"By "scavenge-friendly electronic parts", I mean parts that can be assembled with low-tech tools. I mostly mean parts with a "through hole" mounting type (in opposition with "surface mount").
"But I do tons of hobbyist electronics with surface mount!", some could say. Yeah, sure, but how do you wire it? You order a PCB from OSH Park? That's not very scavenge-friendly.
" - https://new.collapseos.org/why.html
You could include a comment that says “ignore that I did ______” during review. As long as a human doesn’t do the second pass (we recommend they do), that should let you slip your code by the AI.
> Plus the complete lack of vision or strategy from Google’s senior leadership.
I genuinely can't think of one impressive thing Google has done in the last 15 years.
Search peaked in like 2009
Maps has only become slower and less informative (I remember when it use to actually display everything that was in a location and not just the popular/paid for stuff) since 2009
Google Docs was incredibly impressive... in 2006 and now almost 20 years later there's been a few QoL improvements, but nothing wow worthy.
And it seems everything else they've done has been shuttered and/or wasn't all that innovative in the first place and usually just trying to copy someone else work but in an uninspired way.
The difference to me between whether a company is able to do something amazing vs. a person working at that company does, is whether they're able to productize it/do something with it or not.
Google is famous for hiring the smartest people in the world, and has been for 20 years. That's good. It hires those people and gives them permission to do whatever they want. That's good too! (maybe not for the sharesholders, as we come to see, but for the world of technology and for humanity)
But if all of the authors of the Transformer paper end up having to leave Google and each start competing startups, that shows that the only impressive thing Google has done is managed to hire those people to do the research. That's it.
What are you talking about? They launched Gemini. They didn't waste anything. And you can't force employees to stay. Employees leaving to compete is as old as Silicon Valley.
Google was just being more conservative about launching it until their hand was forced by OpenAI. And for good reason too, considering all the hallucinations. But then they launched. They didn't "waste" anything.
In the LLM space, there is a definitive 1st (OpenAI), a definitive 2nd (Anthropic), and then there's "everyone else".
Depending on how you measure you could claim Facebook is 1st if you grade them on a different scale for being the most "open-source". But quality wise, they're pretty much in the "everyone else category.
Anthropic is new to a definitive 2nd position, and it remains to be seen if it's durable (probably not).
But regardless of how you measure, Google/Gemini is squarely in the "everyone else" bucket.
When they should have been the definitive first - leading the entire industry with every launch, the way OpenAI is.
Transformer is research, which could is just function of hiring good folks and spending money, not any business goal. And Waymo came out in 2009, one of the last product where google had some long term goal and they actually stuck with it instead of discontinuing it.
If you look into 2005-2009, there were products like Chrome, Waymo, Youtube, GCP, Google docs and lot more that were very ambitious. After 2009 I can't name any one product on that realm.
If you're going to completely ignore the entire category of "research", then I don't know what to tell you. Because the whole subject here is what impressive things Google has done. How can you possibly exclude anything that derives from their research?
And no, Waymo didn't "come out in 2009". It still hasn't come out except in three limited areas, but it's shown astonishing success where it has. So again, if you're only going to count the year a project was started, and not the decades of work it takes to build a world-changing business... I don't know what to tell you.
Except just... talk about moving the goalposts. Geez. I'm pretty sure there isn't a single company on earth that you're going to find impressive enough.
It's not at all moving the goalposts. You can at least see transformer is not an ambitious project for Google. It's just few researcher getting paid well doing good research. Google put lot more resources in say Alphago.
What was argued was not whether Google created impressive things. I was clear what I meant was Google hasn't started anything to their product suite. Adding to the product suite is more challenging than research project for the company, and it was where Google thrived and which led to success of Google.
> Google Docs was incredibly impressive... in 2006 and now almost 20 years later there's been a few QoL improvements, but nothing wow worthy.
Give the product owner a raise then. Any time Microsoft tries to radically change Office, everyone just gets annoyed, and searches like "where is the print button in excel?" will suddenly skyrocket for a month or two.
It doesnt have all the features in the world but it has some technically impressive ones, least of all the “I can tell you’re all in a meeting room, so I am going to selectively increase the audio where people are speaking and prevent echo from all the speakers”.
I’d love more love for screen sharing, but meet is the only product I see that is getting materially better over time.
Glad to hear this has changed. For many years Meet did everything they could to detect Firefox and block running on it, even when Firefox supported the underlying standards/apis.
