completely. I disable all comments on my blogs and if people want to discuss then Twitter or Reddit seem to make much more sense.

Otherwise you end up spending a ton of time just removing spam, never mind trying to make sense of and respond to the comments that may get randomly left years after the original post.

Perl has a similar syntax, it is quite nice

    sub my_funct{
        my($x, $y) = @_;
        #…
        goto &my_funct($x, $y);
    }

In Perl this is convenient because effectively with just adding the goto keyword you reduce all the memory use and risk of blowing your stack (recursion is otherwise super inefficient in Perl).

Not at all. The papers where I was first author were essentially 100% all my work. Second or third authors were advisor and a committee member that helped edit, closely read proofs and pseudocode. Certainly helpful and deserving of authorship but a steep drop off in contributions from first author (me) to them.

Advisor had family in Europe (I am in US), and would use these conferences to create extended family vacation plans.

Of course when I think back on this I am still pissed! But years have passed and I am doing well in my career. I am able to not get too fixated on it.

And yes, I agree with the people saying it was terrible and very wrong. In terms of all the possible terrible and very wrong things that can happen to a graduate student this is maybe, in the grand scheme of things, about mid-range maybe?

I recall the same things. I remember being bitterly envious of students that seemingly were given topics to examine while I had to spin my wheels for a good year or two coming up with my own ideas to explore.

Looking back my perspective is very different. First of all, my ideas generated a minor spike in publications for the lab all centered around my work. After I graduated I continued to advise new students to continue what I started.

I now think in these teems, which might sound cynical but simply reflect the vicious nature of academia.

    1. The students given topics, the ones I was jealous of, all kind of sucked. They were given topics to advance a short term goal and gtfo. It may seem paradoxical or cruel but a savvy advisor will maximize both short term and long term gains. I was a long term bet, the others were short term plays.

    2. I benefitted greatly from being forced to identify my own topics. This is the one skill that I use every single day. Every hour of every day. As time goes on the ability to evaluate ideas deeply and with some speed effectively defines what it is I do for a living. 

    3. Students given topics were cheated out of more valuable long term skills for the lab’s short term gains. This is not always universally true, of course. Some super stars really can crank through a deep serious topic quickly and continue on to generate novel research if their own. One such person may appear in a university department once every ten or twenty years, they are extraordinarily rare.

    4. I was well aware of the exploitive nature of grad school, did it anyway with a clear head for what I wanted to get out of it and my only disappointments came from when I giddily let my guard down and expected more than what I already realized would be forthcoming. A specific example, my advisor would use my conference paper acceptances to fund their own personal travel and vacations; I was not allowed to present my own first author papers. Silently tolerating that sort of bullshit, in part, allowed me to graduate.

I worked at a small ISP/web host in the late 90s. A very common web server at the time was a cheap white box PC running Windows NT and IIS. We had many of these systems on this interesting “rack” that was more like a multi level table or three d chess board.

These systems could handle a ton of traffic very reliably. Consider that there was very little dynamic content and what there was barely taxed the CPU at all (e.g. a Perl CGI script to query and display some tiny amount of data from (LOL!) an Access database.

Probably not “screwed” at all really. From a professional point of view anyway, since many of the practical things you’ll learn may not match industry expectations by the time you graduate.

What is concerning is this “visceral hatred” for the curriculum. You’d be well served by exploring why that is the case! Ask around, what are graduates of your program up to these days? Say 1, 5, 10 years after graduation. Search around on LinkedIn to get some idea.

I know the disgruntled genius trope is popular on HN, but it is extremely rare irl. What is common irl, however, are delusional undergraduates that feel the faculty at their school is holding them back from a life of fast cars and never ending excitement!

If OP thinks college is somehow eroding some former joy in mathematics than I have no idea how jumping straight to some job will help? Could it be that reality is just not meeting some unrealistic expectations expectations?

If I were this student’s faculty advisor I would spend a lot of time talking to them about this. In my experience this sounds like the tip of a very large iceberg that may require some counseling and intervention.

I do industry R&D. I don’t think I am all that fettered. I do accept that there is a certain “social contract” involved though! I understand my employers business and my work is done with the ambition to advance the companies long term position.

Practically speaking this means that my work is restricted in that it deliver something resembling a product or service within a couple of years. Even if only the thinnest of MVPs of a concept. String enough of those together and you get pretty close to something resembling “unfettered”.

I suppose my point is that industry R&D is not all that bad right now, you just need to be a reasonably responsible corporate citizen.

Correct! Is it not fair to say that Python was at least initially popularized by the perception (right or wrong) that its syntax made it easier to learn? From there it rapidly spread (network effects) within the community of educators and learners and then outwards into the education adjacent field of scientific computing.

I see perception and network effects having at least initially a stronger factor in Python’s popularity than any particular technical advantage.

There is nothing in that press release that could not have been done in the 1980s with Prolog.

Yeah, it’d have been more code but you would not have needed to destroy a forest to train the thing.

This is the NLP trade off of the 21st century. The code is easier to write but the model is completely opaque, and you need to really burn a lot of electricity to make it work.

direct (non-Reddit) links:

https://www.nntp.perl.org/group/perl.perl5.porters/2021/04/m...

https://news.perlfoundation.org/post/cat-transparency-repor-...

Matt Trout (mst) has long accumulated criticism over his behavior but it seems antisemetism was what finally led to his removal.

Removal: bans from all Perl and Raku irc channels, mailing lists, conferences, etc.

Example of MST being told by Sawyer to leave him alone and MST not relenting.

Of course over the years he has done worse to others. Mainly new users, especially women.

https://gist.github.com/wchristian/4e1bcdb761f20985b0a5c97a4...

The main culprit is Matthew Trout(aka mst). He is basically the Perl communities RMS. He’s toxic but has his fans that will defend him and his “quirks”. Whenever he gets in trouble he of course immediately wraps himself in the flag of being neurosivergent or polyamorous or whatever else he feels might best absolve him this time around.

Perl has an MST problem it needs to solve.

The solution to the Stallman problem at the FSF is probably as simple as removing his name from everything, posting some announcement on a “non-free” website.

He’ll never notice to cause trouble about it, he lives in a bubble.

Soon the problem will just solve itself, given his age and various obvious physical issues he’ll be dead in the next few years anyway.

I saw somewhere that Paul Graham whole set of academic credentials are a farce. Through means similar to this story he basically paid others to do the majority of his work, all the way through grad school.

Partly explains why his language projects like Arc have been a hot mess.

When did working at one of these places start to be of any significant status?

I question being concerned about a “gap” when the only outcome is a better chance for working at some advertising company.

if he raped a woman would you still say that? what if he raped a black woman while calling her the N word?

of course you would change your mind, because you have no consistent philosophy.