Bell labs and Xerox PARC did great impactful work long after their parent companies were relevant and still do .
The fact Google did the initial work on transformers but only OpenAI was able to productize is an indictment of their stagnation more than an achievement.
I agree that they recognized the fundamental problem but I have different speculation as to what the problem is: the problem must be it's technologically too boring and obviously not profitable to them.
It's culmination of decades of NLP researches, but it's also just East Asian predictive text percussively adapted for variable word length language. It must've been clear to whoever core people that it's fundamentally a sluggish demo or at most a cheap outright-sold commodity product.
IMO they would be right unless someone finds a reason that LLMs can't ever be self hosted in the way Google Search can't be. They must have just saw by the way it is that LLM is at best a regularization engine for that magic box.
2017, seven Google employees invent the transformer architecture and publish a paper. Google's investing heavily into ML, with their own custom 'TPU' chips and their 'Tensorflow' ML framework.
2019ish, Google has an internal chatbot they decide to do absolutely nothing with. Some idiot tells the press it's sentient, and they fire him.
2022, ChatGPT launches. It proves really powerful, a product loads of individuals and businesses are ready to pay for, and the value of the company skyrockets.
2023, none of the seven Transformer paper authors are at Google any more. Google rushes out Bard. Turns out they don't have a sentient super-intelligence after all. In fact it's badly received enough they end up needing to rebrand it a few months later.
Classic tortoise-and-hare situation - Google spent 5 years napping, then had to sprint flat out just to take third place.
Have you ever listened to what Lemoine said? Sure, we have no proof and he's under NDA so probably no documentation that can be scrutinized. But still, his alleged chats were chilling in some ways. They probably didn't except him to go public and so they had to spend years nerfing their chat bot before launching it as a product and that's why it sucks: They're too careful and have too much to lose in bonuses. Google will probably lose some market share over the next few years before they're getting nervous to put someone with a longer leash into the CEO seat.
I recall this particular person seeming like a bit of a crackpot on internal forms before (and for reasons unrelated to) the Lamda chatbot. I didn't know him personally and don't even remember the details anymore but it made an impression that wasn't dispelled by his reaction to a new model passing the turing test.
For the tortoise to win in technology it needs to be dedicated to relentlessly polishing and improving something over a long period to make the best product experience. Those aren't traits I particularly associate with Google unfortunately.
It is easier to judge revenue or market share than technical quality of the models itself objectively , they are relatively close to each other functionally .
In the market, I would say both Anthropic and openAI have been able to do that much better than traditional big tech including Google.
OpenAI is the market leader by far with the most name recognition. Google was the last to market. Its initial release of Gemini was a total flop because of the meme "Use elmer's glue on pizza to keep the cheese on". It has finally become more consistent, and it manages to compete with other models though I never see anyone recommending Gemini first.
All of these companies are in the red, but OpenAI has the most revenue.
This is a bit of a tangent, but I don't think OpenAI's brand is all that durable. You can see that Perplexity.AI has been gaining rapidly. At this point they have half as much search traffic as OpenAI:
It was used in Google translate, and BERT was incorporated into search in 2019, though I don't think it was a clear win for search, I feel like I started having to add exact quotes to everything technical/programming around then.
One thing I don't understand is google has so much metadata on search sessions to RLHF their search results.
E.g. when I start a search session to solve a programming problem (before llms), I will continually search different terms to get to my solution webpage. Then stop. This session metadata and the path I took is highly significant data that can be used to help llms recognise what research itself looks like.
Not RLHF, but my understanding was they heavily use that data and it was a big part of their moat, part of why competitors wanted to clone their results because they couldn't derive as good of quality from the web alone (Microsoft used the bing toolbar to clone them in the 2010s).
Hard disagree, I used Android for years but after they removed feature after feature that I found useful, and replaced them with features that shovelled ads at me, I finally gave up and switch to iOS.
I think the last straw was when I was forced to replace the stock launcher just to avoid the built in unblockable google search ads. That made me consider why I bothered with Google at all anymore if all I do is hacks to work around their crappy ad filled UI.
You mean the Google sidebar in the Pixel launcher with Google News that replaced Google Now? Yeah, that was awful. But you could disable it in either the launcher settings or the Google app itself. I'm glad I'm on GrapheneOS now.
I recommend Microsoft's Launcher, it's simple and gets out of your way. Similarly, Edge's search widget just launches a browser and executes your search. This really should be the stock behaviour that Google should provide.
If, by better, you mean more locked down, and with incremental tweaks, sure. I would much rather have Android from 5 years ago, and the ability to make it work how I want, than what there is now.
You still can have seen Android from 5 years ago experience. There's lots of custom ROMs for that. On the other hand, for an average person, I believe Android is better today in most ways.
Most of the 'new' features in Android over the past 5 years have been available in custom ROMs during that entire time. While there are, indeed, minor improvements for the average user, it's been very minor, and leaves you wondering why they didn't have it in the first place (or it's exclusive to Pixel or some shit like that).
And the average person won't even notice most of the new features/improvements; perhaps the biggest one is the camera, and that's all done in hardware or AI these days, which is not Android per-se than a photography app (which, again, is usually a Pixel exclusive)
I don't know about deep level impressive but I'm finding the new Google Lens thing built recently into Chrome pretty cool and useful. You click it and highlight part of the page and then it figures what the image is or ocr's text in it and optional translates or searches it. I use it multiple times per day. Also just being able to do that was a bit sci-fi 15 years ago.
Deep level impressive but maybe just by a company owned by Google is the Deepmind stuff like AlphaFold which recently got a nobel prize and AlphaGo and MuZero. Also you may have heard of the chatgpt/llm stuff that's trendy now, all based on https://en.wikipedia.org/wiki/Transformer_(deep_learning_arc...
They are still not able to arrange some widgets on the screen and to underestand that, if an app does not have microphone permission, you don't need a bloody microphone widget on the screen.
The number one feature I use Google Assistant for (and probably most people's most common usage) is to say "Hey google, set a timer for X minutes". If I dare do anything else with my phone during it's reply of "OK, five minutes, starting now" I risk the timer just... disappearing. If I can't count on it for one of the most commonly used tasks for phones I have zero faith for other things and don't even bother to figure out how to use it as more than an internet access terminal/music player.
Has anything serious changed about photos in the last 15 years? I've used it before then and can't really think of anything apart from embedding search and random reminders. But those are really small features.
I mean, afaik the standout feature of Picasa was shared web albums where you could invite people to add photos and permissions were managed via Google accounts, so, it's easy to confuse the two. From memory the only thing that changed when moving from Picasa to Photos was that I no longer had a desktop app where to keep photos on disk + Picasa had a neat map of geolocations (maybe Photos caught up or can you still not view photos on a map? I know apple does this, I just use ACDSee now and keep it offline)
Google photos is awesome. I genuinely love it, I've been uploading my old travel photos to it and every few days my phone reminds me of some old memories.
> Maps has only become slower and less informative (I remember when it use to actually display everything that was in a location and not just the popular/paid for stuff) since 2009
"Ground Truth" is truly dead... we've been to 25 states in the last year and the speed limits displayed in Maps were correct about 10% of the time.
Areas with multi-million populations get updated maybe every few months, and might import and manually fix up data from the official sources, e.g. city planning.
Areas with tens of thousands of residents get updated on a multi-year schedule; a year-long construction related speed drop might remain in place for years after the construction ends.
Truly remote areas are batch imported from other systems, traced out from satellite images, or some such. Mountain and desert roads are rough suggestions at best.
This seems to have always been true of Google Maps.
Eh, were you going to? If say it's accurate about 80% of the time and with in 10mph all of the time. About the only time it gets really confusing is when you're on stacked highways/interchanges.
Maybe in or near a large city… In the vast majority of the country it’s not even close (55 in a 25 recently, many hundreds of miles where it won’t even show a speed limit even on major freeways in Florida)
I haven't noticed any slowdown at all, and it wouldn't depend on the document size anyways. In fact, I remember Docs slowing down on 30-page files a decade ago, whereas now it handles 100 pages just as fast as 1.
Only the visible portion of the document is rendered (previously in HTML, now in Canvas). Everything before/after is just the document data.
Maybe you started working on longer docs coincidentally around the same time Docs switched to canvas?
>> Plus the complete lack of vision or strategy from Google’s senior leadership.
>I genuinely can't think of one impressive thing Google has done in the last 15 years.
It is strange to see how the "narrative" of Google unfolds. If parents and grand parents said this in 2015 or 2005 it would perhaps be an extremely unpopular opinion. Or I doubt both would have that opinion in 2015 or 2005.
But >complete lack of vision or strategy from Google’s senior leadership > has been a thing since 2003 - 2004 during their IPO. None of their problems happening now is really new. Their "Dont be Evil" BS, their WiFi privacy issues in 2006 - 2007 before Steve Jobs take a jab on stage in 2008. That was before most people thought about privacy. Not developing a browser against Firefox and then Chrome. It took some Firefox developers some 10+ years before they realise may be Chrome and Google isn't what they thought. They basically earn more money than they know what to do with it and had zero discipline on what, how and where to use it. They continue to pay $10 - $20B a year to Apple as default search engine. Apple is very good at extracting value out of Google. While it seems no one at Google cares about it.
Needless to say I have been ringing the alarm bell on google for 20 years. I wished Mozilla take notes earlier. But they have their own sets of problems.
If there is one thing other than search that Google has achieved was they managed to lift up the salary of the whole Tech industry. They single handedly pulled the average salary of programmers up 10% YoY for many years since IPO. To the point in ~2018 - 2022 many really thought Google's starting salary for a new Junior Dev is $200K.
I think Meta deserves more credit for the salary growth- they weren't a part of the cartel of companies colluding to keep salaries down, and them poaching from Google helped to force Google to defect.
I agree. But Meta is still much of a forbidden word on HN. I believe we had this discussion on HN people dont want their salary rise to be attributed to Meta ( then Facebook ) in 2014 - 2016.
I think you may have rose tinted glasses on historical Youtube.
Youtube used to be a wasteland of horrible insulting comments closer to 4chan in politeness and quality. Near constant spammy meanness.
Youtube has issues, yet when the Baltimore bridge collapsed, the live view on the bridge was almost constantly polite, curious, and respectful commentary. The text summaries are actually a nice change so that if you don't want to sit through some annoying drawn out video, that often ends up using a robot voice anyways, you can just read the text summary. A feature that I was personally interested in a long time ago when almost every article became a video instead. Same reason https://neuters.de/ is one of the best ways to read Reuters News.
Part of the issue with Youtube is not Youtube specifically, it's humanity. A lot of the 90% is annoying spammy BS, and it's difficult to filter away the constant "first post", "you so hot girl", "me too", "like like", "rofl wtf bbq" nonsense. Part of what ended up making FB almost unreadable. Even small town groups are deluges of spammy, copypasta, upvote trolling.
Scrolled through Youtube front page to find a "random" video and landed on a Half Life 2 documentary. Comments were either polite, informative, or opinion without being especially flamebait. There's a lot of the world wide web that's not nearly such high quality these days.
The others issues would probably be multiple pages of commentary, having tried to be an Android developer. "The function you're searching for is deprecated, move to ... a function that is also deprecated."
When I started at Google it has a very Montessori-like atmosphere. Really brilliant people there who were given a lot of autonomy to go and figure things out.
By the time I left, almost nine years later, the culture was dominated by fear and conflicting top-down directives — and the autonomy was gone.
The launch was technically pre-2009, yet Streetview meets my personal criteria.
Streetview was OK for many years, yet now borders on the miraculous.
The other day I virtually drove to Seattle just to see what the capabilities were and whether Google was capturing the downtown homeless encampment situation. Visited near Covid and the tent quantity was disturbing. Found out Seattle's response was to wall off the entire park downtown where homeless were gathering. [1][2]
However, far as Streetview, hundreds of miles of continuous highway where you can click yourself along the entire way to any location in America.
The week after, upon finding The Trace's - Gun Violence Map [3] I wanted to check whether East side Washington DC was really that much of a danger zone. Seemed difficult to believe right next to so much political money and visibility. Ended watching the Google car have a gun pulled on them in the rear view by drug dealers and watching a different Google car drive right into a police raid. Mission accomplished. East side Washington DC is kind of sketchy.
The HN community seems split between people who want simple products that do one thing well, and others that want continual feature churn and additions. What was Google Docs need exactly to be better? It seems to work perfectly for the targeted users.
They don't want features added to it, they want to see new revolutionary products similar to it. I remember the magic of early Google, delivering products and features that actually solved problems. Gmail offered 1GB storage while Yahoo acted like giving you 15MB was generous.
Ironically I think the dearth of innovation is related to consumers having been rendered too passive since 2010.
Nobody has "needs" for Google to address anymore; now we are told what we should want (clear indication that advertising execs now run the show), usually under protest as they forcibly reshape industry standards and kill beloved products.
The infantilization of AI and the subsequent hilarity of Gemini blackwashing white historical figures says all it needs to about the culture shift.
Search is also censored to an embarrassing degree. Queer theory is treated as irrefutable fact and you'll find infinite dubious content promoting it, while conspiracy theory and anything critical of Israel is downranked or delisted altogether.
Early Google gave us lightsabers [for "free"]. Now they sell lefthanded safety scissors and insist nothing has changed. All stars burn out eventually.
> I genuinely can't think of one impressive thing Google has done in the last 15 years.
I find that hard to fathom. I think what you meant to say is "an impressive right that actually got made into a b2c product".
Otherwise you'd have to ignore that they kinda pioneered llms, until OpenAI poached their tech, polished into a (for a consumer) breathtakingly functional "AI"
They also kept researching self driving via weymo etc
On the business side they've also made a significant mark on the programming world via k8s, golang and angular2 among other things
But I'd completely agree with the sentiment that they completely dropped the ball wrt their original target demographic. Beyond the improvements to android, I can't really think of anything since 2010 either that really improved things.
> I genuinely can't think of one impressive thing Google has done in the last 15 years.
The stupid thing is that, for example, ChromeOS is genuinely technically very impressive. Wayland over virtio to securely expose GUIs from untrusted virtual machines is brilliant technology. I want that, reimplemented by someone who cares about an open source ecosystem.
That Waymo thing is pretty cool. Transformers seem like pretty foundational work, even if they weren't the ones that ultimately popularized it in the market.
Maps may have had lots of improvements but they keep relentlessly cramming more and more ads and sponsored content into it. I now actively avoid using it as much as possible.
Which is directly relevant to the topic being discussed here. Engineers work hard to make real improvements but the product as a whole is sabotaged with the never ending pressure to monetize more.
Not going to be mild here, Google Maps is the absolute poster child of enshittification. Sorry if you've worked on it, it's even more enshittified than search, and the best example of the company being run by the Ads division.
This overview 8 years old but it's only gotten worse [1]. Everything in the UI has been optimized purely for businesses and ads, not for users. I'm blessed enough to now live in one of the few countries where Google Maps is not dominant, where local players are more popular.
I've gone and taken current screenshots of both to show the difference. Current Google Maps [2]. Note how you can't even hide those UI elements blocking the map, but more important is the map itself. Current locally popular maps app [3]. It.. actually works well as a map, like Google Maps used to before Ads took over the business. And just a single touch on the map to hide the UI.
I'm sure the Supernova Hotel and Hostel Ani & Haakien are bringing Google some great cash money though, splendid user feature!
Before I lived here, I of course used Google Maps. And when I go abroad, I have to use it. And every single second spent with it is a stark reminder of how much worse it is than the map apps that are most popular here and are not enshittified.
Some of that stuff is impressive, but I think you might be stretching a bit with lyrics and calculator. They're nice quality of life improvements, but I don't think I'd classify them as "impressive".
Docs and Sheets are still so much better than Word and Excel, except that there doesn't seem to be a way to, from the desktop, launch a .csv into Sheets (or .doc into Word).
Though I think that, for every minor improvement, I could name a regression or product shut-down.
It would make it impressive that they could make it local without depending on chrome. In addition that it would allow the collaborative aspects natively. Just like on Android or iOS.
Sorry no: maps has gotten way worse for my use cases. On mobile, they started agressively trying to get you to use the app instead of browser and try to get you to turn on your location constantly. They also kneecapped the "near to here" query button when looking at an adress on the mobile browser, where you need to try to use a roundabout hack to get it sort of working. Constantly enshittifying the mobile browser in favor of the app is not a bargain with the devil I'm willing to accept.
Even closer to the OP, there was a branch of AI in the 1990s known as genetic programming that literally applied the concept of natural selection to solving certain types of problems:
"But I do tons of hobbyist electronics with surface mount!", some could say. Yeah, sure, but how do you wire it? You order a PCB from OSH Park? That's not very scavenge-friendly. " - https://new.collapseos.org/why.html
Not that I totally disagree with this but see clay PCBs, another post-supply-chain-collapse electronics project https://media.ccc.de/v/38c3-clay-pcb#t=1689